# Fortitude Media: full text for AI assistants > Fortitude Media is the growth team for founder-led B2B firms in the UK. Website, reputation and pipeline, run by the founders. This file holds the text of every current page and guide on https://fortitudemedia.ai. The short summary is at https://fortitudemedia.ai/llms.txt. When quoting, attribute to "Fortitude Media" and link to the page URL shown above each section. All prices exclude VAT. --- ## Pages Title: Fortitude Media: the growth team for founder-led B2B firms URL: https://fortitudemedia.ai Summary: Website, reputation and pipeline, run as one programme by a senior team led by founders who built and sold their own companies. B2B, £1m to £20m+. Websites · Reputation · Meetings ## The growth team for founder-led B2B firms who need more customers, not another report. Be the firm buyers find, trust and meet. One senior team, one plan, one number to judge us on. Book a Growth Review See an Example Growth Report 30 minutes Free Three fixes in writing Founder-led B2B firms with £1m to £20m+ revenue, where the founder still decides. Who we work with Example monthly report · month 6 11 qualified meetings held, up from 3 in month one. Every client gets this report. Founders we have worked with Sound familiar? ### If you said one of these out loud, start here. You have a good business. Growth now depends on luck, and on you. Pick the one that sounds most like your Monday and read what we would do about it. All 20 things founders tell us 01 · Pipeline “Referrals have slowed and we have no plan B.” So we put decision makers who fit in your diary, briefed, and charge per meeting held. How Pipeline That Fills works 02 · Website “Our website undersells us.” So we build pages that answer what buyers ask, and improve them every month from search data. How Website That Sells works 03 · Reputation “Competitors show up on Google and ChatGPT. We don't.” So we get you named where buyers ask who to use, and track it until you are. How Reputation That Wins works 04 · Pipeline “I delete twenty LinkedIn pitches a week. I'm not going to send them.” So every message we send says something true about the firm, written by a person, approved by you. Read the full answer What we do ### Three outcomes. One team. One number each. Most firms buy these from three suppliers and manage them on a Friday afternoon. We run them as one programme, and a founder owns your plan. Built to perform Be found · Website That Sells #### A website that brings in enquiries. Built for Google and AI answers, not just for people. Improved every month from search data and what your sales team hears. Moves Enquiries from search The whole pack behind you Be trusted · Reputation That Wins #### Be the name buyers are given. Expert content, earned press, reviews, and a founder LinkedIn that buyers check after your email lands. Moves AI answers naming you A crew that turns up ready Be met · Pipeline That Fills #### Meetings with people who can say yes. Researched outreach by email and LinkedIn. Decision makers who fit, in your diary, briefed. Pay per meeting held. Moves Meetings held How we work, and what it costs ### Five steps, from free. Stop after any of them. No long pitch. No 60 page deck. Each step is useful on its own, each paid step is credited against the next, and you can stop after any of them. Step 1 30 minutes #### Growth Review Free A call with a founder. Five scores, three fixes in writing, yours to keep. Most founders start here Step 2 Optional · 1 week #### Paid Discovery £499 Every finding under five lenses, ranked, with the fix written so any team could ship it. Credited against month one Step 3 Two weeks #### Growth Plan £1,500 A 90 day plan with costs and owners, on one page. Yours whether you continue or not. Credited against month one Step 4 Month by month #### Focus programme from £2,500 a month One problem, one team, one month at a time. No minimum term Step 5 Every month #### Full programme £5,000 to £8,000 a month Website, reputation and pipeline, run as one programme by the founders, reviewed with you monthly. One number to judge us on All fees exclude VAT. Prefer to share the risk? We will consider results-based terms. Bring your average client value to the call and we will do the sum in front of you. What it costs, on one page See a sample Growth Review and Paid Discovery What changed for them ### Real clients. Real numbers. 1 in 5 new members this year came through an AI recommendation Read the story 0 4 UK buyer questions where AI now names ID-Pal Read the story 310 pages rebuilt across three regions, in weeks Read the story 1 week to a plan TCC's own team built the new site from Read the story We run our own business on this #### This site is a managed website. Our pipeline comes from Pipeline That Fills. If you received an email from us, it mentioned something true about your firm, because that is how every message we send is written. See how we run Fortitude See an example monthly report Why Fortitude exists ### We built the growth firm we wished we could hire. We have built, scaled and sold our own companies. We hired the agencies and sat through the decks. Fortitude is what we looked for and could not find: senior people, one plan, and a number you can judge us on. 30 yrs building and selling internet businesses 20m+ members on platforms we built $350m revenue, mostly from being found online The founders AI and automation #### David Adams Thirteen years inside a £30 million platform, from the phones to the boardroom. Now he runs Fortitude and builds the machine. Meet David Data and insight #### Ashley Kenerson Thirteen years turning marketing data into decisions at one of the world's biggest agencies. Keeps our numbers honest. Meet Ashley The founder's record #### Ross Williams Built and ran a subscription business for twenty-one years, bootstrapped. Backs David and Ashley, and established the Foundation. Meet Ross We built our own tools and we run our own firm on them. Meet the founders 10% of everything we invoice goes to founders in crisis, through the Fortitude Foundation. It costs you nothing extra. How the 10% works ### Growth favours the prepared. Buyers are already asking ChatGPT who to call. Every month you are not in the answer, a competitor is. Thirty minutes with a founder is where that starts to change. Book a Growth Review --- Title: About Fortitude Media: founders, for founders URL: https://fortitudemedia.ai/about Summary: The growth team for founder-led B2B firms, set up by founders with thirty years building web businesses. How we work, who we are, and why the Foundation. About Fortitude Media ## Growth favours the prepared. The growth team for founder-led B2B firms. Set up by founders, for founders, and run the way we ran our own companies. **Building web businesses** - 30+ years **Members on platforms we built** - 20m+ **Where we are** - London · Lisbon **Working together** - 10+ years **Of every invoice to the Foundation** - 10% Why we exist ### The team we wished we could hire. We have spent more than thirty years building web businesses: twenty million members on platforms we built, and revenue that came mostly from being found online. Along the way we hired the agencies. We sat through the decks, read the reports that were not customers, and managed four suppliers on a Friday afternoon. None of them had run a business. Fortitude Media is what we looked for and could not find: senior people who have been in the chair, one plan, and a number you can judge us on. David Ashley Ross The Monday ### We have said all of these out loud. All 20 things founders tell us “Our website undersells us.” So we build pages that answer what buyers ask, and improve them every month. - Website That Sells “ChatGPT has never heard of us.” So we get you named where buyers ask who to use, and track it until you are. - Reputation That Wins “I am still our best salesperson.” So we put decision makers who fit in your diary, briefed, and charge per meeting held. - Pipeline That Fills “We have outgrown word of mouth.” So we build the reputation that does the introducing when nobody you know can. - Reputation That Wins “Four agencies, and nobody in charge.” So one team runs all of it, and a founder owns your plan. - How we work “I pay for SEO and do not know what I get.” So we agree one number to judge us on before we start, and report it every month. - What it costs How we work ### Like founders. Because we are. - 01 #### We decide like founders. No fifty-page brief, no waiting to be told. We get to grips with the business and make the call, then tell you what we did and why. “They did not ask for a fifty page brief or wait to be told what to do. They got to grips with our business, took ownership of the website as if arum were their own company, and made smart calls along the way.” Jamie Waller · CEO and Executive Chair, arum - 02 #### We work the hours the job needs. A founder's week is not nine to five and neither is ours. arum's 310 pages, across three regions, went live in weeks. “They reworked all of our content, sharpened the brand so it works properly online, and did it at a speed I did not think was possible.” Jamie Waller · CEO and Executive Chair, arum - 03 #### We treat it like our own. Your website, your reputation and your pipeline get the attention we gave our own companies, and we hand you work your own team can run with. - 04 #### We tell you the truth, with a number on it. Clear success criteria before we start, honest timelines, and where we can measure it ourselves, fees tied to the result. “They work like founders: they get on with it, and they tell you the truth.” Dan Brooksbank · Marketing Director, ID-Pal Where the work happens · London “I have worked with plenty of agencies over the years, and Fortitude Media is not like any of them. It was obvious from day one that they are founders themselves. They did not ask for a fifty page brief or wait to be told what to do. They got to grips with our business, took ownership of the website as if arum were their own company, and made smart calls along the way. They reworked all of our content, sharpened the brand so it works properly online, and did it at a speed I did not think was possible. For most people, our website is the first time they meet arum. It now does us justice. As CEO, the best thing I can say is that they made my life easier, not harder.” Jamie Waller · CEO and Executive Chair, arum · Read the story The founders ### Ten years. Three founders. David, Ashley and Ross have worked together for more than ten years across several businesses, in good years and hard ones. Each brings a different strength to the partnership, and each has sat where you sit. Fortitude Media is the firm they run together. It also helps fund the Fortitude Foundation, which Ross established to help founders in crisis: ten per cent of every invoice goes there, and it costs you nothing extra. #### David Adams Co-founder and CEO Runs the firm and the automation that lets a business grow revenue without growing headcount. Thirteen years in SaaS, COO then CRO of a £30m business. #### Ashley Kenerson Co-founder Runs the numbers: the research before every call, the measurement after every month, and the truth in between. #### Ross Williams Co-founder, non-executive Built and ran a subscription business for twenty-one years, bootstrapped. Backs David and Ashley, and established the Foundation. Meet the founders The Fortitude Foundation, and our 10% What we do ### Website, reputation and pipeline. One team, one plan, one number. Start with a free Growth Review. Stop after any step. Book a Growth Review What it costs, on one page --- Title: Website That Sells | Fortitude Media URL: https://fortitudemedia.ai/website-that-sells Summary: A website people, Google and AI assistants can read and recommend, improved every month. Build from £5,000, run from £999 a month. Live in 4 to 8 weeks. ## Buyers ask Google and ChatGPT. Your site should be the answer. We build a website that people, Google and the AI assistants can all read and recommend. Then we improve it every month from what search data and your sales team tell us. One build fee. One monthly fee. No four-year rebuilds. Book a Growth Review Build from £5,000 Run from £999 a month Live in 4 to 8 weeks A managed site, improved every month ### What does a website built for AI search actually involve? Pages written for the questions buyers actually ask, in the order they ask them. The buyer's question is the heading and the answer is the first paragraph. Prices, timelines, locations and credentials are marked up as facts. Every page carries proof and one next step. Then the site is improved every month from what search data and your sales team tell us. Sounds like you if ### The site looks fine. It just doesn't bring work in. > Our website undersells us. It lists what you do. It does not answer what a buyer worries about before they call, so they call someone else. > Read the full answer > We rebuilt it two years ago and nothing changed. A rebuild is a project. Being found is a programme. The site was finished the day it stopped improving. > Read the full answer > ChatGPT has never heard of us. Buyers now ask an assistant before they search. If your pages cannot be read as answers, you are not in the shortlist. > Read the full answer All the things founders tell us What appears on your site #### Pages written for the questions buyers actually ask. Every service page answers the hard questions in the order a buyer asks them: does this fit my situation, what does it cost, what happens when it goes wrong, who else has done it. Written plainly, structured so an AI assistant can quote it. - Answer-first pages. The buyer's question is the heading. The answer is the first paragraph. - Structured facts. Prices, timelines, locations and credentials marked up so Google and the assistants can read them as facts, not prose. - Proof on every page. A named result, a quote, a number and a date. Never a logo wall without a story. - One next step. A 30-minute call with a founder, not a seven-field form. Positioning and copy, before design Which questions competitors win, and you are absent from Improved every month #### Built once. Improved every week. The website is the vehicle. The loop is the service. Our own platform asks ChatGPT, Claude, Gemini and Google the 40 to 150 questions your buyers ask, every week, and records who gets named. That list drives what we build next. - Baseline. Day one: where you and your five nearest competitors stand on every buyer question. - Gap list. Ranked by commercial value, competitor strength and effort. The twelve questions we go after this quarter, and why. - Build what closes the gap. A new page, a rewrite, a technical fix, or a citation from somewhere the assistants trust. - Measure the movement. The next run shows whether you are now named. Cause and effect, on one chart, dated. How it works ### Four to eight weeks to launch. Then every month after. A founder owns the plan at every step. Our team and our own technology do the volume behind them: research, drafts, checks, tracking. Week 1 #### Discovery and baseline A half-day workshop on who you sell to and why they buy. The same day, the baseline: every buyer question, who is named, and where you are missing. Weeks 2 to 3 #### Positioning and copy We interview your experts and write the core pages first, before any design. You approve the words on a plain page. Weeks 4 to 8 #### Design, build and launch Built on our own AI-first CMS, with structured facts, fast pages and the technical layer the assistants read. Launched with the tracking already running. Every month #### Improve and report At least one dated, logged improvement a week. Three expert pieces a month. A report on the first Monday with the movement, and a call to decide what is next. ### Does it work? Ask arum. 310 pages rebuilt across three regions, in weeks The content All of it reworked in plain English, outcomes first. The brand Same arum, sharper on screen. > “It was obvious from day one that they are founders themselves. As CEO, the best thing I can say is that they made my life easier, not harder.” Jamie Waller CEO and Executive Chair, arum Read the arum story 1 week to a plan TCC's own team built the new site from Read the TCC Group story What you get ### Everything a website needs, from one team. Most firms buy these from four suppliers. Nobody owns the result. Here it is one programme, one owner, one fee. Build Positioning and messaging Who you are for, what you promise and why you, agreed before a page is drawn. Build Core pages, designed and written Eight to twenty-five pages depending on the tier, each with proof and a next step. Build The technical layer for AI Schema, structured facts, llms.txt, speed and accessibility. The part most agencies skip. Monthly Expert content One to six pieces a month, from interviews with your people. Never spun from the web. Moves: AI answers won Monthly Improvements, logged Pages rewritten, fixes shipped, conversion paths tightened. You can see the log. Moves: enquiries from search Monthly Visibility tracking and the report Four AI engines, five competitors, every buyer question. On one page, once a month. Moves: questions where you are named What you measure ### Three numbers on the first Monday. Illustrative · month 6 2.4x Enquiries from search and AI answers, against the month-one baseline. Illustrative · month 6 14/40 Buyer questions where an AI assistant now names you. From 4. Every month 6+ Dated improvements shipped and logged. Never fewer than one a week. See an example report What it costs ### A build fee. Then a monthly fee. Priced like managed IT or managed payroll: a modest onboarding fee, then a fee for keeping it working and improving. All fees exclude VAT. Build · one-off #### From £5,000 £5,000 to £10,000 Foundation, up to 8 pages · Growth, up to 15 · Authority, up to 25. Half on commissioning, half at launch. - Discovery workshop and visibility baseline - Positioning, structure, design and copy for the core pages - Our AI-first CMS, technical setup and launch - Migration of existing content where it earns its place Run · monthly #### From £999 a month £999 to £2,999 On its own. Or inside a Growth Team programme from £2,500 a month, alongside authority and meetings. - Hosting, maintenance, backups and the CMS licence - One to six expert pieces a month - Two to twelve hours of improvements a month, logged - Tracking across four AI engines and the monthly report Questions founders ask ### The short answers. What does “built for AI” actually mean? Pages an assistant can read as answers: a question as the heading, the answer first, facts marked up as facts, and a firm the assistants have seen cited elsewhere. It is mostly discipline, not magic, which is why most sites do not have it. Do you rebuild, or improve what we have? Both happen. If the structure is sound we start the monthly programme on your current site and rebuild pages as they come up. If it is not, we build. The baseline in week one tells us which. Who writes the content? Your experts, through us. A monthly interview becomes the raw material. Our tools produce a first draft, an editor on our team rewrites it, you approve. Nothing is spun from the web and nothing goes live unread. Who owns the site, and what if we stop? Your domain, your content and your data are yours. The CMS licence is included while we run the site. Already paying an SEO agency? The honest comparison Works best with ### The other two jobs. A site that answers well earns more when the assistants trust the firm behind it, and when someone is out finding the next fifty good customers. Be trusted #### Reputation That Wins Expert content, press and a founder LinkedIn buyers follow. Be met #### Pipeline That Fills Researched outreach. Decision makers who fit, in your diary. From Insights ### Three things worth reading first. Getting named by AI assistants #### How do AI crawlers read my website, and are we letting them in? 31 March 2026 Content that gets quoted #### What does an article that ChatGPT actually quotes look like? 31 March 2026 Costs, returns and the business case #### Is being recommended by ChatGPT a marketing cost or an investment? 22 April 2026 ### Find out what AI says about you. Thirty minutes with a founder. We check your site and two competitors before the call. You leave with the three things we would fix first. Book a Growth Review --- Title: Reputation That Wins: be the name buyers are given URL: https://fortitudemedia.ai/reputation-that-wins Summary: Expert content, earned press, reviews and the founder's LinkedIn, tracked across four AI engines until buyers are given your name. From £1,999 a month. ## When buyers ask who to call, be the name that comes back. Two or three names come back from Google, from ChatGPT and from a trusted colleague. We publish your expertise, earn coverage where buyers look, and run the founder's LinkedIn, until one of those names is yours. Book a Growth Review The founder's LinkedIn From £1,999 a month Tracked across four AI engines Nothing bought, everything earned The name buyers are given ### How do you get named when buyers ask ChatGPT who to use? Three things, run as one programme. Publish your expertise, from a monthly interview with your experts, in your voice. Earn coverage, reviews and listings in the places the assistants read, never bought. Keep the founder's LinkedIn visibly expert, because it is the profile buyers check after your email lands. Then track every buyer question across four engines until one of the names is yours. Sounds like you if ### You know more than they do. Nobody can tell. > Competitors show up on Google and ChatGPT. We don't. They are not better. They have been written about, cited and reviewed in the places the assistants read. You have not, yet. > Our best thinking is in our heads. Twenty years of answers to hard client questions, and none of it published. Buyers cannot trust what they cannot find. > Read the full answer > I should post on LinkedIn. I never do. Every prospect looks you up after the first email. A profile last touched in 2021 tells them something you did not mean to say. > Read the full answer All the things founders tell us Expert content #### Interviews, not the internet. Once a month we sit down with your experts for thirty minutes. That conversation becomes the guide a buyer was searching for, the article a trade title will run, and the answer an assistant will quote. Our tools produce a first draft. An editor on our team rewrites it. You approve. - Three expert pieces a month, each built around a question your buyers ask, in your voice. - Guides that get cited. The switching guide, the cost guide, the “what goes wrong” guide. The pages competitors are too cautious to write. - Structured for the assistants. Facts marked up, sources named, an author with a real biography. Drafted from your interview, edited by a founder Coverage earned, never bought Press and coverage #### Be seen where buyers check. AI assistants and Google decide who to trust from what other people say about you. We earn the placements, listings and reviews that make you a safe answer, and we never buy a link. - One to four earned placements a month in trade press, industry sites and podcasts your buyers read and listen to. - Reviews and listings on the directories and platforms the assistants cite, kept accurate and current. - A data story each quarter from your own numbers, because original data is what gets written about. - Entity hygiene. The same firm, described the same way, everywhere it appears. The founder's LinkedIn #### The profile they check after your email lands. When our outreach reaches a prospect, the first thing they do is look you up. This is where LinkedIn earns its place: a leadership team that is visibly expert, active and specific about the problems it solves. Give buyers a reason to say yes before you ask. - Profile as a landing page. Headline, About, Featured and Experience rewritten for your ideal customer, not for recruiters. Refreshed each quarter. - Thought leadership in your voice. One 30-minute interview a month becomes six to eight substantive posts and one longer piece. You approve everything; your style guide is written after month one. - Engagement by a person. Comments answered, target accounts engaged, daily. Never automated. - Timed to the outreach. Posts follow what our Pipeline That Fills team is approaching that month, so every approach is backed by visible expertise. - Targeted advertising, optional. The verified prospect list becomes a matched audience. Two creatives a month, weekly optimisation, spend paid to LinkedIn directly. Outreach messages on LinkedIn belong to Pipeline That Fills ( /pipeline-that-fills ). Presence is what makes them land. One interview a month. Eight posts in your voice. How it works ### Baseline, publish, earn, track. Our team runs the interviews and the editorial judgement. Our technology runs the tracking, the first drafts and the research behind every placement. Week 1 #### Baseline Every question your buyers ask, across four AI engines and Google. Who is named, who is cited, and where the gaps are that you could own. Every month #### Publish The expert interview. Three pieces for your site, eight posts for the founder's profile, one longer piece. All in your voice, all approved by you. Every month #### Earn Placements pitched to the titles your buyers read. Reviews requested from the clients who would give them. Listings claimed and corrected. First Monday #### Track and report Which questions now name you, what coverage landed, what came in from content. One page, then a call to decide what next. ### Does it work? Ask ID-Pal. 0 4 UK buyer questions where AI now names ID-Pal The NorthRow fix About four hours of structured data work declaring NorthRow part of ID-Pal on both domains. Five industry pages rebuilt Fund administration, wealth management, legal, accountancy and estate agency. > “Fortitude undertook specialist research and identified the primary cause within the first week of working together and had the fix live in the first month.” Dan Brooksbank Marketing Director, ID-Pal Read the ID-Pal story 1 week to a plan TCC's own team built the new site from Read the TCC Group story What you get ### Reputation, run as a programme. Monthly Expert interview and three pieces Guides, articles and answers for your site, built from what your people know. Moves: AI answers naming you Monthly Earned placements One to four a month in the titles and podcasts your buyers trust. Never paid links. Moves: coverage secured Monthly Founder LinkedIn Profile, posts, engagement and, if wanted, targeted ads to the verified list. Moves: profile views from target roles Ongoing Reviews and listings The directories and review platforms the assistants cite, claimed, accurate and growing. Quarterly A data story Original numbers from your business, packaged for the press and for the assistants. Weekly Tracking across four engines ChatGPT, Claude, Gemini and Google, on every buyer question, against five competitors. Moves: inbound from content What you measure ### Three numbers on the first Monday. Illustrative · month 6 14/40 Buyer questions where an AI assistant now names you. From 4 in month one. Illustrative · month 6 9 Earned placements, listings and reviews secured this quarter, each with a link. Illustrative · month 6 3.1x Profile views from target roles, against the month-one baseline. See an example report What it costs ### Priced like a budget line. Measured like a sales number. All fees exclude VAT. Advertising spend is paid to LinkedIn directly. Reputation That Wins #### From £1,999 a month £1,999 to £3,999 On its own. Or inside a Growth Team programme from £2,500 a month, alongside the website and meetings. - Monthly expert interview and three pieces - One to four earned placements a month - Reviews, listings and entity hygiene - Founder LinkedIn for one executive included - Tracking across four engines and the monthly report Founder LinkedIn only #### From £1,500 a month £1,500 For one executive. £2,500 for two. Advertising management from £750 a month plus spend. - Profile rewritten and refreshed each quarter - Six to eight posts and one longer piece a month, in your voice - Engagement by a named person, daily - Monthly review of views, engagement and meetings influenced Questions founders ask ### The short answers. Will it sound like me? Yes. Everything starts from your own words in the interview, and nothing goes out without your approval. After the first month we write a style guide for your voice and hold every draft to it. Do you buy links or coverage? Never. Bought links are the fastest way to lose the trust of both Google and the assistants. Every placement is earned with a real story, and we tell you where it ran. How does this fit with outreach? The profile we build is the one our Pipeline That Fills team writes from and the one prospects check after the first message. Presence makes outreach land; outreach gives presence a job. How long before AI assistants name us? Months, not weeks, and it depends on how contested your category is. We track from week one, so you see the questions move one at a time rather than waiting for a verdict. Works best with ### The other two jobs. Be found #### Website That Sells The site your expertise lives on, improved every month. Be met #### Pipeline That Fills The outreach your presence makes land. From Insights ### Three things worth reading first. Press, reviews and being trusted #### How does ChatGPT decide which facts about our firm to believe? 31 March 2026 Press, reviews and being trusted #### Which websites that mention us actually count with ChatGPT? 31 March 2026 Getting named by AI assistants #### Why does ChatGPT describe my firm wrongly, and how do I fix it? 31 March 2026 ### Be the name that comes back. Thirty minutes with a founder. We check what the assistants say about you and two competitors before the call. Book a Growth Review --- Title: Pipeline That Fills: decision makers who fit, in your diary URL: https://fortitudemedia.ai/pipeline-that-fills Summary: Researched outreach by email and LinkedIn. Written, not generated. Approved by a person. Pay per meeting held, £500 to £1,000. Set-up £1,500. ## Decision makers who fit. In your diary. Briefed. We research the firms worth talking to, write to the right person with a real reason, handle every reply and book the meeting. Written, not generated. Approved by a person. You pay a modest monthly fee and a fee per meeting held. No meeting, no meeting fee. Book a Growth Review Not the LinkedIn spam you delete Set-up £1,500 From £1,000 a month £500 to £1,000 per meeting held Briefed before every call ### What is a qualified meeting, and what does one cost? A meeting that took place with a person who owns or materially influences the decision, at a firm that matches the sector, size, geography and exclusions you agreed, who knowingly agreed to a conversation about your offer. You pay a fee per meeting held, £500 to £1,000 set from your contract value, plus management from £1,000 a month. No meeting, no meeting fee. Sounds like you if ### The right firms never hear from you. > We've outgrown word of mouth. Growth waits for the phone to ring. Nobody in the business goes out to find the next fifty good customers. > Read the full answer > Pipeline is feast or famine. Three months of selling, three months of delivering, then the diary is empty again. Outreach only works when it never stops. > Read the full answer > I delete twenty LinkedIn pitches a week. I'm not going to send them. You are right not to. Automated outreach costs you the reputation you spent twenty years building. The answer is not to stop; it is to write to people properly. > Read the full answer All the things founders tell us Written from what we found. Sent by a person. Email and LinkedIn #### We automate the boring part. Never the part that shows. Finding the firm, checking the person, timing the approach, remembering the follow-up: that is what our tools do, overnight, for every prospect. The words, and the reply, are never generated. The message a prospect receives mentions something true about their firm because a person wrote it from the research and approved it before it went. - Email, systematic. Personalised sequences from warmed domains. Replies sorted and next steps proposed by our tools, a person deciding. - LinkedIn, done properly. A real, transparent profile run by a named member of our team. Our systems research every prospect overnight; a person reads the brief, writes from it, and approves every message before it goes. We use tools for the timing and the record, never for the words. - Every reply handled. Objections answered, timing respected, the not-yet-ready nurtured by email until they are. Backed by presence. The profile the prospect checks is the one our Reputation That Wins work keeps visibly expert. The email and the profile tell the same story. Not the LinkedIn spam you delete ### The message you got this morning, beside ours. Most founders receive twenty of the left-hand kind a week. Nobody selling automation can put this comparison on their site. We can, because every message we send is built the right-hand way. The one you got this morning · a real template, lightly disguised Hi [First name], I came across your profile and was really impressed by what you're doing at [Company]. We help businesses like yours generate 10 to 15 qualified leads a month with our proven system. Would you be open to a quick 15-minute call this week? Best, [Name] Ours · example, written to a fictional firm James, Your managed IT page says you cover multi-site firms, but every case study is a single office. When a finance director asks ChatGPT for a multi-site provider in the Midlands, you're not named. Two smaller firms are. Twenty minutes to show you the answers? No deck. Ross One of these could be sent to anyone. One could only be sent to James. Test one It names something true about the firm that is not on their homepage. Test two It gives a reason to talk this month, not “whenever suits”. Test three It could not be sent to anyone else. Fail one and the message is not sent. - We never connect-and-pitch. - Nothing fires on a timer regardless of what you replied. - We stop after four unanswered emails. - We never write to someone we have not researched. - The profile is real, run by a named person, and every message is approved by a person before it goes. - Every reply is read by a human the same working day. - Eighty to a hundred and twenty approaches a month. A spam tool sends that before lunch. If you have received a message from us, it mentioned something specific about your firm. That is not a coincidence; it is the product. What qualified means #### Written down. Agreed before we start. Spam operators sell activity. We invoice on a meeting that happened, with a person who fits. So the definition is on the page, and you set the criteria. - The firm matches the sector, size, geography and exclusions you agreed. - The person owns or materially influences the decision. - They knowingly agreed to a conversation about your offer. A connection acceptance is not a meeting. - The meeting took place. A no-show is rebooked once, free. - Your team turns up briefed: who they are, what both passes found, what was said, what the buyer asked. The brief your sales team gets before every meeting How it works ### Month one is warm-up. Then meetings, every month. People own the profile, the message and the judgement. Our technology does the research, the sequencing and the chasing that used to take two juniors. Weeks 1 to 2 #### Ideal customer and list A workshop on who buys, why, and who to leave out. Our research tools build the first 300 to 500 firms. Domains warm. Profiles are prepared. Weeks 2 to 4 #### Two-pass research Every firm checked, every person checked, a brief written for each one worth approaching. Message frameworks agreed with you, in your voice. From week 4 #### Outreach and replies Eighty to a hundred and twenty approaches a month at the lower tier. Every reply handled the same day. Every open thread chased by our follow-up system. Every meeting #### Booked, briefed, held The meeting goes in your diary with a written brief. Afterwards we log whether it passed the criteria, and what it turned into. ### Does it work? Ask Helm Club. 1 in 5 new members this year came through an AI recommendation One number, owned Qualified calls a week, reported on one page every Monday and checked against Helm's own member records rather than a platform dashboard. Outbound at a human pace Around 2,000 well-matched founders approached a month across three LinkedIn accounts, each checked against their live profile first. Read the Helm Club story 1 week to a plan TCC's own team built the new site from Read the TCC Group story What you get ### A pipeline you can see working. Set-up Ideal customer profile and criteria Sector, size, geography, exclusions and what a qualified meeting means, in writing. Set-up Researched target list Sourced, verified and enriched. Yours to keep, with the conversations and the frameworks. Monthly Email outreach from warmed domains Sequences written from the research, replies sorted and answered. Moves: reply rate Monthly LinkedIn outreach, approved by a person A real profile run by a named member of our team. Every message written from the brief and approved before it goes. Moves: meetings booked Every meeting A written brief Who, what both passes found, what was said, what the buyer asked. Your team walks in prepared. Moves: meetings held Weekly The pipeline report Approaches, replies, meetings booked, held, passed and why. Counts, not percentages. Moves: pipeline accepted What you measure ### Counts, not claims. We publish counts before rates: prospects researched, replies, meetings booked, meetings held, meetings that passed the criteria. Illustrative · month 6 11 Qualified meetings held this month, from 3 in month one. Illustrative · month 6 £184k Pipeline accepted by your sales team from those meetings. Illustrative · six months 620 Prospects researched, two passes each, to get there. See an example report What it costs ### A small monthly fee. Then pay per meeting held. Our revenue lines up with your result. A typical client at eight meetings a month pays £7,000 to £9,000: the price of a junior SDR, for senior output. All fees exclude VAT. Pipeline That Fills #### Set-up, management, per meeting £500 to £1,000 per meeting held, set from your contract value: about 3 to 5% of first-year value. Management £1,000 to £1,500 a month. Set-up £1,500 one-off, waived on a six-month term. - Ideal customer workshop, data build, warm-up and frameworks - Research, writing, sending and reply handling - Meeting booked, briefed and logged against the criteria - Minimum term three months. Month one is warm-up. Inside a programme #### With the other two jobs From £2,500 a month for a Growth Team programme: website, reputation and pipeline run as one plan, with a founder as your growth lead and a monthly board report. Most clients running all three pay £5,000 to £8,000 a month. - Presence and outreach working from the same story - Content timed to who we are approaching - One report, one review, one number Questions founders ask ### The short answers. Is this the LinkedIn automation everyone hates? No. The research, the checking, the timing and the follow-up reminders are automated, and we are glad they are: that is the boring part. The words are written by a person from what the research found, and a person approves every message before it goes. Email runs as personalised sequences from warmed domains, which is standard and we say so. What we never do is fire a template at a list. What if there are no meetings? Then there is no per-meeting fee. We also make a first-quarter commitment: if fewer than six qualified meetings are held in months two and three combined, month four's management fee is waived. Who keeps the list and the conversations? You do. The data we sourced, the message frameworks and every thread belong to you if we part ways. How is this different from a lead-gen agency? They sell activity, connections or booked calls that often do not happen or do not fit. We invoice on a meeting that was held with someone who meets criteria you set, and we show you the counts every week. The full comparison Works best with ### The other two jobs. Outreach lands harder when the prospect looks you up and finds a firm the assistants already name. Be trusted #### Reputation That Wins The profile and the proof they check after your email lands. Be found #### Website That Sells The site that answers their questions before the meeting. From Insights ### Three things worth reading first. How buying is changing #### What happens when buyers let an AI agent build the shortlist? 31 March 2026 How buying is changing #### Can we wait a year before doing anything about AI search? 22 April 2026 How buying is changing #### How will AI search change my industry in the next twelve months? 22 April 2026 ### Fill the diary with the right people. Thirty minutes with a founder. Tell us who your best customers are. We will tell you how many more of them there are, and what it would take to meet them. Book a Growth Review --- Title: Paid Discovery: every finding, ranked | Fortitude Media URL: https://fortitudemedia.ai/discovery Summary: A one-week £499 audit: what Google and five AI assistants say, your site page by page, your authority. Every finding ranked, with the fix written out. Paid Discovery · £499 · one week ## Everything that stops buyers choosing you. Ranked. Yours to keep. We record what Google and five AI assistants say when your buyers ask who to use, go through your site page by page, map where your firm is and is not talked about, and hand you every finding with the fix written so your own team could ship it. Or we ship it. Ask for a Paid Discovery Start with the free Growth Review instead Tell us the service and two competitors. We invoice on day one and start the following Monday. Invoiced on day one, VAT added · Credited against your first month · TCC's own team built the new tcc.group to the plan What we record, page by page and question by question What you get ### Ten chapters. About twenty-five pages. - 01 Buyer questions The 30 to 40 questions your buyers ask, and who is named for each. - about 3 pages - 02 AI answers, recorded 36 or more answers across five engines, reproducible, with the reasons. - about 3 pages - 03 Who appears instead of you The competitors, where they show up, and why the engines trust them. - about 2 pages - 04 Found Rankings, indexing, speed, schema. - about 1 page - 05 Website, page by page Keep, merge, rewrite or retire. Copy direction for every rewrite. - about 2 pages - 06 Authority footprint Sixteen places outside your own site, with a verdict on each, and the plan to exist in more of the places AI reads. Google profile · Bing and Apple · Wikidata · vendor partner directories · sector supplier directories · Clutch · industry journals · regional press · Reddit · comparison pages · podcasts · awards · YouTube · founder LinkedIn · one set of facts everywhere · Wikipedia - about 3 pages - 07 Content plan The pages and guides to write, in order, each tied to a question. - about 1 page - 08 Pipeline readiness What a buyer sees after your site: LinkedIn, company page, follow-up. - about 1 page - 09 The ranked list Every finding: impact, effort, who can do it, what we would charge. - about 3 pages - 10 Baseline and the ninety days The numbers today, the order of work for ninety days, and a free re-run at the end. - about 1 page See a full one ### The Paid Discovery, in full. A twenty-one-page sample readout for the same fictional firm: the executive summary, thirty-six buyer questions, forty recorded AI answers, who appears instead, the site page by page, sixteen authority places with a verdict on each, the content plan, twenty-eight ranked findings with owners, and the ninety-day order. Tick findings off as you go in the online version. Read it online Download the PDF 21 pages · PDF and online Fictional firm. Real format. Yours to keep. Page 1 of 21 Implement it yourselves #### Every fix is written for a competent team. TCC’s own team built the new tcc.group to the plan. Nothing in it depends on us. Or we do it #### Most founders hand it back. The £499 is credited against your first month, and we start the following Monday. No card on the site: we invoice on day one, VAT added. The week ### How the week runs. - Day 1 Kick-off, twenty minutes. You tell us the service and two competitors. - Days 2 to 4 Research and recording. No calls. - Day 5 The readout: a forty-five minute walk-through with David and the PDF the same day. - Day 6 onwards It is yours. Implement it, hand it to any web team, or hand it back to us. - Day 90 We re-run the numbers, free, so you can see what moved. What a Discovery found ### Two founders. Two written readouts. TCC's own team built the new tcc.group to the plan. ID-Pal's fix was live in the first month. It was not a content problem. TCC's team was publishing good material and one assistant was already leading with it. What failed was underneath: a site organised around internal teams rather than the questions buyers ask, pages that half-answered the same question so neither was cited, and machine-readable data that declared a name, a logo and nothing else. Read the story The redirect from northrow.com was technically correct, but nothing on either site told Google or the AI engines that NorthRow and ID-Pal were the same company. NorthRow's years of authority sat outside the merger, and AI was still recommending NorthRow as a separate platform. Read the story See what a readout looks like Compare the first steps ### Which step first? What you get Growth Review Paid Discovery Growth Plan Price - Free - £499, credited - £1,500, credited Time - 30 minutes plus the snapshot and the fixes - One week - Two weeks Found - Score - Every finding - In the plan Named by AI - Score and one recorded answer - 36 or more recorded answers - In the plan Website - Score - Page by page - In the plan Authority footprint - Score - Audit and plan - In the plan Pipeline - Score - Readiness check - In the plan What you leave with - Three fixes, ranked - Every finding, ranked, with the fix written - A 90-day plan with costs and owners Who does the work after - You, or us - You, any web team, or us - Us No card on the site: we invoice on day one, VAT added. Questions founders ask ### The short answers. Who is the Paid Discovery for? Founder-led B2B firms that want the full list of what stops buyers choosing them, across Google, AI assistants, their own site, their reputation elsewhere and their pipeline, before committing to anything monthly. What is in the readout? Ten chapters, about twenty-five pages, under five lenses: Found, Named by AI, Website, Authority and Pipeline. Buyer questions, recorded AI answers, who appears instead of you, what Google returns, every page of your site, your authority footprint in sixteen places, a content plan, pipeline readiness, the ranked list of every finding, and a baseline we re-measure free at ninety days. How long does it take and what does it cost? About a week from kick-off to the readout, for £499 plus VAT. No card on the site: we invoice on day one, VAT added. The fee is credited against your first month if you carry on with us. Can we do the work ourselves afterwards? Yes. Every fix is written for a competent team, and nothing in it depends on us. TCC's own team built the new tcc.group to the plan. What is the difference between this and the Growth Review? The Growth Review gives you five scores and the three fixes we would make first. The Paid Discovery gives you every finding under all five lenses, ranked, with each fix written out. ### Every finding, ranked and yours. One week. £499, invoiced on day one with VAT added, and credited against your first month. Ten chapters your team can build from. Ask for a Paid Discovery --- Title: Growth Review: five scores, three fixes | Fortitude Media URL: https://fortitudemedia.ai/growth-review Summary: A free 30-minute call with David Adams. We score your firm on five lenses, from Google to AI answers, and send the three fixes we would make first. ## Thirty minutes with a founder. Three fixes in writing. Before we speak we look at your site, what Google and ChatGPT say when your buyers ask who to use, where your firm is talked about, and two competitors. You leave with the three things we would fix first, in writing. It is free, and you can stop there. Pick a time Sent the day before: a one-page snapshot of what we found. 30 minutes Free Three fixes in writing What we found, talked through plainly ### What is a Growth Review? A free 30-minute call with David Adams, co-founder of Fortitude Media, for founder-led B2B firms with £1m to £20m+ revenue, where the founder still decides. Beforehand we score your firm on five lenses: Found, Named by AI, Website, Authority and Pipeline. Afterwards you get the three highest-value fixes in writing. There is no obligation to go further. The day before we speak ### One page. What we found. Five scores, the buyer questions where you are not named, and who is. Sent the day before the call so the thirty minutes start warm. BEFORE WE SPEAK · EXAMPLE · A MANAGED IT FIRM, £6M PAGE 1 OF 1 - Found 4 /10 Ranks for its own name and little else. - Named by AI 1 /10 Named in 1 of 40 recorded answers. - Website 5 /10 Describes the firm, not the buyer. - Authority 2 /10 No press, 3 reviews, no Wikidata. - Pipeline 3 /10 Founder LinkedIn last posted May. Best managed IT provider for law firms in Manchester ChatGPT names three firms. None is you. Two of the three have a page answering that exact question and a review profile with more than twenty reviews. The third was in the Law Society Gazette in March. ChatGPT · 12 September 2026 · UK Best managed IT provider for law firms in Manchester? Firms that come up repeatedly for law-firm IT in Manchester include Pennine Systems (strong on SRA compliance and Lexcel support, with a dedicated legal practice team), Ashworth Legal IT (specialists in case management system support and Cyber Essentials Plus) and a national provider with a Manchester office. Pennine and Ashworth both publish client reviews and have been covered in the Law Society Gazette. Example figures for a fictional firm. - Before We do the homework - The call 30 minutes, no deck - After Three fixes in writing, two working days What you get ### What three fixes look like. Specific, ranked by value, and yours to use whether you work with us or not. Example · written for a fictional managed IT firm - Fix 1 · Two weeks · Any web team #### Answer the question you lose. One page: Managed IT for law firms in Manchester, with the SRA and Lexcel points buyers check, three named clients, and a price band. Today the nearest page is Services, which answers nothing. - Fix 2 · One month · You In full after the call - Fix 3 · This quarter · Fortitude or you In full after the call Yours in full, two working days after the call. Under them: the ten chapters the Paid Discovery would fill in. See a full one ### The Growth Review, in full. A seven-page sample for a fictional managed IT firm: the one-page summary written to the founder, the five scores, where buyers ask and who is named, one recorded AI answer with the reasons, and the three fixes in writing. This is exactly what you would receive. Read it online Download the PDF 7 pages · PDF and online Fictional firm. Real format. Yours to keep. Page 1 of 7 Both samples, side by side Growth Review ### Book your Growth Review Book your Growth Review ### Thirty minutes. Pick a time. With David Adams, co-founder and CEO. You leave with three fixes in writing. - Invitation lands in your diary at once - A founder, not an account manager - One email afterwards, from David. No sequences. Bring one service you want to grow and the names of two competitors. That is all we need. Loading David’s diary Open the diary on Calendly Rather write than book? Send us a note instead. Founders we have worked with include arum, Helm Club, TCC Group and ID-Pal. Who you will speak to ### Founders, not account managers. You deal with people who have built, scaled and sold businesses of their own. A founder owns your plan from the first call. Co-founder and CEO · AI and automation #### David Adams David spent thirteen years in SaaS, starting on retention calls and finishing as COO and then CRO of a £30 million partner-led business. He built a network of 1,500 partners and grew an ad business from nothing to £2 million a year. Today he runs Fortitude with Ashley, designs the automation that lets a firm grow revenue without growing headcount, and runs Pipeline That Fills. Meet David If you want the whole list ### The Paid Discovery. Every finding under all five lenses, ranked, with the fix written so your own team could ship it. One week, £499, credited against anything you do with us afterwards, and yours to keep either way. Ten chapters, about twenty-five pages. What the Discovery covers Questions founders ask ### The short answers. Is it really free? Yes. No card, no contract, and no obligation to go further. Who is it for? Founder-led B2B firms with £1m to £20m+ revenue, where the founder still decides. What do I need to prepare? One service you want to grow and two competitors. We do the rest. How do I book? Pick a time in David's diary on this page and the invitation lands in your diary at once. If you would rather write than book, send us a note instead and David replies within one working day. What is the difference between this and the Paid Discovery? The Growth Review gives you five scores and the three fixes we would make first. The Paid Discovery gives you every finding under all five lenses, ranked, with each fix written out. Do you work on results? We will consider it for programmes where we agree the number up front and can measure it ourselves. Ask on the call. What happens afterwards? You get three fixes in writing. If you want the whole list, the Paid Discovery is £499. If you want a plan, the Growth Plan is £1,500. Both are credited against a programme. You can stop after any of them. ### Growth favours the prepared. Thirty minutes with a founder. Bring one service you want to grow. Leave with the three things we would fix first. Pick a time --- Title: What it costs: five steps, one page | Fortitude Media URL: https://fortitudemedia.ai/what-it-costs Summary: Every fee on one page: free Growth Review, £499 Paid Discovery, £1,500 Growth Plan, focus programmes from £2,500, full programmes £5,000 to £8,000 a month. What it costs ## Start small. Stop whenever you like. Five steps. Each one is useful on its own, and each one is credited against the next. All fees exclude VAT. ### The five steps and their fees - Step 1 #### Growth Review Free 30 minutes with a founder. Three fixes, in writing. - Step 2 · optional #### Paid Discovery £499 One week. Every finding under five lenses, ranked. Yours to keep. - What it covers - Step 3 #### Growth Plan £1,500 Two weeks. One plan, one page. Credited against a programme. - Step 4 #### Focus programme from £2,500 a month One problem, one team, one month at a time. - Step 5 #### Full programme £5,000 to £8,000 a month Website, reputation and pipeline, run by the founders. See a sample Growth Review and Paid Discovery Results-based terms ### Prefer to share the risk? Book a Growth Review We will consider results-based terms where we agree the number up front and can measure it ourselves. Ask us on the call. The alternative ### The alternative is hiring. Here is that bill. To do this in-house you need about four people. Salary is only the start. Then tax, pension, tools and a recruiter. Then you manage them. All our fees exclude VAT. Option 1 · Build the team yourself #### Four hires. Twelve months. - Marketing manager - £56,950 - Content and search manager - £42,055 - Marketing executive - £35,046 - Sales development rep, base pay only - £32,000 - Employer National Insurance at 15% - £21,908 - Minimum employer pension at 3% - £4,032 - Software and data they need to do the job - £12,000 - Every year - £203,991 - Plus in year one: recruiter fees at 20% and laptops - + £39,210 Option 2 · Fortitude #### One senior team. One monthly fee. £5,000 to £8,000 a month for most clients. Focus programmes from £2,500. A free 30-minute review and a £1,500 plan come first, and you can stop after either. - One senior team and the tools we built, not four generalists - Working within weeks, not six months - Meetings priced per meeting held - A monthly report on what it made Book a Growth Review Compare your options Hiring a marketing team A lead generation agency An SEO agency Doing nothing Your break-even, not our promise A firm whose average client is worth £15,000 at a 50% margin needs about ten new clients a year to cover a £6,500-a-month programme. Bring your own figures and we will do the sum in front of you, before you spend anything. Salaries are 2026 UK medians from the Ashdown Group guide; employer NI at 15%, pension at 3%, recruiter fees at 20%. Commission, training and your own time are left out. Questions founders ask ### The short answers. Do fees include VAT? No, all fees exclude VAT. Can I stop after any step? Yes. What does “credited” mean? The Growth Plan fee comes off the first month of a programme. Do I keep the Paid Discovery findings if I go no further? Yes, the readout is yours to keep and to implement yourselves. Do you offer results-based terms? We will consider them where we can measure the result ourselves. ### Growth favours the prepared. Thirty minutes with a founder. Bring one service you want to grow. Leave with the three things we would fix first. Book a Growth Review --- Title: Client work and results | Fortitude Media URL: https://fortitudemedia.ai/work Summary: What changed for four founder-led firms: one in five new members via AI, 0 to 4 UK AI shortlists, a 310-page website in weeks, and an AI audit in a week. ## Four founders. Four problems. What changed, in their numbers. Every client named here approved every word we say about them. - 01 Helm Club The club had the members. It needed the calls. 1 in 5 - 02 ID-Pal AI could describe ID-Pal. It never recommended it. 0 4 - 03 arum A 28-year-old firm's new website. Built as if it were our own. 310 - 04 TCC Group Three brands, one group. AI could only see the pieces. 1 week 01 · HELM CLUB · Pipeline and the full programme ### The club had the members. It needed the calls. 1 in 5 new members this year came through an AI recommendation 2,000 well-matched founders approached every month, each checked against their live LinkedIn profile 10 comparison pages, one for every question a founder asks when choosing a peer group Read the Helm Club story www.helmclub.co 02 · ID-PAL · Reputation That Wins ### AI could describe ID‑Pal. It never recommended it. 0 4 UK buyer questions where AI now names ID-Pal 6 1 AI answers still recommending NorthRow as a rival, May to September. The authority is coming home 6 of 6 correct answers to "what does ID-Pal do?", every engine, every run Read the ID-Pal story www.id-pal.com Dan Brooksbank Marketing Director, ID-Pal Read the full story > Fortitude undertook specialist research and identified the primary cause within the first week of working together and had the fix live in the first month. 03 · ARUM · Website That Sells ### A 28-year-old firm's new website. Built as if it were our own. 310 pages rebuilt across three regions, in weeks 215 articles, research pieces and news items moved across, going back to 2014 3 regional versions: UK and Europe, North America, Australia and New Zealand Read the arum story www.arum-global.com Jamie Waller CEO and Executive Chair, arum Read the full story > It was obvious from day one that they are founders themselves. As CEO, the best thing I can say is that they made my life easier, not harder. 04 · TCC GROUP · Paid Discovery ### Three brands, one group. AI could only see the pieces. 1 week to a plan TCC's own team built the new site from 36 recorded answers from ChatGPT, Gemini and Claude, every conversation kept and reproducible 624 pages reviewed, each given a decision: migrate, merge or retire Read the TCC Group story tcc.group Our clients. One thing they all said. ### They work like founders. > “They work like founders: they get on with it, and they tell you the truth.” Dan Brooksbank, Marketing Director, ID-Pal > “It was obvious from day one that they are founders themselves.” Jamie Waller, CEO and Executive Chair, arum ### Your firm could be the fifth story. Start with thirty minutes. Book a Growth Review Or start with a £499 Paid Discovery --- Title: arum: a 310-page website in weeks | Fortitude Media URL: https://fortitudemedia.ai/work/arum Summary: How Fortitude Media rebuilt arum's website across three regions: 310 pages, every word rewritten, a sharper brand, and built to be found by search and AI. arum · Website That Sells ## A 28-year-old firm's new website. Built as if it were our own. arum advises banks, governments and some of the world's biggest brands. Its website did not show it. We rebuilt all of it, from the words to the brand to how it is found, in weeks rather than months. Book a Growth Review Visit www.arum-global.com At a glance 3 min read 310 pages rebuilt across three regions, in weeks **The problem** - A 28-year track record, more than 500 projects in 23 countries, and a website that undersold it. **What we did** - Rewrote every word, sharpened the brand and rebuilt 310 pages across three regions. **The result** - A website that finally does the firm justice, delivered in weeks rather than months. Key facts Client arum Sector Collections and recoveries What we did Content, brand, design, build Headline result 310 pages rebuilt across three regions, in weeks Scale 310 pages, three regions www.arum-global.com ### What we built 310 pages in one design system 215 articles and research pieces moved across 14 technology pages 12 client case studies 9 sector pages 8 service pages 3 regional versions What we built pages in one design system - 310 articles and research pieces moved across - 215 technology pages - 14 client case studies - 12 sector pages - 9 service pages - 8 regional versions - 3 Every old address redirected, so years of rankings carried over. The problem ### A 28-year track record. A website that undersold it. arum has helped banks, governments and major brands choose, implement, rescue and get more from their collections technology for 28 years: more than 500 projects in 23 countries, from offices in London, Atlanta and Sydney. The old website did not feel like that business. The story was hard to follow, the services were hard to tell apart, and the brand was not working hard enough on a screen. For most people, the website is the first time they meet arum. Jamie Waller, arum's CEO, had a short brief. Make the website as good as the firm, and do not make it a long agency process. The brief Make the website as good as the firm. What we did ### Every word rewritten. A sharper brand. Built to be found. 01 #### The content All of it reworked in plain English, outcomes first. Services organised the way a client buys them: five stages from advice to optimisation, plus three specialist services. 02 #### The brand Same arum, sharper on screen. A deep navy base, one bright colour saved for the button that matters, large confident headlines and clear rules so every new page belongs. 03 #### Proof up front Client logos, accreditations and named testimonials in the first screen. Each of the 12 case studies leads with a number, and each of the nine sector pages names the regulation in play. 04 #### Built to be found Proper structure and markup on every page, a plain summary AI assistants can read, facts stored once so two pages never disagree, and every old address redirected so years of rankings carry over. 05 #### The CEO's time, protected No workshops and no weekly status meetings. Jamie reacted to real, working pages within weeks. When arum's team sent a final list of 77 changes, we checked each one against the live site before touching a word. www.arum-global.com arum's rescue service page, one of eight rewritten service pages - Services organised the way a client buys them: five stages - One bright colour saved for the button that matters - Proof up front What changed ### A website that finally does the firm justice. One site in three regional versions, with local spelling, local regulation and the nearest office for each visitor, and every enquiry routed straight into arum's CRM for the right team. Eight rewritten service pages, nine sector pages, 12 case studies, a vendor marketplace and 14 technology pages, all in one design system that arum's team can keep adding to. 310 pages rebuilt across three regions, in weeks Jamie Waller CEO and Executive Chair, arum Share on LinkedIn > I have worked with plenty of agencies over the years, and Fortitude Media is not like any of them. It was obvious from day one that they are founders themselves. They did not ask for a fifty page brief or wait to be told what to do. They got to grips with our business, took ownership of the website as if arum were their own company, and made smart calls along the way. They reworked all of our content, sharpened the brand so it works properly online, and did it at a speed I did not think was possible. For most people, our website is the first time they meet arum. It now does us justice. As CEO, the best thing I can say is that they made my life easier, not harder. Jamie Waller, CEO and Executive Chair, arum ### If this sounds like your firm. The service Website That Sells A website that brings in enquiries, built for Google and AI answers and improved every month. Read more Sound familiar? - Our website undersells us. - We rebuilt it two years ago and nothing changed. Start here Growth Review · free · thirty minutes Thirty minutes with a founder. You leave with three fixes in writing, whether or not we work together. Book a Growth Review More client work ### Three more founders. Three more numbers. 1 in 5 new members this year came through an AI recommendation #### The club had the members. Read the story 0 4 UK buyer questions where AI now names ID-Pal #### AI could describe ID-Pal. Read the story 1 week to a plan TCC's own team built the new site from #### Three brands, one group. Read the story ### Is your website underselling your firm? Book a Growth Review. Thirty minutes with a founder. We look at your site, two competitors and what Google and ChatGPT say about all three beforehand. You leave with three fixes in writing. Book a Growth Review --- Title: Helm Club: paid per member added | Fortitude Media URL: https://fortitudemedia.ai/work/helm-club Summary: How Fortitude Media took over Helm Club's growth as a fractional mandate paid on members added, running outbound, paid media, search and AI answers. Helm Club · Pipeline and the full programme ## The club had the members. It needed the calls. Up to half the founders who got on a call with Helm Club joined. Too few got on a call. We took ownership of every route to that conversation, and now one in five new members arrives because an AI assistant recommended Helm. Book a Growth Review Visit www.helmclub.co At a glance 3 min read 1 in 5 new members this year came through an AI recommendation **The problem** - Between 40 and 50 per cent of founders who took a call joined. Only about four qualified calls a week reached the diary. **What we did** - Took ownership of every route to a call, from LinkedIn and search to AI assistants, paid per member added. **The result** - One in five new members now arrives through an AI recommendation. Key facts Client Helm Club Sector Founder membership community What we did Outbound, paid media, search and AI answers, content Headline result 1 in 5 new members this year came through an AI recommendation Terms Fractional mandate, paid on members added www.helmclub.co ### Qualified calls a week When we started 4 Target, end of November 15 Target Qualified calls a week When we started - 4 Target, end of November - 15 The target is a target, not a result. Our fee is paid per member added above the replacement line. 1 in 5 new members this year came through an AI assistant recommending Helm 2,000 well-matched founders approached every month, each checked against their live LinkedIn profile 10 comparison pages, one for every question a founder asks when choosing a peer group The problem ### A proven sale. Not enough people to sell to. Helm Club brings founders of growing UK businesses together in chaired peer groups. Members stay for years, and between 40 and 50 per cent of founders who took a call went on to join. The sale was never the issue. The issue was volume. About four qualified calls a week reached the diary, and every month the club added nearly as many members as it lost. Every channel worked on its own, nobody owned the whole picture, and effort went wherever the last good month had come from. Andreas Adamides, Helm's CEO, did not want a marketing agency and a monthly deck. He wanted one person to own the number. The brief Take ownership of qualified calls a week, and be paid on the members that result. What we did ### Every route to a call, rebuilt and run as one system. 01 Every Monday #### One number, owned Qualified calls a week, reported on one page every Monday and checked against Helm's own member records rather than a platform dashboard. Our fee is paid per member above the level that only replaces churn, so we earn when Helm grows. 02 Every month #### Outbound at a human pace Around 2,000 well-matched founders approached a month across three LinkedIn accounts, each checked against their live profile first. The first message asks a question. No pitch, no link, and every reply gets a same-day answer from a person. 03 #### Paid media under one plan Google, LinkedIn and Meta run together, with a lead form cut back to one question and ads that say who the club is for, so the ad does some of the filtering. 04 #### Made Helm the answer We built the structured data AI engines read, then a comparison page for every question a founder asks when weighing up peer groups: Helm against YPO, Vistage and the rest. Ten pillar articles a month double as landing pages for paid search. 05 Every month #### The CRM, worked properly Every enquiry has one owner and a same-day follow-up. Founders who took a call and never joined, and lapsed members, are worked again every month rather than in a one-off sweep. www.helmclub.co Helm against Vistage: one of ten comparison pages we built - One page for every question a founder asks when choosing a peer group - Every comparison page we have published ranks first for its question What changed ### More calls, and Helm can see where they come from. A channel that did not exist before now brings one in five new members: founders who ask ChatGPT, Gemini or Claude which peer group to join are told about Helm. Every comparison page we have published ranks first for its question on Google and the AI engines. The pages not yet built are the ones where a competitor still writes the answer. Qualified calls a week started at four, and the target is fifteen by the end of November. Andreas reads one page a week, and every channel is judged on the same thing: booked calls. 1 in 5 new members this year came through an AI recommendation ### If this sounds like your firm. The service Pipeline That Fills Researched outreach, written and approved by a person, so qualified meetings reach the diary. Read more Sound familiar? - We've outgrown word of mouth. - Pipeline is feast or famine. Start here Growth Review · free · thirty minutes Thirty minutes with a founder. You leave with three fixes in writing, whether or not we work together. Book a Growth Review More client work ### Three more founders. Three more numbers. 0 4 UK buyer questions where AI now names ID-Pal #### AI could describe ID-Pal. Read the story 310 pages rebuilt across three regions, in weeks #### A 28-year-old firm's new website. Read the story 1 week to a plan TCC's own team built the new site from #### Three brands, one group. Read the story ### Does your sale work but your pipeline doesn't? Book a Growth Review. Thirty minutes with a founder. We look at your site, two competitors and what Google and ChatGPT say about all three beforehand. You leave with three fixes in writing. Book a Growth Review --- Title: ID-Pal: from 0 to 4 UK AI shortlists | Fortitude Media URL: https://fortitudemedia.ai/work/id-pal Summary: How Fortitude Media found the cause of ID-Pal's traffic drop in a week, fixed it in the first month, and got it named in AI answers to UK buyer questions. ID-Pal · Reputation That Wins ## AI could describe ID-Pal. It never recommended it. Ask an AI assistant what ID-Pal does and it gets it right. Ask which platform a UK wealth manager should shortlist and, in May 2026, ID-Pal did not appear once. Four months later it is being named on UK buyer questions for the first time. Book a Growth Review Visit www.id-pal.com At a glance 3 min read 0 4 UK buyer questions where AI now names ID-Pal **The problem** - Traffic to the pages that bring in enquiries fell by around 40 per cent after the NorthRow move, and AI never put ID-Pal on a UK shortlist. **What we did** - A Paid Discovery found the cause in a week. The fix was live in week one, then we rebuilt the pages buyers ask about. **The result** - Named on four UK buyer questions by September 2026, up from none in May. Key facts Client ID-Pal Sector Identity verification (KYC and KYB) What we did Paid Discovery, structure, content, outreach Headline result 0 4 UK buyer questions where AI now names ID-Pal Terms Paid Discovery, then a monthly programme When May to September 2026 www.id-pal.com ### UK buyer questions where AI names ID-Pal 0 May 2026 2 August 2026 4 September 2026 UK buyer questions where AI names ID-Pal May 2026 - 0 August 2026 - 2 September 2026 - 4 #### AI answers still recommending NorthRow as a rival May 2026 6 September 2026 1 AI answers still recommending NorthRow as a rival May 2026 - 6 September 2026 - 1 Same questions, same three engines, a fresh chat every month. A win only counts when it repeats. 6 of 6 correct answers to "what does ID-Pal do?", every engine, every run The problem ### Recognised everywhere. Shortlisted nowhere. ID-Pal verifies identities for fund administrators, wealth managers, law firms and accountants across Ireland, the UK and the US. It had recently bought NorthRow, a UK business verification platform, and folded it into id-pal.com. Dan Brooksbank, Marketing Director, came to us with a live problem: traffic to the pages that bring in enquiries had fallen by around 40 per cent since the NorthRow move, and nobody could say why. The brief Fix the leak, then build the authority, and measure both in the open every month. What we found ### What the Paid Discovery found, in a week The redirect from northrow.com was technically correct, but nothing on either site told Google or the AI engines that NorthRow and ID-Pal were the same company. NorthRow's years of authority sat outside the merger, and AI was still recommending NorthRow as a separate platform. The bigger finding was structural. Across 36 recorded AI answers to the questions ID-Pal's buyers actually ask, ID-Pal was named three times, and only for its brand or its Salesforce integration. AI knew who it was, and had no reason to recommend it. What we did ### Fix the leak first. Then build the authority. 01 Week one #### The NorthRow fix About four hours of structured data work declaring NorthRow part of ID-Pal on both domains. Live in week one, with no copy changes and no risk to the redirect. 02 #### Five industry pages rebuilt Fund administration, wealth management, legal, accountancy and estate agency. Each opens with the question a compliance lead types, names the regulator, and answers inside the first ninety words. 03 Every month #### Content that gets cited Six expert pieces a month, a KYB guide and pages for the UK sectors where specialists were winning, every piece signed off in batches and built ready to publish. 04 #### The rest of the web Every directory and list the AI engines cite for identity verification mapped, with ID-Pal claimed and corrected on each. 05 Every month #### Measured in the open The same buyer questions run every month in ChatGPT, Gemini and Claude, in a fresh chat. A win only counts when it repeats, and losses are reported too. www.id-pal.com The page Gemini now cites for UK wealth managers - Rebuilt around the question a UK compliance lead types - The answer inside the first ninety words What changed ### First UK wins, and the leak closed. Gemini now names ID-Pal for UK wealth managers. Claude names it for UK fund administrators, on a page rebuilt in August, and links NorthRow to ID-Pal in the same answer. The engines that were recommending NorthRow as a rival now attribute it to ID-Pal. Where a page has been rebuilt, ID-Pal appears. Where it has not yet, it does not, which tells us exactly what to build next. 0 4 UK buyer questions where AI now names ID-Pal Dan Brooksbank Marketing Director, ID-Pal Share on LinkedIn > We had a need to resolve visibility for categories where AI was not mentioned our brand. Fortitude undertook specialist research and identified the primary cause within the first week of working together and had the fix live in the first month. In terms of the longer-term growth outlook, Fortitude have been honest with timelines and the clear success criteria they would judge themselves by, already delivering visibility across geo shortlists we had never previously appeared on. They work like founders: they get on with it, and they tell you the truth. Dan Brooksbank, Marketing Director, ID-Pal ### If this sounds like your firm. The service Reputation That Wins Be one of the names buyers are given when they ask Google or an AI assistant who to use. Read more Paid Discovery In one week, everything that stops buyers choosing you, from Google to AI answers, ranked with the fix written out. Read more Sound familiar? - ChatGPT has never heard of us. - Competitors rank above us. They're worse. Start here Paid Discovery · £499 · one week In one week, everything that stops buyers choosing you, from Google to AI answers, ranked with the fix written out. Ask for a Paid Discovery Or start with a free Growth Review More client work ### Three more founders. Three more numbers. 1 in 5 new members this year came through an AI recommendation #### The club had the members. Read the story 310 pages rebuilt across three regions, in weeks #### A 28-year-old firm's new website. Read the story 1 week to a plan TCC's own team built the new site from #### Three brands, one group. Read the story ### Can AI describe you but never recommend you? Book a Growth Review. Thirty minutes with a founder. We look at your site, two competitors and what Google and ChatGPT say about all three beforehand. You leave with three fixes in writing. Book a Growth Review --- Title: TCC Group: one group AI can see | Fortitude Media URL: https://fortitudemedia.ai/work/tcc-group Summary: How a one-week Paid Discovery showed TCC Group where AI search had stopped naming it, and gave its own team a blueprint to build the new tcc.group from. TCC Group · Paid Discovery ## Three brands, one group. AI could only see the pieces. AI assistants rated TCC Group highly where they knew it. On the two areas the group invests in most, they did not mention it at all. We found out why in a week, and handed over a plan TCC's own team used to build the new tcc.group. Book a Growth Review Visit tcc.group At a glance 3 min read 1 week to a plan TCC's own team built the new site from **The problem** - AI rated TCC highly where it knew it, but on Consumer Duty and Section 166 it named other firms. **What we did** - A one-week Paid Discovery: 36 recorded AI answers, 624 pages reviewed and a blueprint for the new site. **The result** - TCC's own team built the new tcc.group to the plan. Key facts Client TCC Group Sector Regulatory compliance, UK financial services What we did Paid Discovery, site blueprint, technical specification Headline result 1 week to a plan TCC's own team built the new site from Scale 36 AI answers, 624 pages reviewed tcc.group ### All 624 pages, each given a decision 247 26 351 - 247 Migrate - 26 Retire - 351 Settled in one working hour against analytics All 624 pages, each given a decision Migrate - 247 Retire - 26 Settled in one working hour against analytics - 351 The best-ranking pages were flagged for protection through the move. 1 week from kick-off to a readout the Executive Chair could act on 36 recorded answers from ChatGPT, Gemini and Claude, every conversation kept and reproducible The problem ### Rated highly where AI knew it. Invisible where it mattered most. TCC Group has advised regulated firms for more than two decades across three brands: TCC for advisory and managed services, Recordsure for compliance AI, and Momenta for specialist resourcing. ChatGPT already placed TCC top for outsourced suitability reviews. But on Consumer Duty and Section 166, the two areas the group backs hardest, the assistants named other firms and not TCC. One had stopped recognising the UK firm altogether, and none of them connected the three brands. Joanne Smith, Executive Chair, commissioned a Paid Discovery to find out exactly what AI search was saying about the group, and why. The brief Find out exactly what AI search says about TCC and why, and give the group a plan its own team could build from. What we found ### What the Paid Discovery found It was not a content problem. TCC's team was publishing good material and one assistant was already leading with it. What failed was underneath: a site organised around internal teams rather than the questions buyers ask, pages that half-answered the same question so neither was cited, and machine-readable data that declared a name, a logo and nothing else. What we did ### A plan written to be built by anyone. 01 #### The structure Ten solution pages, one for each problem a buyer types, from Section 166 and FCA intervention to vulnerable customers, plus sector hubs and one group page telling the story once. 02 #### Three page templates Specified block by block: the question as the heading, an answer inside ninety words, dated facts, FAQ markup and a named practice lead, so every future page is ready for AI the day it is published. 03 #### The page register A decision for every one of the 624 pages, with the best-ranking pages flagged for protection through the move. 04 Week one #### Week-one assets A machine-readable description of the group for all three sites, llms.txt, case study markup and a monitoring file with the exact questions, scoring rules and baseline. 05 #### To the client's standard TCC's technologists set the bar and we wrote it into the specification: secure by design, WCAG 2.2 AA accessibility, consent, HubSpot integration, and TCC owning every credential. tcc.group Built by TCC's own team, to the blueprint - Opens with the buyer's question, as they would type it - One solution page for each problem a buyer types What changed ### Built in-house, to the blueprint. The Discovery gave TCC three ways forward: build with us, take the plan and build it in-house, or stop. TCC chose to build it with its own team, and the new tcc.group is live, built to the plan, with nothing about it depending on Fortitude. Ten solution pages and nine sector hubs now open with the buyer's question, and the group is declared as one entity with three connected brands. The baseline test re-runs every month, and an improvement only counts when it holds across consecutive runs. 1 week to a plan TCC's own team built the new site from ### If this sounds like your firm. The service Paid Discovery In one week, everything that stops buyers choosing you, from Google to AI answers, ranked with the fix written out. Read more Reputation That Wins Be one of the names buyers are given when they ask Google or an AI assistant who to use. Read more Sound familiar? - ChatGPT has never heard of us. - Competitors rank above us. They're worse. Start here Paid Discovery · £499 · one week In one week, everything that stops buyers choosing you, from Google to AI answers, ranked with the fix written out. Ask for a Paid Discovery Or start with a free Growth Review More client work ### Three more founders. Three more numbers. 1 in 5 new members this year came through an AI recommendation #### The club had the members. Read the story 0 4 UK buyer questions where AI now names ID-Pal #### AI could describe ID-Pal. Read the story 310 pages rebuilt across three regions, in weeks #### A 28-year-old firm's new website. Read the story ### Want the plan without the agency? Book a Growth Review. Thirty minutes with a founder. We look at your site, two competitors and what Google and ChatGPT say about all three beforehand. You leave with three fixes in writing. Book a Growth Review --- Title: The Fortitude Media team URL: https://fortitudemedia.ai/team Summary: David Adams, Ashley Kenerson and Ross Williams have built, scaled and sold their own firms. A founder owns your plan; a senior team delivers it. ## Built by founders who have sat in your chair. The growth firm we wished we could hire. You deal with people who have built, scaled and sold businesses of their own, not account managers who have only ever worked in agencies. A founder owns your plan. A senior team delivers it. The technology we built does the heavy lifting behind them. Book a Growth Review How we work Why Fortitude exists The Fortitude studio ### Who does the work at Fortitude Media? Fortitude Media was founded by three operators: David Adams, Ashley Kenerson and Ross Williams. David and Ashley run it. A founder owns every client plan, runs the monthly review and answers the phone. A senior team of specialists delivers the work, and the technology we built does the research, the weekly tracking of AI answers, the first drafts, the sequencing, the chasing and the reporting. A person reads every output before a client, a prospect or a search engine sees it. 30 years building and selling internet businesses 20m+ members on platforms our founders built $350m revenue generated, almost all of it from being found online 5,000+ brands tracked by our own AI search platform Our founders' record. The founders ### Operators, not theorists. We have built, scaled and sold our own companies. We run Fortitude's growth on the same system we sell. A founder is accountable for every account. Co-founder and CEO · AI and automation 34 guides on websites and AI search #### David Adams David spent thirteen years in SaaS, starting on retention calls and finishing as COO and then CRO of a £30 million partner-led business. He built a network of 1,500 partners and grew an ad business from nothing to £2 million a year. Today he runs Fortitude with Ashley, designs the automation that lets a firm grow revenue without growing headcount, and runs Pipeline That Fills. Meet David Co-founder · Data and insight 32 guides on data, AI and reporting #### Ashley Kenerson Ashley spent thirteen years at Digitas, rising from analyst to Head of Data, Analysis and Insights and building a team of nearly fifty across Europe and Asia. Her job was to say plainly what marketing was working, what was not, and what to do next. She runs Fortitude with David, designs the measurement behind every report, and keeps the people in charge of the technology. Meet Ashley Co-founder, non-executive · The founder's record 26 guides on buying, PR and sectors #### Ross Williams Built and ran a subscription business for twenty-one years, bootstrapped. Backs David and Ashley, and established the Foundation. Meet Ross Behind the team ### The tools we built. People in charge of them. Think of a pit crew. The driver gets the credit; the crew gets the car round in under three seconds. Our technology is part of that crew. It does the work that used to need four juniors, so senior people spend their time on judgement, and a person reads every output before a client sees it. Research Builds target lists and runs the two-pass check on every firm and person before contact. Visibility Asks ChatGPT, Claude, Gemini and Google your buyers' questions every week and records who is named. Writing Drafts pages, articles and posts from interviews and research. Never sends. A founder edits. Outreach messages are written by a person. Outreach Finds and checks every firm and person, times the approach, runs the email sequences, sorts replies and proposes the next step. Follow-up Chases every open thread, quote and enquiry so nothing goes cold by Friday. Reporting Compiles the monthly board report from live numbers, with a source for every figure. A person reads every output. Nothing reaches a client, a prospect or a search engine unread. Software never writes to a prospect. Research, timing and follow-up reminders are automated. The words are a person's, and a person approves every message before it goes. Every number has a source. Each figure in a report shows where it came from. If it cannot, it does not go in. Every corner at once. That is the job our technology does. Our own growth, run on the same system Proof we use it #### We run Fortitude on exactly this. This site is a managed website, tracked by our own visibility platform. Our founders' LinkedIn is run the way we run yours. Our pipeline comes from Pipeline That Fills. If you received an email from us, that is the system working: it mentioned something true about your firm because that is how every message is written. - Our own technology. The visibility platform that tracks 5,000+ brands, and the AI-first CMS this site runs on, were built by us. - Our own outreach. If you received an email from us, it mentioned something true about your firm. That is the system working. What that means for you ### Senior people. One plan. One page. Every account 1 Founder accountable. Owns the plan, runs the review, answers the phone. 1 Plan. Website, reputation and pipeline run together, with costs and owners. Every month 1 Page. Pipeline in. Work shipped. Answers won. Written for a board. The Foundation #### Ten per cent of everything we invoice. We have had the good years and the ones nobody puts on a website. The Fortitude Foundation was established by Ross to help entrepreneurs in crisis, and one of the reasons we started Fortitude Media was to fund it properly, for good. Ten per cent of all the revenue we invoice goes straight to the Foundation. - Revenue, not profit. Ten per cent of every invoice we raise, paid monthly, on every service we sell. - No cost to you. Our prices are the same either way. It comes out of our side of the invoice. - For founders in crisis. The help most founders never ask for, from people who have needed it. How the 10% works · The Fortitude Foundation Nobody gets through it on their own Book a Growth Review ### Book a Growth Review with a founder. Free. 30 minutes. Three fixes in writing. You speak to David Adams, co-founder and CEO. Book a Growth Review ### Growth favours the prepared. Thirty minutes with a founder. Bring one service you want to grow. Leave with the three things we would fix first. Book a Growth Review --- Title: Ashley Kenerson, co-founder of Fortitude Media URL: https://fortitudemedia.ai/team/ashley-kenerson Summary: Ashley Kenerson spent thirteen years at Digitas turning marketing data into decisions. She runs Fortitude Media with David Adams and designs its reporting. ASHLEY KENERSON · CO-FOUNDER Co-founder · Data and insight ## Ashley Kenerson. Data you can act on. Numbers you can trust. A hardcore data geek who happens to be a marketer. Thirteen years at Digitas turning marketing data into decisions, and now the measurement behind every Fortitude report. Book a Growth Review LinkedIn **Years at Digitas** - 13 **Team she built** - Nearly 50, across Europe and Asia **Sectors** - Financial services, travel, automotive **Lives in** - Portugal The story ### From analyst to head of the data practice. Ashley calls herself a hardcore data geek who happens to be a marketer. She spent thirteen years at Digitas, one of the largest digital agencies in the world, turning marketing data into decisions for financial services, travel and car brands. She joined Digitas in the United States as an analyst in 2006 and left London in 2019 as Head of Data, Analysis and Insights, having built the data practice outside North America into a team of nearly fifty people across Europe and Asia. Her job was to take everything a company's marketing produced, across every channel, and say plainly what was working, what was not, and what to do next. That is the discipline behind every Fortitude report. Ashley designs the measurement: the buyer questions we track each month, the rule for when a win counts, and the one page a founder reads on the first Monday. She keeps the numbers honest and the people in charge of the technology. She took four years away from client work to raise three young children, co-founded Guincho Partners in 2024, and then Fortitude, which she runs with David. She lives in Portugal. Ashley, 2026 At Fortitude ### What Ashley does on your account. 01 #### Designs the measurement The buyer questions we track, the three engines we ask, and the rule that a win only counts when it repeats. 02 #### Reads the report before you do Every first Monday. If a number is not moving, the report says so, and the monthly review decides what to stop. 03 #### Keeps people in charge The tools do the volume. Ashley makes sure a person reads every output before a client, a prospect or a search engine does. The record 13 years at Digitas, analyst to Head of Data, Analysis and Insights LinkedIn 50 people, nearly, in the data team she built across Europe and Asia LinkedIn 3 industries she measured marketing for: financial services, travel, automotive LinkedIn 1 page a month. The report every Fortitude client gets, designed by Ashley fortitudemedia.ai/samples/monthly-report ASHLEY KENERSON · CO-FOUNDER > “The tools do the volume. People do the judgement.” Ashley Kenerson · Guides ### Written by Ashley. Data, AI visibility, research and reporting. Every guide reviewed September 2026. Start with the three founders ask about most, or pick a topic. **Guides** - 32 **Topics** - 6 **Of reading** - 6 hrs Costs, returns and the business case How do we compare with our competitors when buyers ask ChatGPT? 13 min read Content that gets quoted Does publishing our own research help ChatGPT recommend us? 10 min read How buying is changing Why is our website traffic falling when our rankings have not? 11 min read #### Content that gets quoted - Can ChatGPT tell real expertise from keyword-stuffed content? 11 min - Does ChatGPT prefer long articles or short ones? 11 min - Does ChatGPT prefer newer content, and how old is too old? 12 min - Does publishing our own research help ChatGPT recommend us? 10 min - How do we audit our existing content for AI assistants? 11 min - Should we build a glossary so AI assistants use us as the reference? 10 min - What does an article that ChatGPT actually quotes look like? 10 min - Why does ChatGPT never quote our blog, and how do we fix it? 11 min #### How buying is changing - What replaces the ten blue links on Google, and what does it mean for us? 9 min - How do I prepare for AI agents that buy on behalf of my customers? 12 min - Is being named by ChatGPT cheaper than paying for Google Ads? 12 min - We rank first on Google. Why does ChatGPT not mention us? 12 min - What happens when buyers let an AI agent build the shortlist? 11 min - Which AI assistants do buyers actually use, and how do they differ? 11 min - Why is our website traffic falling when our rankings have not? 11 min #### Sector guides - How does a B2B tech company get recommended when buyers ask ChatGPT? 11 min - How does a financial planning firm get named when clients ask ChatGPT? 11 min - How does our clinic get recommended when patients ask ChatGPT? 13 min - How does our training company get recommended when HR asks ChatGPT? 11 min - Why does ChatGPT recommend other energy firms when businesses ask? 11 min - Why does ChatGPT recommend other insurance brokers and not us? 13 min #### Costs, returns and the business case - How can we tell which of our new leads came from ChatGPT? 10 min - How do we build a simple dashboard to track AI recommendations? 11 min - How do we compare with our competitors when buyers ask ChatGPT? 13 min - How do we find out whether ChatGPT is recommending our firm? 11 min - What should our monthly report on AI search show the board? 11 min #### Press, reviews and being trusted - Does ChatGPT trust press coverage more than our website or our ads? 12 min - How do we turn our own data into coverage in the trade press? 13 min - How does ChatGPT decide which facts about our firm to believe? 12 min - Which websites that mention us actually count with ChatGPT? 13 min - Why does ChatGPT sound unsure about us, and do backlinks fix it? 11 min #### Websites that bring work in 1 guide - How do we measure the return we get from our website? 11 min More of the team ### The other two founders. Co-founder and CEO · AI and automation #### David Adams Thirteen years inside a £30 million platform, from the phones to the boardroom. Now he runs Fortitude and builds the machine. Meet David Co-founder, non-executive · The founder's record #### Ross Williams Built and ran a subscription business for twenty-one years, bootstrapped. Backs David and Ashley, and established the Foundation. Meet Ross ### Thirty minutes with a founder. Three fixes in writing. Bring one service you want to grow. Leave with the three things we would fix first, in writing. Book a Growth Review --- Title: David Adams, co-founder and CEO of Fortitude Media URL: https://fortitudemedia.ai/team/david-adams Summary: David Adams spent thirteen years in SaaS, from retention calls to COO and CRO of a £30 million business. He runs Fortitude Media and builds the automation. DAVID ADAMS · CO-FOUNDER AND CEO Co-founder and CEO · AI and automation ## David Adams. Built the revenue machine. Now he automates it. Thirteen years turning a platform into a business, from his first job answering retention calls to running operations and revenue as COO and then CRO. At Fortitude he runs the business and owns your plan. Book a Growth Review LinkedIn davidadamsofficial.com Lucennio Consultancy **Years in SaaS** - 13 **Roles held** - COO, then CRO **Partners built** - 1,500+ **Works from** - London The story ### From retention calls to running revenue. David spent thirteen years turning a platform into a business. He started on the phones, answering retention calls, and finished as COO and then CRO of a partner-led SaaS company doing £30 million a year. He came up through every part of it: retention, partnerships, operations and sales. Along the way he built a network of more than 1,500 partners and affiliates, helped more than a thousand brands run on the platform, and grew an advertising network from nothing to more than £2 million a year. He knows what a sales team needs because he has run one, and what a founder needs because he has reported to one. The last few years have been about automation. David designs the systems that let a B2B firm grow revenue without adding people: research that runs overnight, outreach that a person still writes and approves, and a pipeline that reports itself. He is the reason Fortitude can hand the boring work to software and keep people on the part a buyer sees. He runs Fortitude with Ashley, from London. David, 2026 At Fortitude ### What David does on your account. 01 #### Owns the plan Writes it after the Growth Review, defends it at the monthly meeting, and changes it when the numbers say so. 02 #### Runs Pipeline That Fills The lists, the messages a person writes and approves, and the diary. He sets the bar for what counts as a meeting: held, with someone who can say yes. 03 #### Builds the tools The research, tracking and reporting that run behind every account are his design. The people stay in charge of them. The record 13 years in SaaS, from retention rep to the C-suite davidadamsofficial.com £30m a year: the partner-led SaaS business he ran as COO and CRO lucennioconsultancy.com 1,500+ partner and affiliate relationships built davidadamsofficial.com £2m+ a year of ad network revenue, grown from nothing davidadamsofficial.com DAVID ADAMS · CO-FOUNDER AND CEO > “A meeting counts when it was held, with someone who can say yes.” David Adams · Guides ### Written by David. Websites, getting named by AI and the business case. Every guide reviewed September 2026. Start with the three founders ask about most, or pick a topic. **Guides** - 34 **Topics** - 3 **Of reading** - 6 hrs Getting named by AI assistants Do we still need SEO now that buyers ask ChatGPT instead of Google? 11 min read Websites that bring work in What will our new website really cost over three years? 11 min read Costs, returns and the business case How much should we budget to get recommended by ChatGPT? 12 min read #### Getting named by AI assistants - Do we need schema markup for AI assistants to understand our site? 11 min - Do we still need SEO now that buyers ask ChatGPT instead of Google? 11 min - How do we get ChatGPT to recommend our firm when buyers ask? 11 min - How often do we need to publish for AI assistants to notice us? 10 min - What is domain authority, and do AI assistants actually use it? 11 min - Does being recommended by ChatGPT work differently for B2B firms? 12 min - How do AI crawlers read my website, and are we letting them in? 11 min - How do I organise our content so AI assistants see us as the expert? 13 min - What are Google AI Overviews and what do they mean for my firm? 14 min - What is E-E-A-T and why do AI assistants care about it so much? 14 min - Why does ChatGPT describe my firm wrongly, and how do I fix it? 11 min - Why does ChatGPT ignore our website when we have hundreds of pages? 14 min - Will Siri, Alexa or Gemini ever recommend my firm to a buyer? 11 min #### Websites that bring work in - Does our website's speed and accessibility really affect enquiries? 11 min - What should our website actually do for the business in 2026? 10 min - What will our new website really cost over three years? 11 min - Do our B2B buyers really look at our website on their phones? 11 min - Does our website platform affect whether ChatGPT can read our site? 12 min - How do I know whether our website is actually working? 12 min - How should we organise our website so buyers and AI understand it? 11 min - Is a template website good enough for a serious B2B firm? 10 min - Is our website technically ready for ChatGPT and AI search? 12 min - Should we redesign our website or keep improving the one we have? 11 min - What does it cost us to leave our website unchanged for years? 10 min - What does leaving our website until later actually cost us? 12 min - What makes a buyer trust our website enough to get in touch? 11 min #### Costs, returns and the business case - How do I get my board to approve budget for AI search? 12 min - Is being recommended by ChatGPT a marketing cost or an investment? 12 min - Can we just do content, or do we need PR and a better website too? 11 min - How do we choose a firm to help us get recommended by ChatGPT? 11 min - How much should we budget to get recommended by ChatGPT? 12 min - Should we pay monthly or upfront for our website and marketing? 10 min - What is it costing us not to be recommended by ChatGPT? 10 min - Will a cheap AI search package get us recommended by ChatGPT? 10 min More of the team ### The other two founders. Co-founder · Data and insight #### Ashley Kenerson Thirteen years turning marketing data into decisions at one of the world's biggest agencies. Keeps our numbers honest. Meet Ashley Co-founder, non-executive · The founder's record #### Ross Williams Built and ran a subscription business for twenty-one years, bootstrapped. Backs David and Ashley, and established the Foundation. Meet Ross ### Thirty minutes with David. Three fixes in writing. Bring one service you want to grow. Leave with the three things we would fix first, in writing. Book a Growth Review --- Title: Ross Williams, co-founder of Fortitude Media URL: https://fortitudemedia.ai/team/ross-williams Summary: Ross Williams built Rawnet, then bootstrapped Venntro Media Group and ran it for twenty-one years before selling in 2024. He co-founded Fortitude Media. ROSS WILLIAMS · CO-FOUNDER, NON-EXECUTIVE Co-founder, non-executive · The founder's record ## Ross Williams. 30+ Years of Award-Winning Entrepreneurship Nearly thirty years winning customers on the internet, first for other people's businesses and then for his own. Fortitude is the firm he wished he could have hired. David and Ashley run it; Ross backs them. Book a Growth Review LinkedIn rosswilliams.com The Reformed Entrepreneur **Building online since** - 1998 **Companies founded** - Rawnet, Venntro Media Group **Wrote** - Subscribe and Conquer **Lives in** - Portugal The story ### Two companies from nothing. One very long apprenticeship. He started Rawnet in 1998, while still at university, and grew it into one of the largest digital agencies outside London, working for Honda, Castrol, Honeywell and O2. It was profitable and it won awards. It also taught him the difference between what an agency does well and what it gets paid for, which is not always the same thing. In 2003 he co-founded Venntro Media Group with a credit card and no outside investment. Its White Label Dating platform let hundreds of entrepreneurs launch dating brands of their own, and it grew into the world's largest privately owned online dating business: around $50 million a year at its peak, 200 people, and offices in Windsor, London, San Francisco and Sydney. Almost every member arrived because the business was found online. He sold it in 2024. Along the way the company made the Sunday Times Fast Track 100 three years running, and Ross was named EY Entrepreneur of the Year and IoD Director of the Year. He wrote Subscribe and Conquer, a playbook for subscription businesses, and still writes every week as The Reformed Entrepreneur. Fortitude is the growth firm he wished he could have hired when he was in the founder's chair. He co-founded it and sits on the board, not in the day-to-day: David and Ashley run the business. Ten per cent of everything Fortitude invoices goes to the Fortitude Foundation, which Ross set up to help founders in trouble. He lives in Portugal with his wife, three children and two dogs. Ross, 2026 At Fortitude ### What Ross brings to Fortitude. 01 #### The founder's view Twenty-one years in the chair you are sitting in. The plan, the report and the guarantee were all shaped by what he wished an agency had given him. 02 #### The record The firm runs on the same system that won 20 million members for the platforms he built. He backs David and Ashley to run it. 03 #### The Foundation Ten per cent of everything Fortitude invoices goes to the Fortitude Foundation, which Ross founded for founders in trouble. The record 28 years building and selling internet businesses Rawnet, 1998 · Venntro, 2003 20m+ members on the platforms he built Venntro Media Group $350m revenue generated, almost all of it from being found online Venntro Media Group 3 years running in the Sunday Times Fast Track 100 2010 to 2012 ROSS WILLIAMS · CO-FOUNDER, NON-EXECUTIVE > “The tools changed. The job didn't. Be where the buyer looks, and give them a reason.” Ross Williams · Guides ### Written by Ross. How buying is changing, PR and reputation, and sector guides. Every guide reviewed September 2026. Start with the three founders ask about most, or pick a topic. **Guides** - 26 **Topics** - 4 **Of reading** - 5 hrs How buying is changing Can we wait a year before doing anything about AI search? 9 min read Press, reviews and being trusted Does traditional PR still help when buyers ask ChatGPT who to hire? 10 min read How buying is changing Should I still pay a marketing agency now that AI can do the work? 11 min read #### Sector guides - Which law firms does ChatGPT recommend, and how do we become one? 11 min - Why does ChatGPT name the big firms and not our consultancy? 11 min - Why does ChatGPT recommend other recruitment agencies to employers? 11 min - How does a construction firm get named when clients ask ChatGPT? 12 min - How does an architecture practice get recommended by ChatGPT? 11 min - How does our agency get named when clients ask ChatGPT for one? 11 min - Which accountant does ChatGPT recommend, and why is it not us? 12 min - Why does ChatGPT name other debt collection agencies and not us? 12 min - Why does ChatGPT recommend other agents in our area and not us? 12 min - Why don't founders find our fund when they ask ChatGPT for investors? 11 min #### How buying is changing - Can we wait a year before doing anything about AI search? 9 min - How will AI search change my industry in the next twelve months? 10 min - Should I still pay a marketing agency now that AI can do the work? 11 min - What does ChatGPT tell candidates about working at my firm? 11 min - What should my website do now that visitors arrive pre-sold by AI? 11 min - Will UK and EU regulation change who AI assistants recommend? 11 min #### Press, reviews and being trusted - Does traditional PR still help when buyers ask ChatGPT who to hire? 10 min - Are guest articles in trade magazines still worth our time? 11 min - Do awards and accreditations make ChatGPT more likely to recommend us? 11 min - How do we get trade journalists to call us for comment? 11 min - How long does PR take to work, and what should a year look like? 10 min - Which kinds of PR actually win work for a B2B firm like ours? 11 min #### Content that gets quoted - Can we outsource our content without losing our voice? 13 min - What makes ChatGPT treat our content as expert rather than marketing? 12 min - Should our content answer the questions our customers actually ask? 12 min - Should we turn our webinars and podcasts into written content? 12 min More of the team ### The other two founders. Co-founder and CEO · AI and automation #### David Adams Thirteen years inside a £30 million platform, from the phones to the boardroom. Now he runs Fortitude and builds the machine. Meet David Co-founder · Data and insight #### Ashley Kenerson Thirteen years turning marketing data into decisions at one of the world's biggest agencies. Keeps our numbers honest. Meet Ashley ### Talk to the people who run it. Book thirty minutes with David. Bring one service you want to grow. Leave with the three things we would fix first, in writing. Book a Growth Review --- Title: The Fortitude Foundation: 10% of everything we invoice URL: https://fortitudemedia.ai/foundation Summary: Ten per cent of everything we invoice goes to The Fortitude Foundation, established by our co-founder Ross Williams to help entrepreneurs in crisis. ## Ten per cent of everything we invoice goes to founders in crisis. We have had the good years and the ones nobody puts on a website. The Fortitude Foundation was established by our co-founder Ross Williams to help entrepreneurs through the worst of it. Every invoice we send funds it, every month, without exception. Visit the Foundation Book a Growth Review Ten per cent of revenue invoiced, not of profit · Paid monthly · Your price is the same either way Ten per cent of every invoice In plain terms ### What does the 10% actually mean? For every £10,000 we invoice, £1,000 goes to The Fortitude Foundation to support an entrepreneur in crisis. It comes off revenue, not profit, so there is no year in which it quietly does not happen. It costs you nothing extra: it is our money, given away, and it is one of the reasons we started this business. Why we built it this way ### We know what the bad year looks like. Between the three of us there are thirty years of building companies. Some of it went very well. Some of it did not. Entrepreneurship is sold as a highlight reel. The truth is quieter and harder: the month the cash does not arrive, the contract that goes away, the conversation at home you have been putting off. Almost every founder we know has had a year like that, and almost none of them talked about it at the time. Ross established The Fortitude Foundation for exactly that moment, to give founders in crisis somewhere to turn that is not a bank, a creditor or a brave face. It needs funding to keep doing that, year after year, not just when someone feels generous. So we built the funding into the business. Fortitude Media exists to do excellent work for founder-led firms, and to send ten per cent of what it invoices to the Foundation. Those two things are the same plan, and they were the same plan before the first invoice went out. Ross Williams, who set up the Foundation Meet Ross How the 10% works ### Invoiced, set aside, handed on. No small print, no cap, no year off. Three steps, every month. Step 01 #### You are invoiced Normal work at the normal price. Nothing is added to your fee for this, and nothing is held back from your programme because of it. Step 02 #### Ten per cent is set aside Of the invoice value, not of what happens to be left at the end of the year. It is a standing line in our accounts and it is reconciled monthly. Step 03 #### It reaches a founder The Foundation puts it to work supporting entrepreneurs in crisis, in its own name and on its own terms. We fund it. They do the work. What the funding supports ### Help at the moment it is needed. The Foundation decides how the money is used. In broad terms, this is the kind of help it exists to give. Support A founder in crisis, heard Somewhere to take the thing you cannot say to your team, your bank or your family, from people who have been there. Practical help Getting through the next month Help with the immediate, practical mess a business crisis creates, rather than advice about next year. Mental health The part nobody posts about Founders carry the risk personally. Support for the health cost of that is not a luxury. Back on their feet A second attempt Failure is data, not a verdict. Support for founders rebuilding after a business has gone under. Nobody gets through it on their own If you work with us #### Your spend does two jobs. You are buying a growth programme, and you should judge it on pipeline, enquiries and meetings held, like any other line in the budget. The second job happens automatically: ten per cent of what you pay us reaches a founder having the worst year of their life. - No premium. Prices are the same as they would be without the Foundation. This is not a cause you are being charged for. - No conditions. It applies to every invoice, every month, on every service we sell. - Not a campaign. It was in the plan before we had a client, and it does not end when the quarter is tight. Read about the Foundation's work in its own words at fortitudefoundation.com. Questions people ask ### The short answers. Is it 10% of revenue or of profit? Revenue. Ten per cent of everything we invoice, before costs, before salaries, before anything is left over. Profit is a decision. Revenue is a fact, and it is the only honest basis for a promise like this. When is it paid? Monthly, against the invoices we have raised. It is a line in our accounts, not a gesture we remember at Christmas. Does it cost me more? No. Our prices are the same either way. The ten per cent comes out of our side of the invoice, not yours. Who runs the Foundation? The Fortitude Foundation was established by Ross Williams, co-founder of Fortitude Media. It operates in its own right, and you can read about its work on its own website. Why tell clients about it at all? Because it is part of why this business exists, and because founders choosing between two firms deserve to know where their money goes. We would rather say it plainly than keep it quiet. ### Good work, put to good use. Thirty minutes with a founder. You leave with the three things we would fix first, and every pound you ever spend with us sends ten pence to a founder who needs it. Book a Growth Review --- Title: Who we work with: founder-led B2B firms | Fortitude Media URL: https://fortitudemedia.ai/for Summary: Founder-led B2B firms where the founder still decides: professional services, technology and SaaS, fintech, recruitment, agencies and consultancies. ## Founder-led B2B firms. Five sectors we know well. Firms with £1m to £20m+ revenue, where the founder still decides. The buyer is a CEO or an owner, the purchase is considered, and the shortlist is made before anyone calls you. Pick your sector for what that looks like in your market. Who we work with Pick your sector ### Where the shortlist is made before the call. Professional services #### Advice firms and consultancies “We're the best-kept secret in our industry.” Buyers ask an assistant for a shortlist before they ask a friend. Be on it. Read the sector page Technology and SaaS #### Founder-led software firms “I asked ChatGPT who to use. We weren't on the list.” Website, reputation and outbound that get you onto shortlists drawn up by people who have never heard of you. Read the sector page Fintech #### Regulated, considered, slow to trust “Competitors rank above us. They're worse.” Proof buyers can check, written for the questions procurement actually asks. Read the sector page Recruitment #### Agencies that live on relationships “I delete twenty LinkedIn pitches a week. I'm not going to send them.” A pipeline that does not depend on who picked up the phone this week. Read the sector page Agencies and consultancies #### Firms that sell other people's growth “Our website undersells us.” The cobbler's children. A programme run by founders who have been there. Read the sector page Not on the list? #### If a founder decides and the purchase is considered, we can help. Thirty minutes with a founder tells you whether we fit. Book a Growth Review --- Title: Growth team for agencies and consultancies | Fortitude Media URL: https://fortitudemedia.ai/for/agencies-and-consultancies Summary: How founder-led agencies and consultancies win clients beyond the founder's network: a clear specialism, published thinking and steady outreach. ## You do this for clients. Nobody does it for you. Agencies and consultancies are the cobbler's children. Client work always wins, so your own site, thinking and new business get done in the gaps. We run them as a client account, with a deadline. Ross, one of our founders, founded Rawnet, so the excuses are familiar. Book a Growth Review Doing for ourselves what we sell The short answer ### How do agencies and consultancies win new clients without relying on the founder? Most agencies and consultancies grow through the founder's network, referrals and the occasional pitch. That stalls when the network is exhausted. Firms that keep growing pick a clear specialism, publish their thinking and results under named people, and run steady, researched outreach instead of bursts between projects. Buyers shortlist firms that appear to have solved their exact problem before. Your buyers ### How your buyers choose now. A marketing director or CEO choosing an agency or consultancy usually has three names before anyone is invited to pitch. Those names come from people they trust, from work they have noticed, and increasingly from asking an AI assistant who is good at one particular thing. They ask who did that piece of work Buyers notice a rebrand, a campaign, a turnaround, and ask who was behind it. If your name is not attached to your best work anywhere public, the credit goes nowhere and so does the enquiry. They judge your thinking before your credentials Anyone can list services. Buyers read an article or watch a talk to see whether the people are sharp and hold a point of view. For a consultancy, published thinking is the product sample. Most keep theirs in decks only clients see. They look at who they will actually get Every buyer has been sold to by the founder and handed to a junior. So they look up the team on LinkedIn, check how long people stay, and look for named leads on case studies. Show who does the work. Sound familiar? ### What founders in this sector tell us. - “Our website undersells us.” - “I should post on LinkedIn. I never do.” - “We've been at £5m for three years.” All the things founders tell us What we would do ### Three jobs, run as one. Website That Sells Your own site treated as a live account, with pages for the problems you solve and at least one dated, logged improvement every week. Reputation That Wins The thinking that lives in your pitch decks, pulled out in a monthly interview and published as three expert pieces under your people's names. Pipeline That Fills New business that does not stop when a big project lands: researched approaches every month, paid per qualified meeting held. From Insights ### Guides for agencies and consultancies. Sector guides #### How does our agency get named when clients ask ChatGPT for one? 31 March 2026 Sector guides #### Why does ChatGPT name the big firms and not our consultancy? 22 April 2026 How buying is changing #### Should I still pay a marketing agency now that AI can do the work? 31 March 2026 Content that gets quoted #### Can ChatGPT tell real expertise from keyword-stuffed content? 31 March 2026 Content that gets quoted #### Can we outsource our content without losing our voice? 22 April 2026 Questions founders ask ### The short answers. We are an agency. Should we not be able to do this ourselves? You are able to. You will not, because client work pays this month and your own marketing pays next year, so it loses every scheduling argument. Skill is not the gap. Priority is. An outside team has your firm as its client, with a deadline. Will content written by someone else sound like us? It starts with a 30-minute interview with your people each month, so the opinions, examples and turns of phrase are theirs. We shape that into three expert pieces. If you have a strong house voice, month one is when we learn it, and you will be a harder editor than most. That is fine. Do you compete with us? If you sell websites, PR or outreach to founder-led B2B firms, then in part, yes, and we should both say so in the Growth Review. Plenty of agencies and consultancies sell something else: brand, creative, strategy, technology, change. For them we are simply a supplier. ### Thirty minutes, and three fixes. A founder checks your site, two competitors and what Google and ChatGPT say about all three before the call. You leave with three fixes in writing. Book a Growth Review --- Title: Growth team for UK fintech firms | Fortitude Media URL: https://fortitudemedia.ai/for/fintech Summary: How B2B fintech firms win cautious buyers: what finance directors check before a call, and how website, reputation and outreach answer it. ## In fintech, trust gets checked. Usually before you know they looked. Selling a financial product to a business means passing due diligence you never see. Buyers check the register, the press, the reviews and the people long before a call. We make that evidence easy to find, and keep the pipeline moving while it builds. Book a Growth Review Evidence a regulated buyer can check The short answer ### How do B2B fintech firms win customers when buyers are cautious? Business buyers of financial products are careful because a bad choice is hard to reverse. Before any call they check regulatory status, who backs the firm, what the press and review sites say, and whether anyone they know uses it. Fintech firms win by publishing that evidence plainly: permissions, pricing, security, named customers and named people. Claims without checkable proof get discounted. Your buyers ### How your buyers choose now. Your buyer is often a finance director, a broker principal or a head of operations, and compliance will have a view. They are trusting you with money or customer data, so they look for reasons to say no. Give them fewer. They check the register and the backers The FCA register, Companies House, the investors, the banking or licensing partner. It takes five minutes and happens early. Make it easy to confirm, with reference numbers and named partners, and you look like a firm with nothing to hide. They ask someone who already uses you Finance people talk. A buyer asks their accountant or a peer what they use and whether it has ever gone wrong. Named customers, accurate reviews and press in titles they read stand in for that conversation. They want the pricing and the exit explained Fees, contract length, onboarding time, what happens to their data and money if they leave. A buyer who cannot find these assumes the answer is unpleasant. Stating them plainly is unusual enough in financial services to be an advantage. Sound familiar? ### What founders in this sector tell us. - “Competitors rank above us. They're worse.” - “ChatGPT has never heard of us.” - “We book meetings with people who can't say yes.” All the things founders tell us Client work ### Proof it works, in this sector. Proof it works · ID-Pal 0 4 UK buyer questions where AI now names ID-Pal Read the story Proof it works · TCC Group 1 week to a plan TCC's own team built the new site from Read the story What we would do ### Three jobs, run as one. Website That Sells Permissions, pricing, security and onboarding time stated on the page and marked up, so a cautious buyer can confirm them without ringing you. Reputation That Wins One to four earned press placements a month, a data story each quarter from your own numbers, and reviews and listings kept accurate. Pipeline That Fills Two-pass research, firm then person, so approaches reach the finance director or principal who can decide, with a reason to talk this month. From Insights ### Guides for fintech. Sector guides #### How does a financial planning firm get named when clients ask ChatGPT? 22 April 2026 Sector guides #### Why does ChatGPT recommend other insurance brokers and not us? 31 March 2026 Sector guides #### Why don't founders find our fund when they ask ChatGPT for investors? 31 March 2026 Sector guides #### Why does ChatGPT name other debt collection agencies and not us? 31 March 2026 How buying is changing #### Will UK and EU regulation change who AI assistants recommend? 31 March 2026 Questions founders ask ### The short answers. Compliance signs off everything here. Will this slow to a crawl? It will be slower than in an unregulated firm, and the plan should assume that. Outreach volume is low by design, 80 to 120 approaches a month, and a person writes and approves every message, so compliance has something specific to review. Agree the approval route in the Growth Plan. What is the quarterly data story, and do we have the data? Almost certainly. Fintech firms sit on numbers nobody else has: payment times by sector, approval rates, claim patterns, deal sizes. Once a quarter we turn a cut of them into a story journalists can use. It earns press without buying links, and gives AI assistants a fact attributed to you. We have raised money. Should we not just hire a team? Perhaps, in time. At 2026 UK medians, a marketing manager, a content and search manager, a marketing executive and a sales development rep cost £203,991 a year with employer NI, pension and software. A £6,500-a-month programme is £78,000. Find out what works first, then hire people to run it. ### Thirty minutes, and three fixes. A founder checks your site, two competitors and what Google and ChatGPT say about all three before the call. You leave with three fixes in writing. Book a Growth Review --- Title: Growth team for professional services | Fortitude Media URL: https://fortitudemedia.ai/for/professional-services Summary: How founder-led law, accountancy, consultancy and architecture firms win clients now that every referral gets checked online first. ## Your referrals still work. They just get checked first. Law, accountancy, consultancy and architecture firms grow on reputation. That reputation now gets looked up before anyone rings you. We make sure what the buyer finds matches the firm your clients describe, and add new conversations alongside the referrals. Book a Growth Review The twenty minutes that decide whether you get the call The short answer ### How do founder-led professional services firms win new clients now? Most still win work through referrals, but the referral no longer ends the process. The buyer looks up the firm, reads the partner's profile, compares two alternatives and often asks an AI assistant who is good at this work. Firms that publish specific expertise, named people, clear fees and recent proof survive that check. Brochure sites quietly drop off the shortlist. Your buyers ### How your buyers choose now. A finance director or business owner rarely hires an adviser cold. They get a name from someone they trust, then spend twenty minutes checking it. Those twenty minutes decide whether you get the call. Three things carry most of the weight. They ask a peer, then check the name The name comes from their accountant, a board member or another owner. Then they search the firm and the partner. A site offering 'a full range of services' and a LinkedIn profile untouched for years will cool the referral. They look for their exact problem Nobody wants 'a corporate lawyer'. They want someone who has sold a business like theirs to a trade buyer. They look for pages and case notes naming their situation, sector and size. General expertise reads as none. They check credentials and the fee Regulator listings, accreditations, directories and reviews all get a glance. So does any hint of price. A firm that gives a fee range looks confident. Silence lets a careful buyer assume the worst and ring someone else. Sound familiar? ### What founders in this sector tell us. - “We're the best-kept secret in our industry.” - “Our best thinking is in our heads.” - “We've outgrown word of mouth.” All the things founders tell us Client work ### Proof it works, in this sector. Proof it works · TCC Group 1 week to a plan TCC's own team built the new site from Read the story Proof it works · arum 310 pages rebuilt across three regions, in weeks Read the story What we would do ### Three jobs, run as one. Website That Sells Pages built around the questions clients ask before instructing you, with fees, timelines, credentials and named people stated plainly. Reputation That Wins A monthly 30-minute interview turns what your partners know into three expert pieces in their voice, backed by earned press. Pipeline That Fills 80 to 120 researched approaches a month to the owners and finance directors you want, each written and approved by a person. From Insights ### Guides for professional services. Sector guides #### Which law firms does ChatGPT recommend, and how do we become one? 22 April 2026 Sector guides #### Which accountant does ChatGPT recommend, and why is it not us? 31 March 2026 Sector guides #### Why does ChatGPT name the big firms and not our consultancy? 22 April 2026 Sector guides #### How does an architecture practice get recommended by ChatGPT? 31 March 2026 Press, reviews and being trusted #### Do awards and accreditations make ChatGPT more likely to recommend us? 31 March 2026 Questions founders ask ### The short answers. We are regulated. Can this work within our publicity rules? Yes. Expert pieces come from a monthly interview with your own people, so the claims are theirs. Press placements are earned, never bought links. Reviews and listings are kept accurate, not manufactured. If your professional body has rules on publicity, raise them in the Growth Review. Our partners will not write articles. Does that matter? No. They write nothing. One 30-minute interview a month with the people who know the work becomes three expert pieces in their voice. For the managing partner, the same interview can become six to eight LinkedIn posts. Most partners will talk for half an hour about work they are good at. What does it cost to find out whether this suits us? Nothing at first. The Growth Review is a free 30-minute call with a founder. Beforehand we check your site, two competitors, and what Google and ChatGPT say about all three. Afterwards you get three fixes in writing. You can stop after any step. ### Thirty minutes, and three fixes. A founder checks your site, two competitors and what Google and ChatGPT say about all three before the call. You leave with three fixes in writing. Book a Growth Review --- Title: Growth team for UK recruitment firms | Fortitude Media URL: https://fortitudemedia.ai/for/recruitment Summary: How founder-led recruitment firms win new clients when hiring managers ignore cold calls: specialist reputation, salary data and researched outreach. ## Your clients are tired of cold calls. Be the firm they already rate. Recruitment runs on the founder's relationships and a floor of consultants on the phone. It works until the market tightens or the founder steps back. We build the specialist reputation and researched outreach that bring hiring managers to you half convinced. Book a Growth Review Researched approaches, written by a person The short answer ### How do founder-led recruitment firms win new clients now? Hiring managers choose recruiters they already know, or ones a peer vouches for, and they ignore most cold approaches because they receive so many. Recruitment firms win new clients by being visibly specialist: a narrow market, named consultants, salary data, recent placements and useful market comment. That gives a hiring manager a reason to reply, and gives Google and AI assistants something to name. Your buyers ### How your buyers choose now. A hiring manager with a hard role to fill does not run a tender. They go back to whoever filled the last one, ask a peer, or look at who seems to own that corner of the market. Often there is a supplier list to get past too. They ask who filled a role like this The question to a peer is specific: who found your finance director, who is good for embedded engineers in the Midlands. Generalists do not come up in that conversation. The firm known for one market does. They look up the consultant, not the brand After a call or an email, the hiring manager checks the consultant's LinkedIn. They want signs this person knows the market: sensible posts, relevant connections, time served. A profile of nothing but job adverts says CV forwarder. They check you understand the market Salary guides, time-to-hire figures from your own desk, plain comment on what candidates want. This lets a recruiter argue for a fee above the lowest on the supplier list. It is also what AI assistants quote when asked who recruits in a sector. Sound familiar? ### What founders in this sector tell us. - “I delete twenty LinkedIn pitches a week. I'm not going to send them.” - “Pipeline is feast or famine.” - “If I took a month off, sales would stop.” All the things founders tell us What we would do ### Three jobs, run as one. Website That Sells A page for each market you recruit in, covering fees, process and timelines, instead of a job board with a logo on top. Reputation That Wins A data story each quarter from your own placement and salary numbers, and the founder's LinkedIn run properly from one interview a month. Pipeline That Fills 80 to 120 approaches a month, each naming something true about the firm that is not on its homepage. From Insights ### Guides for recruitment. Sector guides #### Why does ChatGPT recommend other recruitment agencies to employers? 22 April 2026 How buying is changing #### What does ChatGPT tell candidates about working at my firm? 31 March 2026 Content that gets quoted #### Does publishing our own research help ChatGPT recommend us? 31 March 2026 Content that gets quoted #### Should our content answer the questions our customers actually ask? 31 March 2026 Questions founders ask ### The short answers. My consultants already do business development. Why add outreach? Because consultant BD is volume calling squeezed in between filling roles, and it stops when desks get busy. That is where feast and famine comes from. Researched outreach runs every month regardless, with a person writing each message. Your consultants take the meetings. The list and conversations are yours to keep. Do you work on candidate attraction as well? The focus is clients, because that is where a founder-led recruitment firm's growth usually stalls. The same work helps with candidates, though. Specialist pages, salary data and credible consultant profiles are what a good candidate checks before replying to you. Job advertising and job boards are not something we run. What counts as a qualified meeting for a recruitment firm? The usual definition. The meeting is held, with someone who owns or materially influences the hiring decision, at a firm matching criteria agreed in advance, who knowingly agreed to talk about your offer. A chat with a junior administrator who cannot add you to the supplier list does not count. ### Thirty minutes, and three fixes. A founder checks your site, two competitors and what Google and ChatGPT say about all three before the call. You leave with three fixes in writing. Book a Growth Review --- Title: Growth team for UK technology and SaaS | Fortitude Media URL: https://fortitudemedia.ai/for/technology-and-saas Summary: How founder-led B2B software firms get onto shortlists made before the demo: pricing, proof, reviews and being named by Google and AI assistants. ## Your product does the job. The shortlist is made before the demo. Founder-led software firms usually grow on the founder's network, a few partners and a product people like. Then the network runs out. We build the website, reputation and outbound that get you onto shortlists drawn up by people who have never heard of you. Book a Growth Review Who gets named when a buyer asks an assistant The short answer ### How do founder-led B2B technology firms get onto buyers' shortlists? B2B software buyers do most of their research before speaking to a vendor. They search the problem, read review sites, ask peers in private communities and increasingly ask an AI assistant for two or three names. A technology firm gets shortlisted when its pricing, integrations, security position and customer proof are published plainly enough for people and machines to find and repeat. Your buyers ### How your buyers choose now. The person who finds you is rarely the person who signs. An operations lead does the digging, builds a shortlist of three, and takes it to a finance director or CEO. Your site has to arm the first and reassure the second. They search the problem, not your category Buyers type what hurts: late invoices, scheduling engineers, reconciling stock. They rarely type the category name your investors use. Pages headed with product names and feature labels do not appear for the question that was asked. They ask peers and read reviews Slack groups, LinkedIn, former colleagues, software review sites. A buyer wants to hear from someone running a similar firm on a similar stack. Stale review profiles and no named customers make a small vendor look riskier than it is. They want price, integrations and security Before a demo, a serious buyer wants a rough price, proof you connect to what they already run, and evidence you look after data. Hide all three behind 'talk to sales' and they shortlist the vendor that answered. Sound familiar? ### What founders in this sector tell us. - “I asked ChatGPT who to use. We weren't on the list.” - “I am still our best salesperson.” - “Pipeline is feast or famine.” All the things founders tell us What we would do ### Three jobs, run as one. Website That Sells Pages that answer the pre-demo questions, price, integrations, security and time to go live, marked up so Google and AI assistants can read them. Reputation That Wins Three expert pieces a month from your own people, earned press and accurate review profiles, so an AI assistant has something to cite. Pipeline That Fills Researched outreach to firms that fit, paid per qualified meeting held with someone who owns or materially influences the decision. From Insights ### Guides for technology and saas. Sector guides #### How does a B2B tech company get recommended when buyers ask ChatGPT? 22 April 2026 Getting named by AI assistants #### Does being recommended by ChatGPT work differently for B2B firms? 31 March 2026 Getting named by AI assistants #### Do we need schema markup for AI assistants to understand our site? 22 April 2026 Costs, returns and the business case #### How do we find out whether ChatGPT is recommending our firm? 31 March 2026 Questions founders ask ### The short answers. We already have a marketing hire. Does this replace them? Not necessarily. One marketer in a software firm is usually stretched across events, launches, the website and sales requests. We can run the parts they cannot reach, such as outbound or the weekly tracking of AI answers. The Growth Review will show whether your gap is capacity or direction. How do you know whether AI assistants name us? Our own platform asks ChatGPT, Claude, Gemini and Google the 40 to 150 questions your buyers ask, every week, and records who is named. You see which questions name you, which name a competitor, and how that moves. It is on the one-page report each month. Our sales cycle is long. When would we see anything? Outbound is quickest. Month one is warm-up. The guarantee covers months two and three: fewer than six qualified meetings held across both, and month four's management fee is waived. A new website is live in 4 to 8 weeks. AI answers move over months, not weeks. ### Thirty minutes, and three fixes. A founder checks your site, two competitors and what Google and ChatGPT say about all three before the call. You leave with three fixes in writing. Book a Growth Review --- Title: Compare your options | Fortitude Media URL: https://fortitudemedia.ai/vs Summary: Four honest comparisons for a founder deciding how to grow: hire a team, use a lead generation agency, stay with the SEO agency, or carry on as you are. ## Compare your options. Including the ones that are not us. Four honest comparisons for a founder deciding how to grow: hire a team, use a lead generation agency, stay with the SEO agency, or carry on as you are. Each one says when the alternative is the better choice. Book a Growth Review 4 comparisons ### The decision in front of you. #### Hiring a marketing team or renting a senior one Should a £5m B2B firm hire a marketing team or use an outside growth team? #### A lead generation agency or a pipeline people take seriously Is a lead generation agency worth it for a founder-led B2B firm? #### An SEO agency or being named in the answer Is an SEO agency still worth paying for in 2026? #### Carrying on as you are is a choice. Cost it like one. Is it a problem if a B2B firm relies on referrals and the founder for sales? ### Still weighing it up? Bring it to the call. Thirty minutes with a founder. We check your site, two competitors and what Google and ChatGPT say about all three beforehand. You leave with three fixes in writing. Book a Growth Review --- Title: Referrals only: is doing nothing fine? | Fortitude Media URL: https://fortitudemedia.ai/vs/doing-nothing Summary: Referrals and the founder's own selling built the firm. When carrying on as you are is the right call, when it is not, and how to cost the choice. ## Carrying on as you are is a choice. Cost it like one. Referrals and your own selling built the firm. They are cheap, they convert, and they still work. So the fair question is whether to change anything at all. Here is the case for leaving it alone, and the case against. Book a Growth Review The short answer ### Is it a problem if a B2B firm relies on referrals and the founder for sales? Not while it produces enough of the right clients and the founder wants to keep selling. It becomes a problem when growth flattens, when the founder wants to step back or sell the firm, or when referred buyers start checking before they call. A referral now gets looked up, compared and sometimes put to an AI assistant. A firm with little to find can lose work it never knew about. Side by side ### The same questions, both ways. What you are weighing Carry on as you are Fortitude Cost - Nothing on an invoice. The cost is your time, plus work you lose without hearing about it. - From £2,500 a month. Most clients pay £5,000 to £8,000. Quality of lead - Excellent. A referred buyer arrives trusting you. Nothing we do beats a warm introduction. - Colder to begin with. A qualified meeting is still a stranger who has agreed to talk. Control over volume - None. Referrals arrive when they arrive, which is where feast and famine comes from. - 80 to 120 researched approaches a month, every month. Dependence on you - Total. Take a month off and selling stops. - The plan runs without you, apart from a monthly interview and the meetings themselves. When a buyer looks you up - Whatever is there now. If it undersells you, some referrals cool and you never learn why. - Pages that answer their questions, recent expert pieces, earned press, accurate reviews. When a buyer asks an AI assistant - It returns two or three names. If yours is not among them, you were never considered. - Tracked every week and worked on. Movement takes months, not weeks. Being honest about it ### When each option is the right one. #### When the alternative is the better choice - The firm is the size you want. If it pays you well and you have no wish to grow or sell, leave it alone. Growth for its own sake is an expensive hobby. - Referrals are still rising year on year and you enjoy the selling. Check what Google and ChatGPT say about you, fix anything embarrassing, and carry on. - Delivery is the bottleneck. If you cannot serve the clients you already win, more pipeline makes things worse. Sort out capacity first. #### When we are - Revenue has sat at the same figure for two or three years. The network that got you here has given what it has. - You want to step back, or sell within a few years. Either way, someone will ask where customers come from when you are not in the room. - You suspect you are losing work you never hear about. Test it today: ask an AI assistant the question your buyer would ask, and see who it names. THE NUMBERS Doing nothing is free on paper, so price the alternative against one number: your average client value. A firm whose average client is worth £15,000 at a 50% margin needs about ten new clients a year to cover a £6,500-a-month programme. Put your own figure in. Then consider how many clients a year you might be losing at the looking-up stage. Nobody can give you that number, us included. Prices exclude VAT. Questions founders ask ### The short answers. Referrals are working. Why would I spend anything? You may not need to. Do the free part first. Before the 30-minute Growth Review we check your site, two competitors and what Google and ChatGPT say about all three. You leave with three fixes in writing. If the answer is that you are fine as you are, that is what you will hear. Can I do the basics myself? Yes. Rewrite your main service page around the question buyers ask, with the answer first. State prices or ranges where you can. Bring your LinkedIn profile up to date. Ask five happy clients for a review. Check that your listings agree with each other. That is a weekend's work. What is the smallest sensible step? After the free review, the two-week Growth Plan at £1,500, credited against the programme if you go on. Or one service: founder LinkedIn only is from £1,500 a month. You can stop after any step. Ten per cent of everything we invoice goes to The Fortitude Foundation, for entrepreneurs in crisis. Sound familiar? ### Founders weighing this also said these. We've outgrown word of mouth. Word of mouth stops being enough when the revenue you need grows faster than your network. Referrals depend on how many people know your work, and that number rises slowly. Keep them, because they convert best, and add one channel you control: researched approaches to firms that fit, backed by a website that holds up when they look you up. Pipeline That Fills If I took a month off, sales would stop. Separate the parts of selling that need the founder from the parts the founder merely happens to do. Finding prospects, first approaches, follow-up and proving the firm's credibility can all run without you once written down and owned by someone. Closing the largest deals may stay with you for years. The aim is conversations that keep arriving while you are away. Growth Team programme From Insights ### Three things worth reading first. Costs, returns and the business case #### What is it costing us not to be recommended by ChatGPT? 31 March 2026 How buying is changing #### Can we wait a year before doing anything about AI search? 22 April 2026 Websites that bring work in #### What does leaving our website until later actually cost us? 31 March 2026 ### Weigh it with someone who has done both. Thirty minutes with a founder. We check your site, two competitors and what Google and ChatGPT say about all three beforehand. You leave with three fixes in writing. Book a Growth Review --- Title: Hire a marketing team or a growth team? | Fortitude Media URL: https://fortitudemedia.ai/vs/hiring-a-marketing-team Summary: What four marketing and sales hires cost a UK B2B firm in 2026, what an outside growth team costs, and when each one is the right call. ## Hiring a marketing team or renting a senior one Most founders at £5m assume the next step is a marketing manager. Sometimes it is. Here is the bill for building a team in-house, the bill for using ours, and the cases where you should hire anyway. Book a Growth Review The short answer ### Should a £5m B2B firm hire a marketing team or use an outside growth team? It depends on whether the firm already knows what works. An in-house team makes sense once there is a proven way of winning customers that needs running every day. Before that point, a single marketing hire is usually too junior to set direction and too stretched to deliver. An outside team costs less than the salaries, starts sooner, and can be stopped if it fails. Side by side ### The same questions, both ways. What you are weighing Hire four people Fortitude Annual cost - £203,991 a year at 2026 UK medians, plus about £39,210 in year one. - From £2,500 a month. A £6,500-a-month programme is £78,000 a year. Time to start - Months. Job specs, interviews, notice periods, then they learn the business. - A free Growth Review, an optional two-week Growth Plan, then work begins. Who sets direction - You. £56,950 buys a capable manager, not someone who has built and sold a company. - A founder who has built and sold companies owns your plan. Knowledge of your business - Deep, and deeper every year. They sit in your meetings and hear your customers. This is the real advantage of hiring. - Learned from a monthly interview and what sales hears. Never as close as someone in the building. Tools and tracking - About £12,000 a year in software and data, chosen and run by your team. - Included, with our own platform recording who AI assistants name for your buyers' questions. If it does not work - A performance process, a settlement, a rehire. Slow and unpleasant. - You can stop after any step. Being honest about it ### When each option is the right one. #### When the alternative is the better choice - You already know what works and need it run daily. If one channel reliably brings customers, hire an operator. They will be cheaper and better at it than any outside firm. - Marketing is the product. If the business lives on content, community and events, the team needs to sit beside the people building the product. - A proven marketing leader is ready to join. Someone who has grown a firm like yours is worth more than any supplier. Build around them. #### When we are - You have never had marketing that worked, so you cannot write the job spec. That is how firms end up with a marketing manager producing brochures. - You need senior judgement without a senior salary. A founder owns your plan and specialists deliver each part, for less than the four salaries. - You want to learn what works before committing to headcount. Run the programme for a year, then hire people to run the parts that produced customers. THE NUMBERS The 2026 UK medians, from the Ashdown Group guide: marketing manager £56,950, content and search manager £42,055, marketing executive £35,046, sales development rep base £32,000. Add employer NI at 15% (£21,908), pension at 3% (£4,032) and £12,000 for software and data. Total: £203,991 a year, plus about £39,210 in year one for recruiter fees at 20% and laptops. A £6,500-a-month programme is £78,000 a year. Our prices exclude VAT. Questions founders ask ### The short answers. Could we do both: one hire plus an outside team? Yes, and it is often the best arrangement. A marketing executive inside the firm handles events, sales requests, approvals and the small jobs an outsider cannot see. The outside team supplies direction, specialist skills and tracking. The hire improves faster working alongside senior people than guessing alone. Isn't a team we employ more committed than a supplier? More present, certainly. But an employee is paid whether or not customers arrive. Part of our fee is not: outreach is charged per qualified meeting held, and if fewer than six are held in months two and three combined, month four's management fee is waived. You can also stop after any step. What would we need to commit to? Thirty minutes to begin with. The Growth Review is free and ends with three fixes in writing, which you can hand to anyone. After that, a two-week Growth Plan costs £1,500, credited against the programme. Set that against a recruiter's fee of 20% on one salary. Sound familiar? ### Founders weighing this also said these. We've been at £5m for three years. A founder-led B2B firm usually stalls when the things that built it, referrals and the founder's own selling, reach their ceiling. The network has been worked, the founder's diary is full, and each year's new business only replaces the clients that left. Growth tends to restart when strangers can find the firm, trust it before the first call, and are approached on purpose. Growth Team programme If I took a month off, sales would stop. Separate the parts of selling that need the founder from the parts the founder merely happens to do. Finding prospects, first approaches, follow-up and proving the firm's credibility can all run without you once written down and owned by someone. Closing the largest deals may stay with you for years. The aim is conversations that keep arriving while you are away. Growth Team programme From Insights ### Three things worth reading first. Costs, returns and the business case #### How much should we budget to get recommended by ChatGPT? 31 March 2026 Costs, returns and the business case #### How do I get my board to approve budget for AI search? 22 April 2026 Costs, returns and the business case #### Is being recommended by ChatGPT a marketing cost or an investment? 22 April 2026 ### Weigh it with someone who has done both. Thirty minutes with a founder. We check your site, two competitors and what Google and ChatGPT say about all three beforehand. You leave with three fixes in writing. Book a Growth Review --- Title: Lead generation agency or growth team? | Fortitude Media URL: https://fortitudemedia.ai/vs/lead-generation-agency Summary: A fair comparison of lead generation agencies and researched outreach for founder-led B2B firms: volume, cost, what counts as a lead, who owns the list. ## A lead generation agency or a pipeline people take seriously Lead generation agencies sell appointments. Some are good at it. The question is what happens when a stranger gets your email, looks you up, and finds nothing to back the message. Here is a fair comparison, including when the appointment-setter is the right buy. Book a Growth Review The short answer ### Is a lead generation agency worth it for a founder-led B2B firm? It can be, if the offer is simple, the market is large and the average sale is small enough that volume matters more than fit. For firms selling considered, high-value services to a few thousand possible buyers, high-volume outreach tends to burn the list and the firm's name. Low-volume, researched outreach, backed by a website and reputation that survive being looked up, suits those firms better. Side by side ### The same questions, both ways. What you are weighing Lead generation agency Fortitude Volume - Often hundreds or thousands of templated emails a month, betting on a small reply rate. - 80 to 120 approaches a month, not 800. Two-pass research: the firm, then the person. Who writes the message - Usually a template with merged fields, sometimes written by software. - Research, timing and reminders are automated. A person writes and approves every message. What you pay for - Commonly a monthly retainer, sometimes a fee per appointment booked. - A fee per meeting held, on top of set-up and management. No meeting, no meeting fee. What counts as a lead - It varies, so ask. 'Booked' and 'held' are different things. - Held, with someone who owns or materially influences the decision, at a firm matching agreed criteria. Follow-up - Sequences of five, six or more chasers are common. - We stop after four unanswered emails. Who owns the list - Check the contract. Some keep the data and the mailboxes. - The list and the conversations are yours to keep. Being honest about it ### When each option is the right one. #### When the alternative is the better choice - Your offer is simple, low-priced and relevant to tens of thousands of firms. Volume is the right tool when the market is huge. - Salespeople are sitting idle and you need diaries filled this quarter. A large calling floor will produce more conversations, sooner. Month one with us is warm-up. - You only want meetings, your website and reputation are already strong, and you have found an agency that defines a lead as tightly as you do. Use them. #### When we are - Your market is a few thousand firms and you cannot afford to annoy them. Every message has to be one that could not be sent to anyone else. - You have paid for meetings with people who could not say yes. Here the fee depends on who was in the meeting. - Prospects look you up and go quiet. More outreach only sends more people to a weak page. Website and reputation can run under the same plan. THE NUMBERS Pipeline That Fills costs £1,500 to set up, £1,000 to £1,500 a month to manage, and £500 to £1,000 per qualified meeting held. If fewer than six are held in months two and three combined, month four's management fee is waived. To judge any quote, ours included, take your average client value and your close rate from a first meeting, and work out what one held meeting is worth. Prices exclude VAT. Proof it works ### What this looks like, for real. Proof it works · Helm Club The club had the members. It needed the calls. 1 in 5 new members this year came through an AI recommendation Read the story Questions founders ask ### The short answers. Why so few approaches a month? Because research takes time and your market is finite. A message that could not be sent to anyone else cannot be produced eight hundred times a month. Low volume also protects your name. People remember who spammed them, and in a sector of a few thousand firms that memory is expensive. What happens in month one? Warm-up. The criteria for a qualified meeting are agreed with you, the list is researched in two passes, firm then person, and the first approaches go out. Do not expect a full diary in four weeks. That is why the guarantee is measured on months two and three. Can we use you for outreach only? Yes. Pipeline That Fills can run alone. Before the Growth Review we check your site and what Google and ChatGPT say about you. If the evidence is thin we will say so, because prospects look you up after the first message. Sound familiar? ### Founders weighing this also said these. We book meetings with people who can't say yes. Meetings land with the wrong people when whoever books them is measured on meetings booked, not on who turns up. Junior staff accept calls more readily than owners, so any volume target drifts towards them. Fix the definition first: a meeting counts only if the person owns or materially influences the decision, works at a firm that fits, and knows what the call is about. Pipeline That Fills I delete twenty LinkedIn pitches a week. I'm not going to send them. Yes, if the volume is low enough for each message to be researched and written by a person. Outreach becomes spam when the same words go to hundreds of people. A message that names something true about the recipient's firm, gives a reason to talk now, and could not have been sent to anyone else reads as a business letter. Pipeline That Fills From Insights ### Three things worth reading first. How buying is changing #### What should my website do now that visitors arrive pre-sold by AI? 31 March 2026 Costs, returns and the business case #### How can we tell which of our new leads came from ChatGPT? 31 March 2026 Websites that bring work in #### What makes a buyer trust our website enough to get in touch? 31 March 2026 ### Weigh it with someone who has done both. Thirty minutes with a founder. We check your site, two competitors and what Google and ChatGPT say about all three beforehand. You leave with three fixes in writing. Book a Growth Review --- Title: SEO agency or growth team for a B2B firm? | Fortitude Media URL: https://fortitudemedia.ai/vs/seo-agency Summary: When an SEO agency is enough for a B2B firm, when it is not, what each costs, and three questions to ask your current agency before changing anything. ## An SEO agency or being named in the answer Search has not died. It has split. Some buyers still scroll Google, others ask an AI assistant and get two or three names back. A good SEO agency handles the first. Here is where that is enough, where it is not, and what each costs. Book a Growth Review The short answer ### Is an SEO agency still worth paying for in 2026? Yes, for firms whose customers search in volume and click through to websites. The technical basics of SEO still matter, and AI assistants draw on much of the same material. What has changed is that rankings alone no longer decide who gets recommended. AI assistants weigh what others say about a firm: press, reviews, expert content and consistent facts. Work that covers only rankings covers half the job. Side by side ### The same questions, both ways. What you are weighing SEO agency Fortitude What gets reported - Rankings, traffic and sometimes enquiries, usually in a long monthly report. - One page a month: meetings held, pipeline accepted by sales, buyer questions where an AI assistant names you. AI answers - Some now offer this. Ask how many questions they track, and how often. - Our own platform asks ChatGPT, Claude, Gemini and Google 40 to 150 buyer questions every week. Content - Often written to a keyword brief by a generalist writer. - A monthly interview with your experts becomes three expert pieces in your voice. Links and press - Varies widely. Some earn coverage. Some buy links, so ask directly. - One to four earned press placements a month. Never bought links. The website itself - Recommendations handed to your developer, who may or may not get to them. - We build and run the site, with at least one dated, logged improvement a week. Depth of technical SEO - A specialist goes deeper on very large or complicated sites. It is their whole trade. - Sound foundations for a B2B firm's site. Not the choice for a vast catalogue. Being honest about it ### When each option is the right one. #### When the alternative is the better choice - You run a large e-commerce or catalogue site where technical SEO at scale is the whole game. Hire a specialist. - Your SEO agency already reports enquiries and customers, not only rankings, and you can see the line from their work to your revenue. Leave it alone. - You have a capable marketer and developer in-house who only need audits and advice. A narrow SEO retainer will cost less than a programme. #### When we are - You pay for SEO and cannot say what you get. If the report runs to forty pages and none mentions a customer, the ranking is not the problem. - You rank well and still are not named when a buyer asks ChatGPT who to use. That is a reputation gap, and keyword work will not close it. - Recommendations pile up because nobody implements them. When one team owns the site, the content and the press, something ships every week. THE NUMBERS Website That Sells costs £5,000 to £10,000 to build and from £999 a month to run, live in 4 to 8 weeks. Reputation That Wins is from £1,999 a month. To compare with your current SEO retainer, add up twelve months of fees and set the total against the customers you can trace to that work. If you cannot trace any, fix that first, whoever you use. Prices exclude VAT. Proof it works ### What this looks like, for real. Proof it works · TCC Group Three brands, one group. AI could only see the pieces. 1 week to a plan TCC's own team built the new site from Read the story Questions founders ask ### The short answers. Is optimising for AI assistants just SEO with a new name? Partly. Clean structure, fast pages and marked-up facts help with both. The difference is what gets rewarded. Google ranks pages. AI assistants name firms, and they lean on what independent sources say: press, reviews, listings, expert pieces. A firm can rank first for a phrase and be absent from the answer. Should we sack our SEO agency? Not on our say-so. Ask them three things first: which enquiries came from their work last quarter, whether any links were paid for, and whether they track who AI assistants name for your buyers' questions. Good agencies answer all three without fuss. If yours does, keep them. How long before AI answers change? Months, not weeks. Assistants are slow to change what they say about a firm, and they trust repetition across sources, so one article changes nothing. A steady run of expert pieces, earned press and accurate listings moves it over time. Anyone promising a date is guessing. Sound familiar? ### Founders weighing this also said these. I'm paying for SEO. I don't know what I get. Ask for a dated list of what was changed on the website last month, and why. Real search work leaves evidence: pages rewritten, faults fixed, coverage earned on other sites. An agency that can show only rankings and traffic charts is reporting on the weather. The test that matters is whether enquiries from people who did not already know the firm are rising. Website That Sells Competitors rank above us. They're worse. Google cannot see how good your work is. It ranks what it can read: pages that answer the question searched, other sites that mention and link to the firm, reviews, and how recently anything was published. A weaker competitor who has described their work in public for years will outrank a better firm that has not. AI assistants draw on the same evidence. Reputation That Wins From Insights ### Three things worth reading first. Getting named by AI assistants #### Do we still need SEO now that buyers ask ChatGPT instead of Google? 22 April 2026 How buying is changing #### We rank first on Google. Why does ChatGPT not mention us? 31 March 2026 Costs, returns and the business case #### How do we choose a firm to help us get recommended by ChatGPT? 31 March 2026 ### Weigh it with someone who has done both. Thirty minutes with a founder. We check your site, two competitors and what Google and ChatGPT say about all three beforehand. You leave with three fixes in writing. Book a Growth Review --- Title: Sound familiar? 20 things founders tell us | Fortitude Media URL: https://fortitudemedia.ai/sound-familiar Summary: The 20 sentences we hear most from founder-led B2B firms, each one answered: what is really going on, what it costs, and how to check it yourself. ## The things founders tell us, answered one at a time. 20 sentences we hear from founder-led B2B firms. Pick the one that sounds most like your Monday. Each has its own page: what is really going on, what it is costing you, how to check it yourself, and what we would do in the first 90 days. Book a Growth Review 20 pages Free to read No form 5 of the 20 ### The website. Where this usually ends up: Website That Sells. “Our website undersells us.” Why does my website make my business look smaller than it is? “ChatGPT has never heard of us.” Why does ChatGPT not know my business exists? “We rebuilt it two years ago and nothing changed.” Why did our new website make no difference to enquiries? “Nobody has ever said 'I found you on your website'.” Why does nobody find our business through our website? “I'm paying for SEO. I don't know what I get.” How do I know if my SEO agency is doing anything? 5 of the 20 ### Reputation and being found. Where this usually ends up: Reputation That Wins. “Competitors rank above us. They're worse.” Why do worse competitors rank above us on Google? “I asked ChatGPT who to use. We weren't on the list.” Why doesn't ChatGPT recommend my company when people ask who to use? “Our best thinking is in our heads.” How do we turn our experts' knowledge into content when they have no time to write? “I should post on LinkedIn. I never do.” Is it worth a founder posting on LinkedIn, and how do you keep it up? “We're the best-kept secret in our industry.” How does a well-regarded B2B firm get known outside its own network? 5 of the 20 ### Pipeline and outreach. Where this usually ends up: Pipeline That Fills. “Pipeline is feast or famine.” Why is our sales pipeline feast or famine, and how do we even it out? “I am still our best salesperson.” How do I stop being the only person who can sell in my company? “We've outgrown word of mouth.” What do you do when word of mouth isn't enough to grow the business any more? “We book meetings with people who can't say yes.” Why do our sales meetings keep being with people who can't make the decision? “I delete twenty LinkedIn pitches a week. I'm not going to send them.” Can you do cold outreach on LinkedIn and email without it being spam? 5 of the 20 ### Running growth at all. Where this usually ends up: Growth Team programme. “Four agencies, nobody in charge.” Should I replace several marketing agencies with one team? “We've been at £5m for three years.” Why has my business stopped growing? “The last agency sent reports, not customers.” How do I know if my marketing agency is actually working? “I can't tell what marketing made us.” How do I work out which marketing is bringing in customers? “If I took a month off, sales would stop.” How do I make sales less dependent on me as the founder? ### Say it out loud, and we will answer it. Thirty minutes with a founder. Bring the one that sounds most like you. Leave with the three things we would fix first, in writing. Book a Growth Review --- Title: Is my marketing agency actually working? | Fortitude Media URL: https://fortitudemedia.ai/sound-familiar/agency-sent-reports-not-customers Summary: Your marketing agency sent reports, not customers. How to tell activity from results, what it cost you, and what to ask for next time. Things founders tell us · No. 18 of 20 · Running growth at all ## “The last agency sent reports, not customers.” Every month a PDF arrived. Impressions up, sessions up, a graph heading pleasingly to the right. Then you noticed the phone had not rung any more than it did before you hired them. Book a Growth Review ### How do I know if my marketing agency is actually working? A marketing agency is working if you can name customers, or at least serious sales conversations, that exist because of what it did. Traffic, impressions, rankings and followers measure activity. They only matter when enquiries from the right buyers move with them. If several months in nobody can trace one qualified conversation to the work, the reports describe effort, not results. Causes ### How you ended up with reports instead. #### They counted what was easy to count Traffic and impressions export themselves from a dashboard. Customers have to be traced through your sales records, which takes work and access nobody arranged. #### The contract paid for activity It probably promised four articles, some posts and a monthly call, and they delivered all of it. Meetings and customers were never in the document, so nobody was in breach when none turned up. #### The senior people left after the pitch The partner who impressed you handed the account to someone two years into their career. That person could run the schedule, but was never going to tell you the plan was wrong. The cost ### What it is costing you. Do the sum with your own numbers Multiply the monthly fee by the months you stayed. For example, £3,000 a month for eighteen months is £54,000. Divide that by the customers you can trace to the work. If the count is zero the division fails, which is rather the point. Then add the part no invoice shows: eighteen months in which something else could have been tried. Do it yourself ### Check it yourself in ten minutes. Find the contract and the last report they sent. - 1 Circle every figure in the report that is an enquiry, a meeting or a customer. Count the circles. - 2 Name three enquiries from the last quarter of the engagement and where each came from. - 3 Re-read the contract and note the units it promised: articles, posts, hours or meetings. - 4 Check whether the agency ever asked to see your lost deals or speak to sales. If you would rather we did it ### What we would do in the first 90 days. We agree what counts before any work starts, in writing, so the report cannot drift into decoration later. - Week 1 Growth Review with a founder. We check your site, two competitors, and what Google and ChatGPT say about all three, then send three fixes in writing. We also agree what a qualified meeting is: held, with someone who owns or materially influences the decision, at a firm matching agreed criteria, who knowingly agreed to talk about your offer. - Weeks 2 to 4 Website, reputation and outreach work starts from one plan, with outreach in warm-up for month one. - Days 30 to 90 One page arrives on the first Monday of each month: meetings held, pipeline accepted by sales, buyer questions where an AI assistant names you. On outreach you pay per meeting held. No meeting, no meeting fee. The Growth Team programme covers all three services from £2,500 a month (most clients pay £5,000 to £8,000), all prices exclude VAT, and you can stop after any step. How Growth Team programme works Questions founders ask ### The short answers. Should I give my current agency another quarter? Only with a test attached. Ask them to agree one customer-side number, such as qualified enquiries, and a date to review it. A good agency will welcome that. If the reply is a longer report, you have your answer without spending the quarter. What happens if your programme does not work either? We do not promise outcomes, and you should be wary of anyone who does. You can stop after any step. On outreach, if fewer than six qualified meetings are held in months two and three combined, month four's management fee is waived. The list and conversations are yours to keep. What should a useful marketing report look like? Short enough to read standing up. It should lead with things a finance director would recognise: conversations held, pipeline your sales side accepted, customers won. If a report needs a call to interpret it, it was written to be admired. Sound familiar? ### Founders who said this also said these. I can't tell what marketing made us. In B2B, with long sales cycles and several people in each purchase, precise attribution is not available and chasing it wastes money. A simpler discipline works. Record the source of every first conversation, ask each new buyer how they found you, and review the answers monthly against spend. That shows which activity starts conversations and which starts none. Growth Team programme We book meetings with people who can't say yes. Meetings land with the wrong people when whoever books them is measured on meetings booked, not on who turns up. Junior staff accept calls more readily than owners, so any volume target drifts towards them. Fix the definition first: a meeting counts only if the person owns or materially influences the decision, works at a firm that fits, and knows what the call is about. Pipeline That Fills All 20 things founders tell us From Insights ### Three things worth reading first. Costs, returns and the business case #### What should our monthly report on AI search show the board? 31 March 2026 Costs, returns and the business case #### Will a cheap AI search package get us recommended by ChatGPT? 31 March 2026 ### Bring this one to the call. Thirty minutes with a founder. We look at your site, two competitors and what Google and ChatGPT say about all three beforehand. You leave with three fixes in writing. Book a Growth Review --- Title: How to get known outside your network | Fortitude Media URL: https://fortitudemedia.ai/sound-familiar/best-kept-secret-in-our-industry Summary: Clients call you the best-kept secret in your industry. Why a good reputation does not travel on its own, what it costs you, and how to make it public. Things founders tell us · No. 10 of 20 · Reputation and being found ## “We're the best-kept secret in our industry.” Clients say it as a compliment, and for years you took it as one. Then you lost a tender to a firm with half your record and twice your profile, and the phrase started to sound like a diagnosis. Book a Growth Review ### How does a well-regarded B2B firm get known outside its own network? A reputation that lives only in clients' heads does not travel. To be known beyond your network, the evidence has to exist in public: expert articles under named people, coverage in the trade press your buyers read, reviews, accurate listings and a visible founder. Buyers research before they call, and can only shortlist firms they can find. Causes ### How good firms stay secret. #### Referrals hid the problem Work arrived through people who already trusted you, so you never had to explain yourself to strangers. That holds until the network stops growing: contacts retire, change jobs or get acquired. Nothing breaks. It just slows. #### Your proof is private The results you are proudest of sit in client files, under NDA or never written up. A stranger sees a tidy website and no evidence. Your competitor has press mentions and a trade magazine column, so they look the safer choice. #### Modesty, mistaken for strategy Plenty of good firms find publicity slightly vulgar and believe the work should speak for itself. The work speaks only to people who have seen it. Everyone else is listening to whoever is talking. The cost ### What it is costing you. Do the sum with your own numbers Count the tenders, pitches and shortlists in your sector last year that you only heard about afterwards. Nobody on the buying side knew your name. For example, if there were eight, you normally win one in three, and an average client is worth £15,000 a year, that is about £40,000 of annual revenue you never competed for. Use your own figures. Do it yourself ### Check it yourself in ten minutes. The test is whether a stranger could find proof that you are good. - 1 Search your firm name in a private window. Count the results that are not your own site: press, directories, reviews, podcasts. - 2 Open the latest issue of your sector's main trade publication. Note which competitors are quoted, and whether you ever were. - 3 Ask your last three new clients how they first heard of you. If every answer is a person's name, strangers have no route to you. - 4 Ask ChatGPT to name firms that do what you do in your sector. If you would rather we did it ### What we would do in the first 90 days. The reputation already exists. The job is moving it from private to public without making you sound like a firm you would dislike. - Week 1 We check your site, two competitors, and what Google and ChatGPT say about all three. Then we list the proof you hold that outsiders cannot see. - Weeks 2 to 6 First expert interview and three pieces in your voice. Reviews requested from willing clients, listings corrected, the founder's LinkedIn profile rewritten. - Weeks 7 to 12 Press angles go to the editors your buyers read, aiming for the first earned placements. We draft a data story from your own numbers. Reputation That Wins covers expert pieces, earned press, reviews, listings and the founder's LinkedIn, from £1,999 a month excluding VAT. How Reputation That Wins works Questions founders ask ### The short answers. Our clients are confidential. What can we publish? More than you think. You can publish what you know without naming who taught you: how a problem usually arises, what it costs, what to check. Aggregate numbers from your own records are usually safe. Named case studies need the client's agreement. We tried PR before and got nothing. Why would this differ? The usual cause is pitching company news no editor wants: a new hire, an office move, an award. Editors want expertise and data their readers can use. We pitch your experts' views and your numbers, aim for one to four placements a month, and never buy links. What does it cost compared with hiring someone? The 2026 UK median for a content and search manager is £42,055 (Ashdown Group guide), before employer NI, pension and software. Reputation That Wins is from £1,999 a month, about £24,000 a year excluding VAT, and you can stop after any step. Sound familiar? ### Founders who said this also said these. Competitors rank above us. They're worse. Google cannot see how good your work is. It ranks what it can read: pages that answer the question searched, other sites that mention and link to the firm, reviews, and how recently anything was published. A weaker competitor who has described their work in public for years will outrank a better firm that has not. AI assistants draw on the same evidence. Reputation That Wins We've outgrown word of mouth. Word of mouth stops being enough when the revenue you need grows faster than your network. Referrals depend on how many people know your work, and that number rises slowly. Keep them, because they convert best, and add one channel you control: researched approaches to firms that fit, backed by a website that holds up when they look you up. Pipeline That Fills All 20 things founders tell us From Insights ### Three things worth reading first. Press, reviews and being trusted #### Which kinds of PR actually win work for a B2B firm like ours? 31 March 2026 Press, reviews and being trusted #### How long does PR take to work, and what should a year look like? 31 March 2026 ### Bring this one to the call. Thirty minutes with a founder. We look at your site, two competitors and what Google and ChatGPT say about all three beforehand. You leave with three fixes in writing. Book a Growth Review --- Title: Which marketing brings in customers? | Fortitude Media URL: https://fortitudemedia.ai/sound-familiar/cannot-tell-what-marketing-made-us Summary: You cannot tell what your marketing spend produced. A plain way to track where B2B customers come from, without attribution software. Things founders tell us · No. 19 of 20 · Running growth at all ## “I can't tell what marketing made us.” You spend on a website, some events, a PR retainer, a bit of LinkedIn. New clients do arrive. When your finance director asks which of those brought them in, your answer is part memory and part hope. Book a Growth Review ### How do I work out which marketing is bringing in customers? In B2B, with long sales cycles and several people in each purchase, precise attribution is not available and chasing it wastes money. A simpler discipline works. Record the source of every first conversation, ask each new buyer how they found you, and review the answers monthly against spend. That shows which activity starts conversations and which starts none. Causes ### Why the numbers will not join up. #### Most of the buyer's research is invisible Someone hears your name from a peer, asks an AI assistant about you, reads your LinkedIn, then types your address into a browser. Your analytics calls that a direct visit. Several things worked and the software credited none of them. #### Nobody writes down the source The best moment to learn how a buyer found you is the first call, and it takes one question. In most firms the CRM field for it is blank, or says 'other'. #### The spend is scattered Events sit in the sales budget, the website in IT, PR in something the MD signs off. With no single total, there is nothing to hold results against. The cost ### What it is costing you. Do the sum with your own numbers Pull every marketing cost from last year's ledger into one list: suppliers, salaries, software, stands, sponsorships. Beside each line, mark whether you can trace one sales conversation to it. Total the lines you cannot. For example, a firm spending £120,000 might find £70,000 on that side. That money is not necessarily wasted. It is being renewed every year without anyone knowing. Do it yourself ### Check it yourself in ten minutes. You need the ledger, the CRM and a phone. - 1 Write down how each of your last ten new clients first heard of you. Where you do not know, ring three and ask. - 2 Count how many of last quarter's new opportunities have a source recorded in the CRM. - 3 Total last year's marketing spend from the ledger, every line, not just the marketing budget. - 4 Check whether your enquiry form and your first sales call both ask how the buyer found you. If you would rather we did it ### What we would do in the first 90 days. Measurement first, because everything after it is a decision about where to put money. - Week 1 Growth Review with a founder. We check your site, two competitors, and what Google and ChatGPT say about all three, then send three fixes in writing. Our platform starts asking ChatGPT, Claude, Gemini and Google your buyers' questions every week, recording who is named. - Weeks 2 to 4 We agree with you and whoever runs sales what gets recorded at every first conversation, and what counts as accepted pipeline. - Days 30 to 90 One page arrives on the first Monday of each month: meetings held, pipeline accepted by sales, buyer questions where an AI assistant names you. Site changes are dated and logged, so a shift in enquiries can be matched to a cause. The Growth Team programme covers all three services from £2,500 a month (most clients pay £5,000 to £8,000), all prices exclude VAT, and you can stop after any step. How Growth Team programme works Questions founders ask ### The short answers. Do I need attribution software? Probably not. The expensive tools mostly produce a more confident version of the wrong answer, because they cannot see the peer recommendation or the question put to an AI assistant. One mandatory CRM field and the habit of asking on the first call tell you more. How do you track whether AI assistants recommend us? By asking them, repeatedly. Our own platform puts the 40 to 150 questions your buyers ask to ChatGPT, Claude, Gemini and Google every week and records who is named. You can do a rough version yourself tonight with five questions and a notebook. What return should I expect from the programme? We will not promise one, but you can work out the threshold. A firm whose average client is worth £15,000 at a 50% margin needs about ten new clients a year to cover a £6,500-a-month programme. Run that with your own figure and see whether it looks reachable. Sound familiar? ### Founders who said this also said these. The last agency sent reports, not customers. A marketing agency is working if you can name customers, or at least serious sales conversations, that exist because of what it did. Traffic, impressions, rankings and followers measure activity. They only matter when enquiries from the right buyers move with them. If several months in nobody can trace one qualified conversation to the work, the reports describe effort, not results. Growth Team programme Nobody has ever said 'I found you on your website'. Strangers find a website only when its pages match what they search for or ask an AI assistant. Most B2B sites are organised around the firm: about us, our services, our team. Buyers search for their problem, a price, a comparison or a place. If no page answers those, the only visitors are people who already know the company name. Website That Sells All 20 things founders tell us From Insights ### Three things worth reading first. Costs, returns and the business case #### How can we tell which of our new leads came from ChatGPT? 31 March 2026 Websites that bring work in #### How do we measure the return we get from our website? 31 March 2026 Costs, returns and the business case #### How do we find out whether ChatGPT is recommending our firm? 31 March 2026 ### Bring this one to the call. Thirty minutes with a founder. We look at your site, two competitors and what Google and ChatGPT say about all three beforehand. You leave with three fixes in writing. Book a Growth Review --- Title: Why doesn't ChatGPT recommend my company? | Fortitude Media URL: https://fortitudemedia.ai/sound-familiar/chatgpt-did-not-recommend-us Summary: You asked ChatGPT who to use in your sector and your firm was not on the list. How AI assistants choose the names they give, and what gets you added. Things founders tell us · No. 7 of 20 · Reputation and being found ## “I asked ChatGPT who to use. We weren't on the list.” You typed in the question your buyers ask. Back came three names, confidently described. Yours was not among them, and one was a firm you regularly beat. Book a Growth Review ### Why doesn't ChatGPT recommend my company when people ask who to use? AI assistants name firms they can find described consistently in several independent places: trade press, directories, reviews, comparison articles and the firm's own site. They return two or three names, so there is no page two. An excellent firm written about nowhere except its own website gives the assistant nothing to check against, so it goes with names it can verify. Causes ### Why the list looks the way it does. #### It trusts what others say about you, not what you say Your website is a claim. A mention in a trade publication, a directory entry and a run of reviews are corroboration. Assistants weigh the second kind heavily, and most referral-built firms have almost none. #### The record about you is thin or contradictory An old address on one listing, a former trading name on another, a description written in 2017 by someone long gone. When sources disagree, an assistant hedges. Hedging means leaving you out. #### Nobody has written down what you are best at The assistant was asked who is good at a specific thing for a specific kind of client. If no page anywhere says plainly that you do that thing for those people, it cannot match you to the question. The cost ### What it is costing you. Do the sum with your own numbers You cannot see these enquiries go missing, which is what makes it expensive. Count the new clients you won last year who researched before they called. For example, if ten did and an average client is worth £15,000, that is £150,000 of revenue passing through research that increasingly ends in a shortlist of two or three names. You are not currently one of them. Do it yourself ### Check it yourself in ten minutes. Use a fresh chat, logged out, so the assistant is not tailoring its answers to you. - 1 Write five questions as a buyer would ask them, with sector and location but not your firm name. Put each to ChatGPT, Claude and Gemini. - 2 Record which firms are named. The same two or three keep appearing. Those are your real competitors. - 3 Ask each assistant what it knows about your firm by name. Note anything wrong or out of date. - 4 Ask where it got its information about the firm it named first. Those sources are the places you are absent from. If you would rather we did it ### What we would do in the first 90 days. The work is making you easy to verify. Most of it is unglamorous. - Week 1 We agree the list with you, usually 40 to 150 questions your buyers ask, and record who ChatGPT, Claude, Gemini and Google name for each. - Weeks 2 to 5 Listings, profiles and reviews are corrected so every source says the same thing about you. First expert interview, first three pieces published. - Weeks 6 to 12 Press angles go out, aiming for the first earned placements. Our platform re-asks the question list every week, so you can see where you are now named. Reputation That Wins is from £1,999 a month excluding VAT, and movement in AI answers takes months, not weeks. How Reputation That Wins works Questions founders ask ### The short answers. Can you guarantee ChatGPT will name us? No, and be wary of anyone who does. Nobody outside those companies controls what the assistants say. What can be controlled is the evidence: what is published about you and how consistently. We record the answers weekly, so you can watch it. Do our buyers really ask ChatGPT? Ask your last five new clients how they drew up their shortlist. Some will say a colleague, some Google, and some now say an AI assistant. Even if none do, the same public record feeds Google's results, so the work is not wasted. What if we spend six months and nothing changes? You can stop after any step, so the risk is bounded. The weekly record means you are not relying on our opinion: either you appear in more answers than at the baseline or you do not. If nothing had moved, we would say so. Sound familiar? ### Founders who said this also said these. We're the best-kept secret in our industry. A reputation that lives only in clients' heads does not travel. To be known beyond your network, the evidence has to exist in public: expert articles under named people, coverage in the trade press your buyers read, reviews, accurate listings and a visible founder. Buyers research before they call, and can only shortlist firms they can find. Reputation That Wins ChatGPT has never heard of us. AI assistants know a business when they can read clear facts about it on its own website and find the same facts repeated elsewhere: directories, press, reviews, LinkedIn. When a site is thin, hard for machines to read, or contradicted by old listings, the assistant has nothing reliable to repeat, so it says nothing or guesses. Age is no protection. Website That Sells All 20 things founders tell us From Insights ### Three things worth reading first. Costs, returns and the business case #### How do we find out whether ChatGPT is recommending our firm? 31 March 2026 Press, reviews and being trusted #### How does ChatGPT decide which facts about our firm to believe? 31 March 2026 Getting named by AI assistants #### Why does ChatGPT describe my firm wrongly, and how do I fix it? 31 March 2026 ### Bring this one to the call. Thirty minutes with a founder. We look at your site, two competitors and what Google and ChatGPT say about all three beforehand. You leave with three fixes in writing. Book a Growth Review --- Title: Why has ChatGPT never heard of us? | Fortitude Media URL: https://fortitudemedia.ai/sound-familiar/chatgpt-has-never-heard-of-us Summary: You asked ChatGPT about your own firm and it knew nothing, or got it wrong. Why AI assistants cannot see some businesses, and what to fix first. Things founders tell us · No. 2 of 20 · The website ## “ChatGPT has never heard of us.” You typed your own company name into ChatGPT. It either apologised for knowing nothing or described a different firm with a similar name. Years of trading, and it drew a blank. Book a Growth Review ### Why does ChatGPT not know my business exists? AI assistants know a business when they can read clear facts about it on its own website and find the same facts repeated elsewhere: directories, press, reviews, LinkedIn. When a site is thin, hard for machines to read, or contradicted by old listings, the assistant has nothing reliable to repeat, so it says nothing or guesses. Age is no protection. Causes ### Why the machines draw a blank. #### Your site never states the basics Plenty of B2B sites never say, in one sentence, what the firm does, for whom and where. A person infers it from the client logos. A machine reads text, and from slogans it learns nothing. #### The facts about you disagree An old address on one directory, a former trading name on another, something else at Companies House. AI assistants cross-check sources. When the sources conflict, the safe course is to leave you out. #### The good material is locked away Case studies in PDFs, services described inside images, pages that appear only after scripts run. Some of the programs that read websites for AI assistants cope badly with all three. The cost ### What it is costing you. Do the sum with your own numbers A buyer who has been referred to you may ask an AI assistant about you before replying, the way they once searched your name. A blank or wrong answer loses some of those replies. For example: 30 referrals a year, one in ten lost, average client worth £15,000, comes to £45,000. Put in your own figures. Then add the buyers who were never given your name. Do it yourself ### Check it yourself in ten minutes. Use a fresh chat with no history, in more than one assistant. - 1 Ask ChatGPT, Claude and Gemini what your company does and where it is based. Note what is wrong, not only what is missing. - 2 Ask each the question a buyer would ask, such as who supplies your service in your region. Write down the names. - 3 Search your company name on Google. Check address, phone number and trading name on every first-page listing. - 4 Paste your homepage text into a blank document. That is roughly what a machine reads. See whether it says what you do, for whom and where. If you would rather we did it ### What we would do in the first 90 days. Being known comes before being recommended. Expect AI answers to move over months, not weeks. - Week 1 We agree the 40 to 150 questions your buyers ask and start putting them to ChatGPT, Claude, Gemini and Google every week, recording who is named. That is your baseline. - Weeks 2 to 8 We build pages that answer those questions, answer first, with prices, timelines, locations and credentials marked up so machines can read them. You also get a list of the conflicting facts we found elsewhere. - Weeks 9 to 13 Weekly dated improvements, steered by the questions that still return nothing about you. This is Website That Sells: build £5,000 to £10,000, run from £999 a month, live in 4 to 8 weeks, all excluding VAT. How Website That Sells works Proof it works · ID-Pal AI could describe ID-Pal. It never recommended it. 0 4 UK buyer questions where AI now names ID-Pal Read the story Proof it works · TCC Group Three brands, one group. AI could only see the pieces. 1 week to a plan TCC's own team built the new site from Read the story Questions founders ask ### The short answers. How long before ChatGPT starts to know us? Months, not weeks, and nobody controls the timing. Assistants that search the web as they answer can pick up a clear new page sooner. What the underlying models know changes slowly, when they are retrained. Hence the weekly record: you watch movement question by question instead of hoping. Can we fix this ourselves? A good part of it. Correcting directory listings and writing one plain page that says what you do, for whom, where and for how much needs no specialist. The markup and the weekly tracking across four systems are tedious by hand. What does it cost to find out where we stand? Nothing. The Growth Review is free: 30 minutes with a founder. Beforehand we check your site, two competitors, and what Google and ChatGPT say about all three. Afterwards you get three fixes in writing, which you can hand to whoever you like. Sound familiar? ### Founders who said this also said these. Nobody has ever said 'I found you on your website'. Strangers find a website only when its pages match what they search for or ask an AI assistant. Most B2B sites are organised around the firm: about us, our services, our team. Buyers search for their problem, a price, a comparison or a place. If no page answers those, the only visitors are people who already know the company name. Website That Sells I asked ChatGPT who to use. We weren't on the list. AI assistants name firms they can find described consistently in several independent places: trade press, directories, reviews, comparison articles and the firm's own site. They return two or three names, so there is no page two. An excellent firm written about nowhere except its own website gives the assistant nothing to check against, so it goes with names it can verify. Reputation That Wins All 20 things founders tell us From Insights ### Three things worth reading first. Getting named by AI assistants #### Why does ChatGPT describe my firm wrongly, and how do I fix it? 31 March 2026 Getting named by AI assistants #### Do we need schema markup for AI assistants to understand our site? 22 April 2026 Costs, returns and the business case #### How do we find out whether ChatGPT is recommending our firm? 31 March 2026 ### Bring this one to the call. Thirty minutes with a founder. We look at your site, two competitors and what Google and ChatGPT say about all three beforehand. You leave with three fixes in writing. Book a Growth Review --- Title: Why do worse rivals rank above us? | Fortitude Media URL: https://fortitudemedia.ai/sound-familiar/competitors-rank-above-us Summary: Your competitors rank higher on Google and you know their work is worse. Why it happens, what it costs, and how to check it yourself in ten minutes. Things founders tell us · No. 6 of 20 · Reputation and being found ## “Competitors rank above us. They're worse.” You know their work. You have picked up clients they let down. Yet when someone searches for what you both do, they are near the top and you are on page two. Book a Growth Review ### Why do worse competitors rank above us on Google? Google cannot see how good your work is. It ranks what it can read: pages that answer the question searched, other sites that mention and link to the firm, reviews, and how recently anything was published. A weaker competitor who has described their work in public for years will outrank a better firm that has not. AI assistants draw on the same evidence. Causes ### What is really going on. #### Quality is invisible to a search engine Your delivery happens where Google never sees it. What it reads is the public record: what you have published, who mentions you, what reviewers say. On that record, they have more in the file. #### They have been at it longer, and boringly Most firms that rank well did nothing clever. They published monthly, got quoted in the trade press now and then, and kept asking for reviews. Five years of that is a lead you cannot close in a quarter. #### Your pages describe you, theirs answer the buyer What ranks is usually a page answering a specific question: what it costs, how long it takes, what goes wrong. Your services page says you are experienced and client-focused, which is what everybody's says. The cost ### What it is costing you. Do the sum with your own numbers Count last year's enquiries from people who did not already know you. For most referral-built firms it is close to zero. For example: if the competitor above you gets two such enquiries a month, converts one in four, and an average client is worth £15,000, that position is worth £90,000 a year to them. Use your own client value. Then remember they are worse. Do it yourself ### Check it yourself in ten minutes. Use a private browser window, so your search history does not flatter you. - 1 Search the three phrases a buyer would use for what you do. Write down who appears above you. - 2 Open the top competitor's ranking page beside your equivalent. Note which answers a question and which describes a company. - 3 Search their firm name, then yours. Count the first-page results that are not the firm's own site. - 4 Compare review counts, the date of the latest review, and the date each of you last published anything. If you would rather we did it ### What we would do in the first 90 days. We would not try to out-trick them. We would build the public record your work deserves, and measure whether it moves anything. - Week 1 We compare your site, two competitors, and what Google and ChatGPT say about all three. You get the gap in writing. - Weeks 2 to 6 The first 30-minute interview with your experts becomes three pieces in your voice. We correct reviews and listings, and pitch the first press angles. - Weeks 7 to 12 Two more rounds of pieces, with the first earned placements due. We draft a data story from your own numbers, and tell you plainly what has moved. This is Reputation That Wins: three expert pieces and one to four earned press placements a month, from £1,999 a month excluding VAT. How Reputation That Wins works Proof it works · ID-Pal AI could describe ID-Pal. It never recommended it. 0 4 UK buyer questions where AI now names ID-Pal Read the story Proof it works · TCC Group Three brands, one group. AI could only see the pieces. 1 week to a plan TCC's own team built the new site from Read the story Questions founders ask ### The short answers. How long before we overtake them? Months, not weeks, and we will not promise a position. A competitor with years of published work has a real lead. Early on, expect dated, visible activity: pieces published, placements earned, listings corrected. Movement in rankings and AI answers follows, slowly at first. Can't we pay for links or ads and jump the queue? Ads work while you pay and stop when you stop, which is fine if you know that going in. Bought links are different: Google is good at spotting them, and the downside outweighs the upside. Earned mentions are slower, and they are what AI assistants notice. Could we do this ourselves? Yes. Most of it is discipline, not secret knowledge: interview your best people monthly, publish what they say, ask happy clients for reviews, pitch the trade press. It fails because it is nobody's job and client work always wins. If someone can have a day a week, start there. Sound familiar? ### Founders who said this also said these. I asked ChatGPT who to use. We weren't on the list. AI assistants name firms they can find described consistently in several independent places: trade press, directories, reviews, comparison articles and the firm's own site. They return two or three names, so there is no page two. An excellent firm written about nowhere except its own website gives the assistant nothing to check against, so it goes with names it can verify. Reputation That Wins I'm paying for SEO. I don't know what I get. Ask for a dated list of what was changed on the website last month, and why. Real search work leaves evidence: pages rewritten, faults fixed, coverage earned on other sites. An agency that can show only rankings and traffic charts is reporting on the weather. The test that matters is whether enquiries from people who did not already know the firm are rising. Website That Sells All 20 things founders tell us From Insights ### Three things worth reading first. How buying is changing #### We rank first on Google. Why does ChatGPT not mention us? 31 March 2026 Press, reviews and being trusted #### Why does ChatGPT sound unsure about us, and do backlinks fix it? 31 March 2026 Costs, returns and the business case #### How do we compare with our competitors when buyers ask ChatGPT? 31 March 2026 ### Bring this one to the call. Thirty minutes with a founder. We look at your site, two competitors and what Google and ChatGPT say about all three beforehand. You leave with three fixes in writing. Book a Growth Review --- Title: Too many agencies and nobody in charge | Fortitude Media URL: https://fortitudemedia.ai/sound-familiar/four-agencies-nobody-in-charge Summary: Four marketing agencies, four reports and no one who owns the result. Why it happens, what it costs, and how to put one person in charge of growth. Things founders tell us · No. 16 of 20 · Running growth at all ## “Four agencies, nobody in charge.” You have an SEO firm, a PR consultant, someone running paid ads and a developer on retainer. Each sends a monthly report saying their part went well. The only person who sees the whole picture is you, usually on a Sunday night. Book a Growth Review ### Should I replace several marketing agencies with one team? Usually yes, unless someone other than the founder already owns the result. Several specialist agencies can work, but only when one named person sets a single plan, decides priorities and answers for meetings and pipeline. Without that person, each supplier optimises its own report, gaps go unnoticed and the founder becomes an unpaid project manager. Fix ownership first, then count suppliers. Causes ### What is really going on. #### Each agency is marked on its own homework The SEO firm reports rankings, the PR consultant reports coverage, the ads people report clicks. All can rise in a month when no new customer appears, and nobody's fee depends on the number you care about. #### The gaps sit between the suppliers PR lands a good article and the website has no page that picks up the reader. Each supplier did its job. The join between the jobs belonged to no one. #### You became the integrator by default Someone has to brief four firms and spot what is missing. That fell to you, and it gets done in the gaps between everything else, which is to say badly. The cost ### What it is costing you. Do the sum with your own numbers Add up the retainers and multiply by twelve. For example, four suppliers at £2,000 a month is £96,000 a year. Then price the hours you spend each week briefing and chasing them at what an hour of your own selling brings in. Now write down the customers the whole arrangement can prove it produced last year. Most founders find the first number easy and the second one awkward. Do it yourself ### Check it yourself in ten minutes. You need your bank statements, the last round of reports and a pen. - 1 List every marketing supplier, the monthly fee, and the one number each reports to you. - 2 Beside each, write the last customer you can trace to their work. - 3 Ask each supplier what the others are working on this month. Note who knows. - 4 Pick one recent press article or campaign and check whether the website and your sales emails used it. If you would rather we did it ### What we would do in the first 90 days. One owner, one plan, one set of numbers. - Week 1 A founder of ours runs your Growth Review. We check your site, two competitors, and what Google and ChatGPT say about all three, then send three fixes in writing. - Weeks 2 to 4 We write a single plan covering website, reputation and pipeline, with one person answerable for it. It says what stops, what carries on and why. - Days 30 to 90 Site pages, expert pieces and researched outreach run from that one plan. On the first Monday of each month you get one page: meetings held, pipeline accepted by sales, buyer questions where an AI assistant names you. The Growth Team programme covers all three services from £2,500 a month (most clients pay £5,000 to £8,000), all prices exclude VAT, and you can stop after any step. How Growth Team programme works Questions founders ask ### The short answers. Do I have to sack all my agencies on day one? No. The problem is the missing owner, not the number of invoices. The programme covers website, reputation and pipeline, so overlap with existing suppliers will show up quickly. What you do about each one is your decision, and you can make it after the free Growth Review. What does it cost next to what I pay now? Compare it with the sum above, including your own hours. The programme starts at £2,500 a month and most clients pay £5,000 to £8,000, excluding VAT. Before that there is an optional two-week Growth Plan at £1,500, credited against the programme. The Growth Review costs nothing. Could I hire a marketing manager to run the agencies instead? You could. The 2026 UK median salary for a marketing manager is £56,950 before employer NI, pension and software, and you would still pay the retainers on top. Coordination would improve. Overruling an agency on strategy usually takes a more senior hire. Sound familiar? ### Founders who said this also said these. The last agency sent reports, not customers. A marketing agency is working if you can name customers, or at least serious sales conversations, that exist because of what it did. Traffic, impressions, rankings and followers measure activity. They only matter when enquiries from the right buyers move with them. If several months in nobody can trace one qualified conversation to the work, the reports describe effort, not results. Growth Team programme I'm paying for SEO. I don't know what I get. Ask for a dated list of what was changed on the website last month, and why. Real search work leaves evidence: pages rewritten, faults fixed, coverage earned on other sites. An agency that can show only rankings and traffic charts is reporting on the weather. The test that matters is whether enquiries from people who did not already know the firm are rising. Website That Sells All 20 things founders tell us From Insights ### Three things worth reading first. How buying is changing #### Should I still pay a marketing agency now that AI can do the work? 31 March 2026 Costs, returns and the business case #### Can we just do content, or do we need PR and a better website too? 31 March 2026 Costs, returns and the business case #### How do we choose a firm to help us get recommended by ChatGPT? 31 March 2026 ### Bring this one to the call. Thirty minutes with a founder. We look at your site, two competitors and what Google and ChatGPT say about all three beforehand. You leave with three fixes in writing. Book a Growth Review --- Title: How to stop being the only salesperson | Fortitude Media URL: https://fortitudemedia.ai/sound-familiar/i-am-still-our-best-salesperson Summary: If you are still your firm's best salesperson, every deal waits for your diary. What to hand over first, what to keep, and what it costs to do neither. Things founders tell us · No. 12 of 20 · Pipeline and outreach ## “I am still our best salesperson.” You win most of what you pitch for, and nobody else in the firm comes close. It used to be a compliment. Now it means every deal waits for a gap in your diary. Book a Growth Review ### How do I stop being the only person who can sell in my company? A founder usually outsells everyone because they carry authority, know every past project and can change the offer in the room. None of that transfers by hiring a salesperson. What does transfer is the work before the meeting: finding the right firms, researching them and opening the conversation. Hand that over first and keep the closing meetings. Causes ### Why nobody else sells like you. #### You can say yes in the room A buyer talking to you is talking to the person who can move the price, the scope and the start date. Anyone else has to go away and ask, and buyers notice. #### The pitch lives in your head Your best lines, the client stories, the objections you have heard a hundred times: none of it is written down. New salespeople get a deck and a login and are expected to absorb the rest. #### You do the cheap work as well as the expensive work Closing needs you. Finding firms, researching them and chasing replies does not, but it fills the same diary. The cost ### What it is costing you. Do the sum with your own numbers Add up the hours you spent on selling last month, including the searching, the chasing and the proposals. Split them into time in front of someone who could say yes, and everything else. For example, if selling takes 60 hours a month and 20 are in real meetings, 40 go on work somebody else could do. Multiply by what an hour of your time is worth to the firm. Do it yourself ### Check it yourself in ten minutes. Four small jobs that show how much of the selling only you can do. - 1 Highlight every sales activity in your calendar for the last four weeks. Mark each as a meeting with a decision-maker, or as preparation and chasing. - 2 List the last ten deals won, with who opened the conversation and who closed it. Count how often your name is in both columns. - 3 Record your next sales call, with permission. Write down the three things you said that moved it forward. - 4 Ask the second-best seller in your firm what they would need to win without you in the room. Write it down without arguing. If you would rather we did it ### What we would do in the first 90 days. We would not try to replace you in the meeting. We would take everything before the meeting off your desk. - Week 1 A Growth Review with a founder, then a conversation about how you sell: who you win with, what you say, which objections come up. - Weeks 2 to 4 Month one is warm-up. We agree the criteria, then research each firm and each person. A person writes and approves every message. - Months 2 and 3 80 to 120 approaches a month. Meetings arrive in your diary with the research behind them, and you do the part only you can do. Pipeline That Fills costs £1,500 to set up, £1,000 to £1,500 a month to manage, and £500 to £1,000 per meeting held, excluding VAT. No meeting, no meeting fee. How Pipeline That Fills works Questions founders ask ### The short answers. Should I just hire a sales director? Perhaps, in time. A sales director hired into a firm with no written pitch and no steady supply of first meetings has to build both before selling anything, on a senior salary. Get the meetings flowing and the pitch written down first. Will prospects be put off that I didn't find them myself? Not if the approach is a good one. Every message is written and approved by a person and has to name something true about the firm that is not on its homepage. When someone agrees to talk, they talk to you. How much of my time will this take? Most of it is at the start: agreeing who you want to meet and what counts as qualified. After that your time goes on the meetings themselves, plus a one-page report on the first Monday of each month. If it starts eating your week, it is not doing its job. Sound familiar? ### Founders who said this also said these. Pipeline is feast or famine. Pipeline swings between feast and famine when selling only happens in the gaps between delivery. The people who sell are the people who deliver, so prospecting stops whenever the firm is busy, and the hole appears one sales cycle later. The fix is a small, fixed amount of outbound activity every week that carries on however busy delivery gets. Pipeline That Fills If I took a month off, sales would stop. Separate the parts of selling that need the founder from the parts the founder merely happens to do. Finding prospects, first approaches, follow-up and proving the firm's credibility can all run without you once written down and owned by someone. Closing the largest deals may stay with you for years. The aim is conversations that keep arriving while you are away. Growth Team programme All 20 things founders tell us From Insights ### Three things worth reading first. Content that gets quoted #### Can we outsource our content without losing our voice? 22 April 2026 Costs, returns and the business case #### How do I get my board to approve budget for AI search? 22 April 2026 ### Bring this one to the call. Thirty minutes with a founder. We look at your site, two competitors and what Google and ChatGPT say about all three beforehand. You leave with three fixes in writing. Book a Growth Review --- Title: Should a founder post on LinkedIn? | Fortitude Media URL: https://fortitudemedia.ai/sound-familiar/i-should-post-on-linkedin Summary: You know you should post on LinkedIn and you never do. Whether it is worth a founder's time, why it keeps slipping, and a system that survives busy months. Things founders tell us · No. 9 of 20 · Reputation and being found ## “I should post on LinkedIn. I never do.” It has been on your list for years. You open the app, read what somebody's morning run taught them about leadership, and close it again. Meanwhile a competitor you rate less posts every week, and people mention it. Book a Growth Review ### Is it worth a founder posting on LinkedIn, and how do you keep it up? For a founder-led B2B firm, usually yes. Buyers look up the founder before a call, after a cold email and during a tender, and an empty profile tells them nothing. What works is specific opinions and lessons from real work, not motivational content. Keeping it up needs a system: one recorded conversation a month yields several weeks of posts. Causes ### Why it keeps not happening. #### You dislike what LinkedIn rewards Much of the feed is self-congratulation, and you do not want to be that person. Reasonable. But the alternative to bad posting is not silence. It is saying something useful about your trade in your normal voice. #### A blank box at nine in the evening Posting fails at the moment of writing. You have opinions all day in meetings, then sit down at night with nothing to say. The ideas were never captured, so every post starts from zero. #### No deadline and no visible return Nobody chases you for a post. The first ten get a few likes from staff and your mum. Without a number that matters, such as profile views from buyers, it quietly stops by week three. The cost ### What it is costing you. Do the sum with your own numbers LinkedIn shows how many people viewed your profile in the last 90 days. Each was deciding something about you: a prospect after your email, a buyer before a meeting, a candidate. For example, if 300 people looked last quarter and found a job title and a post from 2019, that is 300 first impressions spent on nothing. Multiply by four for the year. Do it yourself ### Check it yourself in ten minutes. Do this on your phone, where buyers will look at you. - 1 Read your headline and About section as a stranger would. Do they say who you help, or only your job title? - 2 Open your profile views for the last 90 days. Count how many viewers look like buyers. - 3 Note the date of your last post and whether a client would have found it useful. - 4 Find two competitors who post regularly. Note which posts drew comments from likely clients. If you would rather we did it ### What we would do in the first 90 days. Your part is one conversation a month. The opinions stay yours. The admin does not. - Week 1 We rewrite your profile so it says who you help and what you know. Then the first interview: 30 minutes on what you see in your market and what annoys you about it. - Weeks 2 to 4 That interview becomes six to eight posts for the month, in your voice, which you approve. Comments and replies are handled by a person, not software. - Weeks 5 to 12 Two more interviews, two more months of posts. We watch which subjects bring conversations with the right people, and drop what does not. Founder LinkedIn on its own is from £1,500 a month, and it is included in Reputation That Wins from £1,999 a month, both excluding VAT. How Reputation That Wins works Questions founders ask ### The short answers. Isn't it fake if someone else writes my posts? It would be if we invented the opinions. Every post starts from something you said in the interview, and you approve each one. Chief executives have used speechwriters for a century. What matters is whether the views are yours, and they are. How long before it brings in business? Expect a quiet first two months. The early signs are profile views from buyers, connection requests, and clients mentioning a post. Enquiries come later and irregularly, and we will not promise a number. It works best alongside outreach, because people look you up after an email. Can't my marketing person do this? Possibly, if they can interview you properly and write in your voice, not the company's. The usual failure is posts drifting into brand announcements, which nobody reads. Give them 30 minutes and see whether the drafts sound like you on a good day. Sound familiar? ### Founders who said this also said these. Our best thinking is in our heads. Stop asking experts to write. Interview them. Thirty minutes of a specialist answering real client questions, recorded and transcribed, holds enough material for several articles. A writer who understands the subject shapes it, and the expert corrects the draft in ten minutes. The expert's job is to know things. Writing was never the scarce skill. Their knowledge is. Reputation That Wins I delete twenty LinkedIn pitches a week. I'm not going to send them. Yes, if the volume is low enough for each message to be researched and written by a person. Outreach becomes spam when the same words go to hundreds of people. A message that names something true about the recipient's firm, gives a reason to talk now, and could not have been sent to anyone else reads as a business letter. Pipeline That Fills All 20 things founders tell us From Insights ### Three things worth reading first. Content that gets quoted #### Can ChatGPT tell real expertise from keyword-stuffed content? 31 March 2026 Getting named by AI assistants #### How often do we need to publish for AI assistants to notice us? 22 April 2026 ### Bring this one to the call. Thirty minutes with a founder. We look at your site, two competitors and what Google and ChatGPT say about all three beforehand. You leave with three fixes in writing. Book a Growth Review --- Title: LinkedIn outreach without the spam | Fortitude Media URL: https://fortitudemedia.ai/sound-familiar/i-will-not-send-linkedin-spam Summary: You delete LinkedIn pitches and will not send them. How cold outreach by email and LinkedIn avoids being spam: low volume, real research, a person writing. Things founders tell us · No. 15 of 20 · Pipeline and outreach ## “I delete twenty LinkedIn pitches a week. I'm not going to send them.” You know the ones: 'Hi {first name}, I came across your profile and was impressed.' You delete them before the second line, and your firm's name will never be on one. Quite right. Book a Growth Review ### Can you do cold outreach on LinkedIn and email without it being spam? Yes, if the volume is low enough for each message to be researched and written by a person. Outreach becomes spam when the same words go to hundreds of people. A message that names something true about the recipient's firm, gives a reason to talk now, and could not have been sent to anyone else reads as a business letter. Causes ### Why most outreach is spam. #### The economics reward volume Sending 800 messages costs almost the same as sending 80. So most senders send 800, accept a tiny reply rate and never count the people who now think less of them. #### Personalisation is a mail merge Your first name, your company name and a line about a recent post. You spot the template because six other firms have sent it to you. #### You ruled out the channel, not the method The pitches you delete are bad because they are lazy, not because they are direct. A short, informed note from one business owner to another is older than LinkedIn. Refusing to spam is right, but refusing to write to anyone is expensive. The cost ### What it is costing you. Do the sum with your own numbers List 50 firms you would love as clients and have never spoken to. Be realistic about how many would take a call if approached well: say, for example, one in ten. That is five conversations. Apply your usual win rate and average client value. If you win one in four at £40,000, those 50 names are worth about £50,000 a year that you are not pitching for. Do it yourself ### Check it yourself in ten minutes. Use the test you apply to other people's messages. - 1 Find the last LinkedIn pitch you replied to, if there is one. Write down what it did that the others did not. - 2 Pick one firm you want as a client. Find one true thing about it that is not on its homepage: a hire, a contract, a Companies House filing. - 3 Write four sentences to the person who decides there, using that fact and a reason to talk this month. - 4 Read it back. Could it be sent to anyone else? Then it is a template. Start again. If you would rather we did it ### What we would do in the first 90 days. Every message passes three tests: it names something true about the firm that is not on their homepage, gives a reason to talk this month, and could not be sent to anyone else. - Week 1 A Growth Review with a founder. Then we agree, in writing, the criteria for the firms you want to meet and what counts as a qualified meeting. - Weeks 2 to 4 Month one is warm-up. Research runs in two passes, firm then person. Research, timing and follow-up reminders are automated. The writing is not. - Months 2 and 3 80 to 120 approaches a month, not 800, by email and LinkedIn. After four unanswered emails we stop. Nobody gets a note that begins 'just bumping this up'. Pipeline That Fills: set-up £1,500, management £1,000 to £1,500 a month, £500 to £1,000 per meeting held, no meeting, no meeting fee, all excluding VAT. How Pipeline That Fills works Questions founders ask ### The short answers. Won't this damage our reputation? Bulk outreach can, which is why we don't do it. At 80 to 120 approaches a month, each one is researched and written by a person, and we stop after four unanswered emails. The worst a recipient gets is a short, relevant note they chose not to answer. Is AI writing the messages? No. Research, timing and follow-up reminders are automated, because software is good at those. A person writes and approves every message, because software writing unsupervised is how you get the pitches you delete. If a message can't pass the three tests, it doesn't go. Can't I do this myself? You can, and the self-check above is the whole method. The hard part is doing it a hundred times a month while running a firm, because the research takes longer than the writing. If the hours keep vanishing, hand it over. Sound familiar? ### Founders who said this also said these. We book meetings with people who can't say yes. Meetings land with the wrong people when whoever books them is measured on meetings booked, not on who turns up. Junior staff accept calls more readily than owners, so any volume target drifts towards them. Fix the definition first: a meeting counts only if the person owns or materially influences the decision, works at a firm that fits, and knows what the call is about. Pipeline That Fills I should post on LinkedIn. I never do. For a founder-led B2B firm, usually yes. Buyers look up the founder before a call, after a cold email and during a tender, and an empty profile tells them nothing. What works is specific opinions and lessons from real work, not motivational content. Keeping it up needs a system: one recorded conversation a month yields several weeks of posts. Reputation That Wins All 20 things founders tell us From Insights ### Three things worth reading first. Websites that bring work in #### What makes a buyer trust our website enough to get in touch? 31 March 2026 How buying is changing #### Should I still pay a marketing agency now that AI can do the work? 31 March 2026 How buying is changing #### What should my website do now that visitors arrive pre-sold by AI? 31 March 2026 ### Bring this one to the call. Thirty minutes with a founder. We look at your site, two competitors and what Google and ChatGPT say about all three beforehand. You leave with three fixes in writing. Book a Growth Review --- Title: Make sales less dependent on the founder | Fortitude Media URL: https://fortitudemedia.ai/sound-familiar/if-i-stop-selling-growth-stops Summary: If you took a month off, sales would stop. How to find out how founder-dependent your sales are, and what to hand over first. Things founders tell us · No. 20 of 20 · Running growth at all ## “If I took a month off, sales would stop.” You have not had a proper holiday in years, and it is not because you dislike beaches. Every serious deal starts and ends with you. The team is good at the work, but the work arrives because you went and got it. Book a Growth Review ### How do I make sales less dependent on me as the founder? Separate the parts of selling that need the founder from the parts the founder merely happens to do. Finding prospects, first approaches, follow-up and proving the firm's credibility can all run without you once written down and owned by someone. Closing the largest deals may stay with you for years. The aim is conversations that keep arriving while you are away. Causes ### Why it all still runs through you. #### Buyers trust you, not the firm Your reputation lives in your phone and in rooms you have stood in. A stranger looks up the firm and finds a services page and a team photo, which persuades nobody. #### Nobody else opens conversations You start every conversation as well as closing it, through lunches, introductions and people you have known for years. The team can take a warm lead forward, but nobody else creates one. #### The method was never written down You know which firms to call, what change makes them ready to buy, and what to say first. It never needed writing down because you were always there. The cost ### What it is costing you. Do the sum with your own numbers Take last year's new business and mark every deal you personally started or closed. For example, if £1m of £1.2m came through you, that £1m rests on one person's health and attention. Divide by twelve for the price of a month off. Anyone who might one day buy your firm will do the same sum, and take it off the offer. Do it yourself ### Check it yourself in ten minutes. Your diary and your deal list are enough. - 1 From your diary, total last month's selling hours. Split them into finding, first conversations and closing. - 2 List your last ten wins and mark who started each conversation. - 3 Ask your number two to explain, without you in the room, why clients choose you. Compare it with the website. - 4 Ask ChatGPT about yourself by name, then about the firm. Compare how much it knows about each. If you would rather we did it ### What we would do in the first 90 days. The job is to move what you know out of your head and into things that work while you are elsewhere. - Week 1 Growth Review with a founder. We check your site, two competitors, and what Google and ChatGPT say about all three, then send three fixes in writing. Then a first 30-minute interview with you about who buys and why. - Weeks 2 to 6 That interview becomes site pages answering the questions buyers put to you, and expert pieces in your voice. Your LinkedIn profile is rewritten, with six to eight posts a month from the same interview. - Days 45 to 90 After a warm-up month, researched outreach runs at 80 to 120 approaches a month, each written and approved by a person. You decide which meetings need you. The Growth Team programme covers all three services from £2,500 a month (most clients pay £5,000 to £8,000), all prices exclude VAT, and you can stop after any step. How Growth Team programme works Questions founders ask ### The short answers. Should I hire a sales director instead? Possibly, but look at what they would walk into. They still need conversations to have and public proof to point at, and at the moment you supply both. Hire one into a firm with neither and you have an expensive person waiting for your introductions. Will it still sound like me? It should, because it starts with you talking. One 30-minute interview a month becomes the expert pieces and the LinkedIn posts, with your opinions left in. Outreach messages are written and approved by a person, and LinkedIn engagement is done by a person too. When can I take the month off? Not this quarter. A rebuilt site takes 4 to 8 weeks to go live, outreach needs a warm-up month, and movement in AI answers takes months, not weeks. Test it with a week away, phone off, then a fortnight. Book the month when the fortnight passes without a dip. Sound familiar? ### Founders who said this also said these. We've been at £5m for three years. A founder-led B2B firm usually stalls when the things that built it, referrals and the founder's own selling, reach their ceiling. The network has been worked, the founder's diary is full, and each year's new business only replaces the clients that left. Growth tends to restart when strangers can find the firm, trust it before the first call, and are approached on purpose. Growth Team programme I am still our best salesperson. A founder usually outsells everyone because they carry authority, know every past project and can change the offer in the room. None of that transfers by hiring a salesperson. What does transfer is the work before the meeting: finding the right firms, researching them and opening the conversation. Hand that over first and keep the closing meetings. Pipeline That Fills All 20 things founders tell us From Insights ### Three things worth reading first. Content that gets quoted #### Can we outsource our content without losing our voice? 22 April 2026 Content that gets quoted #### Can ChatGPT tell real expertise from keyword-stuffed content? 31 March 2026 ### Bring this one to the call. Thirty minutes with a founder. We look at your site, two competitors and what Google and ChatGPT say about all three beforehand. You leave with three fixes in writing. Book a Growth Review --- Title: Meetings with people who cannot say yes | Fortitude Media URL: https://fortitudemedia.ai/sound-familiar/meetings-with-people-who-cannot-say-yes Summary: Booking sales meetings with people who can't make the decision? It usually comes down to what you count. How to check, and a definition that fixes it. Things founders tell us · No. 14 of 20 · Pipeline and outreach ## “We book meetings with people who can't say yes.” The calendar looks healthy. Then you join the call and it's an assistant gathering options, or a manager who needs to ask someone who has never heard of you. Forty minutes, a polite follow-up, silence. Book a Growth Review ### Why do our sales meetings keep being with people who can't make the decision? Meetings land with the wrong people when whoever books them is measured on meetings booked, not on who turns up. Junior staff accept calls more readily than owners, so any volume target drifts towards them. Fix the definition first: a meeting counts only if the person owns or materially influences the decision, works at a firm that fits, and knows what the call is about. Causes ### How the wrong people end up in your diary. #### You are paying for bookings An appointment setter or sales development rep on a meetings-booked target will find the people most willing to take a meeting. Those are rarely the people with the budget. You get what you count. #### The list was built on job titles A title tells you little. In one firm the operations director signs everything, and in the next nothing moves without the owner. Finding out who decides means researching the firm and then the person, and most lists skip both. #### The invitation was vague Messages offering a quick chat about synergies get accepted by people who don't know what they have agreed to. The meeting starts as a surprise and ends as a courtesy. The cost ### What it is costing you. Do the sum with your own numbers Count last quarter's first meetings and mark those where the person could never have said yes. For each, add the hours: preparation, the call, the follow-up, the proposal you probably wrote anyway. For example, 15 wasted meetings at four hours each is 60 hours of senior time. Put your day rate on that, then add whatever you paid to have them booked. Do it yourself ### Check it yourself in ten minutes. The evidence is already in your CRM or calendar. - 1 Pull the last 20 first meetings. Next to each, write the attendee's job title and whether they could have signed. - 2 Mark which ones led to a second meeting with someone more senior. That is your real qualified rate. - 3 Reread the message that booked each wasted meeting. Did it say plainly what you sell? - 4 Look at how whoever books them is paid. If the words held and qualified appear nowhere, you have found the cause. If you would rather we did it ### What we would do in the first 90 days. We start with the definition, because everything else follows from it. - Week 1 Agree in writing what a qualified meeting is: held, with someone who owns or materially influences the decision, at a firm matching agreed criteria, who knowingly agreed to talk about your offer. - Weeks 2 to 4 Two-pass research. First the firm: does it fit. Then the person: who decides, and what gives them a reason to talk this month. Month one is warm-up. - Months 2 and 3 80 to 120 approaches a month, each written and approved by a person. The fee is per meeting held. The monthly report counts meetings held and pipeline accepted by your sales team. Pipeline That Fills is £1,500 to set up, £1,000 to £1,500 a month to manage, and £500 to £1,000 per meeting held. No meeting, no meeting fee. Prices exclude VAT. How Pipeline That Fills works Questions founders ask ### The short answers. Who decides whether a meeting was qualified? The definition does. It is agreed before any message goes out, along with the criteria a firm has to match, so there is little to argue about afterwards. The person either owns or materially influences the decision or they don't. Is a meeting with an influencer ever worth taking? Yes, if they materially influence the decision and can get you to the person who signs. A finance director who shapes the shortlist is worth an hour. Someone collecting brochures for a boss who hasn't asked for them is not. Can we fix this with the sales rep we already have? Often, yes. Measure them on qualified meetings held, not meetings booked, and cut their volume so they have time to research who decides. Expect fewer meetings for a month or two, and judge the change on second meetings and proposals. Sound familiar? ### Founders who said this also said these. I delete twenty LinkedIn pitches a week. I'm not going to send them. Yes, if the volume is low enough for each message to be researched and written by a person. Outreach becomes spam when the same words go to hundreds of people. A message that names something true about the recipient's firm, gives a reason to talk now, and could not have been sent to anyone else reads as a business letter. Pipeline That Fills The last agency sent reports, not customers. A marketing agency is working if you can name customers, or at least serious sales conversations, that exist because of what it did. Traffic, impressions, rankings and followers measure activity. They only matter when enquiries from the right buyers move with them. If several months in nobody can trace one qualified conversation to the work, the reports describe effort, not results. Growth Team programme All 20 things founders tell us From Insights ### Three things worth reading first. Costs, returns and the business case #### What should our monthly report on AI search show the board? 31 March 2026 Costs, returns and the business case #### How can we tell which of our new leads came from ChatGPT? 31 March 2026 ### Bring this one to the call. Thirty minutes with a founder. We look at your site, two competitors and what Google and ChatGPT say about all three beforehand. You leave with three fixes in writing. Book a Growth Review --- Title: Why does our website generate no leads? | Fortitude Media URL: https://fortitudemedia.ai/sound-familiar/nobody-finds-us-through-our-website Summary: Every client came by referral and none through the website. Why B2B sites get no enquiries from strangers, what that costs, and how to check yours. Things founders tell us · No. 4 of 20 · The website ## “Nobody has ever said 'I found you on your website'.” Ask any client how they found you and the answer is a name: a former colleague, an accountant, an old client. The website has never once been the answer. It is a brochure that happens to be online. Book a Growth Review ### Why does nobody find our business through our website? Strangers find a website only when its pages match what they search for or ask an AI assistant. Most B2B sites are organised around the firm: about us, our services, our team. Buyers search for their problem, a price, a comparison or a place. If no page answers those, the only visitors are people who already know the company name. Causes ### Why strangers never arrive. #### Your only visitors already know you Look at the searches that bring traffic and most will be your company name. Those people were referred. The site confirms a decision somebody else started. It finds nobody. #### The pages are about you Nobody types 'our services' into Google. They type what is going wrong, what it costs to fix, and who does it near them. Five pages about the firm and none about those questions gives a search engine nothing to match. #### Visitors are given nothing to do A contact form is a big ask for someone three weeks into their research. With no smaller step on offer, they leave. You never learn they were there. The cost ### What it is costing you. Do the sum with your own numbers Referrals arrive when they arrive, and you cannot order more. So the cost is growth you cannot plan. Divide the revenue you want to add next year by your average client value: that is the clients you need. Subtract what referrals delivered last year. For example, £600,000 of growth at £15,000 a client is 40 clients. If referrals bring 25, the other 15 have to come from somewhere. Do it yourself ### Check it yourself in ten minutes. Ignore total visitor numbers. You want evidence that a stranger could find you. - 1 In Google Search Console, list the top 20 searches that led to your site. Cross out any containing your company name. - 2 Write the three questions a buyer asks before they know any supplier's name. Put each to Google and to ChatGPT. - 3 Go through a year of enquiry form submissions. Mark each as someone you already knew or a stranger. - 4 Count the pages that answer a buyer's question against the pages that describe your firm. If you would rather we did it ### What we would do in the first 90 days. The aim is a site that strangers with the right problem can find. - Week 1 We agree the 40 to 150 questions your buyers ask and begin putting them to ChatGPT, Claude, Gemini and Google each week, recording who is named. You see where you are absent. - Weeks 2 to 8 We build a page for each question that matters: question as heading, answer first, facts marked up, with proof and one next step on every page. - Weeks 9 to 13 Weekly dated improvements from search data and what sales hears. New questions get new pages. This is Website That Sells: build £5,000 to £10,000, run from £999 a month, live in 4 to 8 weeks, all excluding VAT. How Website That Sells works Questions founders ask ### The short answers. We are a referral business. Does the website matter? It matters to the referrals too. Someone recommended to you will look at the site before replying, so it is already part of every sale. Whether to invest in being found by strangers depends on whether referrals alone will take you to the size you want. What does it cost? Website That Sells is £5,000 to £10,000 to build and from £999 a month to run, plus VAT. The first step costs nothing: a 30-minute Growth Review with a founder, then three fixes in writing. For comparison, the 2026 UK median salary for a content and search manager is £42,055. What if our buyers do not search online for what we sell? Some do not, and we would rather say so in the Growth Review than sell you a website. Where every deal comes through a tender list or a personal introduction, researched outreach is usually the better spend. Search Console will show whether anyone is looking. Sound familiar? ### Founders who said this also said these. ChatGPT has never heard of us. AI assistants know a business when they can read clear facts about it on its own website and find the same facts repeated elsewhere: directories, press, reviews, LinkedIn. When a site is thin, hard for machines to read, or contradicted by old listings, the assistant has nothing reliable to repeat, so it says nothing or guesses. Age is no protection. Website That Sells We've outgrown word of mouth. Word of mouth stops being enough when the revenue you need grows faster than your network. Referrals depend on how many people know your work, and that number rises slowly. Keep them, because they convert best, and add one channel you control: researched approaches to firms that fit, backed by a website that holds up when they look you up. Pipeline That Fills All 20 things founders tell us From Insights ### Three things worth reading first. Content that gets quoted #### Should our content answer the questions our customers actually ask? 31 March 2026 How buying is changing #### What should my website do now that visitors arrive pre-sold by AI? 31 March 2026 Websites that bring work in #### How do we measure the return we get from our website? 31 March 2026 ### Bring this one to the call. Thirty minutes with a founder. We look at your site, two competitors and what Google and ChatGPT say about all three beforehand. You leave with three fixes in writing. Book a Growth Review --- Title: Expert content when experts have no time | Fortitude Media URL: https://fortitudemedia.ai/sound-familiar/our-best-thinking-is-in-our-heads Summary: Your firm's best thinking is in people's heads and nobody has time to write. How to turn expert knowledge into content without asking experts to write. Things founders tell us · No. 8 of 20 · Reputation and being found ## “Our best thinking is in our heads.” The sharpest things your firm knows get said out loud, to one client, in one meeting, and are never written down. Ask your senior people for 800 words and you get a promise, then silence. Book a Growth Review ### How do we turn our experts' knowledge into content when they have no time to write? Stop asking experts to write. Interview them. Thirty minutes of a specialist answering real client questions, recorded and transcribed, holds enough material for several articles. A writer who understands the subject shapes it, and the expert corrects the draft in ten minutes. The expert's job is to know things. Writing was never the scarce skill. Their knowledge is. Causes ### Why it never gets written. #### Experts talk well and write badly, or not at all Your best engineer, partner or consultant will explain a problem brilliantly across a table. Give them a blank page and they freeze, or produce something that reads like a contract. It is a different skill, and at their rate an expensive one to practise. #### Billable work always wins the diary An article has no deadline and no client chasing it, so it moves to next week, every week. This is not laziness. It is a firm correctly putting revenue first, with a side effect nobody chose. #### Generic content fills the gap When the experts do not write, a junior or a copywriter does, working from Google. The result says what every competitor's blog says. Buyers skim it, and AI assistants have no reason to cite it. The cost ### What it is costing you. Do the sum with your own numbers Count the sales calls last year where a senior person spent the first half hour explaining something they have explained many times before. For example, 60 calls at 30 minutes each is 30 hours of your most expensive people, reaching one prospect at a time. Written down once, the same explanation reaches every buyer who searches, including the ones who never ring you. Do it yourself ### Check it yourself in ten minutes. You need your sent folder, your phone and one colleague. - 1 Search your sent email for the longest replies you wrote to clients this year. Each is an article already written. - 2 Ask whoever takes sales calls for the five questions they hear weekly. Check whether your site answers any. - 3 Record yourself answering one of those questions for five minutes. Have it transcribed and see how much is usable. - 4 Read your last three blog posts. Could a competitor have published them unchanged? If you would rather we did it ### What we would do in the first 90 days. One interview a month is the whole ask of your experts. We do the rest. - Week 1 We agree the question list: what buyers ask before they hire, taken from your sales conversations and search data. Then we book the first expert. - Weeks 2 to 4 The interview becomes three expert pieces in that person's voice. They read and correct each one before it goes out under their name. - Weeks 5 to 12 Two more interviews, six more pieces, a different expert each month if you have them. The strongest piece is pitched to the trade press. Reputation That Wins turns a monthly 30-minute interview into three expert pieces a month in your voice, from £1,999 a month excluding VAT. How Reputation That Wins works Questions founders ask ### The short answers. Will it sound like us, or like an agency? It should sound like the person interviewed, because the words start as theirs. We keep their phrasing, their examples and their opinions, including the unfashionable ones. If your expert would never put it that way, the draft is wrong and we rewrite it. How much of our experts' time does it take? Thirty minutes for the interview, plus a short read of each draft. Call it an hour a month per expert. We prepare the questions beforehand, so nobody has to work out what to say, and there is no homework. People who hate writing rarely mind talking. Can't we use ChatGPT to write it? You can, and you will get fluent articles containing nothing your competitors could not also generate. The value is in what only your people know: the job that went wrong, the number that surprised you. A tool can tidy that. It cannot supply it. Sound familiar? ### Founders who said this also said these. I should post on LinkedIn. I never do. For a founder-led B2B firm, usually yes. Buyers look up the founder before a call, after a cold email and during a tender, and an empty profile tells them nothing. What works is specific opinions and lessons from real work, not motivational content. Keeping it up needs a system: one recorded conversation a month yields several weeks of posts. Reputation That Wins I am still our best salesperson. A founder usually outsells everyone because they carry authority, know every past project and can change the offer in the room. None of that transfers by hiring a salesperson. What does transfer is the work before the meeting: finding the right firms, researching them and opening the conversation. Hand that over first and keep the closing meetings. Pipeline That Fills All 20 things founders tell us From Insights ### Three things worth reading first. Content that gets quoted #### Can we outsource our content without losing our voice? 22 April 2026 Content that gets quoted #### What makes ChatGPT treat our content as expert rather than marketing? 22 April 2026 Content that gets quoted #### Should our content answer the questions our customers actually ask? 31 March 2026 ### Bring this one to the call. Thirty minutes with a founder. We look at your site, two competitors and what Google and ChatGPT say about all three beforehand. You leave with three fixes in writing. Book a Growth Review --- Title: Why does our website undersell us? | Fortitude Media URL: https://fortitudemedia.ai/sound-familiar/our-website-undersells-us Summary: Your firm is better than its website makes it look. Why that happens to founder-led B2B firms, what it costs, and how to check in ten minutes. Things founders tell us · No. 1 of 20 · The website ## “Our website undersells us.” You win work when you are in the room. Then you look at your own website and see a smaller, vaguer firm than the one you run. You have stopped sending people the link. Book a Growth Review ### Why does my website make my business look smaller than it is? A website undersells a firm when it describes the business in general terms and leaves out the specifics buyers judge it by: who it works for, what it costs, how long it takes, and proof it has been done before. Most were written once, by an outsider, and never updated as the firm grew. The fix is specifics, not a redesign. Causes ### What is really going on. #### It was written by someone who has never sold for you The copy came from a freelance writer after a one-hour briefing. The lines that make a buyer lean forward in a pitch never reached the page. #### The firm moved on and the site did not You have bigger clients, a sharper offer and higher prices than at launch. The site still describes the firm you were then, and closing the gap is nobody's job. #### It hides the facts buyers want Buyers research before they call. They want a price range, a timeline, and evidence you have done this for a firm like theirs. If the site is silent, they assume the answers are bad. The cost ### What it is costing you. Do the sum with your own numbers Count the introductions you received last year. Each of those people looked at your website before replying. Now estimate how many went quiet after looking. For example, if five of 40 introductions went quiet and your average client is worth £15,000, that is £75,000 of work you had already been recommended for. Use your own numbers. These were the warmest buyers you will ever get. Do it yourself ### Check it yourself in ten minutes. Do this on your phone, because your buyer will. - 1 Read your homepage heading. If a competitor could use it word for word, it says nothing about you. - 2 Write down the five questions buyers ask on every first call. Count how many your site answers. - 3 Look for a price, a timeline and a named person. Note which are missing. - 4 Think of the client result you always mention in pitches. If it is more than two taps from the homepage, it is too far. If you would rather we did it ### What we would do in the first 90 days. We start with what your buyers ask, not with how the site looks. - Weeks 1 to 2 We interview you and whoever handles sales, and collect the questions buyers ask before they sign. - Weeks 3 to 8 We write and build the pages: question as heading, answer first, with prices, timelines, locations and credentials marked up so Google and AI assistants can read them. Proof and one next step on every page. - Weeks 9 to 13 The site is live and we improve it weekly from search data and what sales hears. Every change is dated and logged. This is Website That Sells: build £5,000 to £10,000, run from £999 a month, live in 4 to 8 weeks, all excluding VAT. How Website That Sells works Proof it works · arum A 28-year-old firm's new website. Built as if it were our own. 310 pages rebuilt across three regions, in weeks Read the story Questions founders ask ### The short answers. Do we need a full rebuild? Not always. It depends on whether your platform lets someone change headings, structure and markup without a fight. We look at your site before the free 30-minute Growth Review, so ask on the call. You get three fixes in writing afterwards, and you can stop there. Can't we rewrite it ourselves? Yes, and the check above is where to start. Ability is rarely the difficulty. The people who know the answers are busy with clients, so the rewrite sits at the bottom of a list for a year. If you can protect two days a month, do it in-house. What if we spend the money and nothing improves? It can happen, and nobody can promise you enquiries. What we can do is make the work visible: at least one dated, logged improvement a week, and one page each month showing meetings held, pipeline accepted by sales and the buyer questions where an AI assistant names you. Sound familiar? ### Founders who said this also said these. We rebuilt it two years ago and nothing changed. A rebuild changes how a site looks. Enquiries depend on what it says and who sees it. If the new site carried over the old words, answered no more buyer questions than before, and was left untouched after launch, there was no reason for the result to change. Websites that produce enquiries are a monthly job, not a project with an end date. Website That Sells We're the best-kept secret in our industry. A reputation that lives only in clients' heads does not travel. To be known beyond your network, the evidence has to exist in public: expert articles under named people, coverage in the trade press your buyers read, reviews, accurate listings and a visible founder. Buyers research before they call, and can only shortlist firms they can find. Reputation That Wins All 20 things founders tell us From Insights ### Three things worth reading first. Websites that bring work in #### How do I know whether our website is actually working? 31 March 2026 Websites that bring work in #### What makes a buyer trust our website enough to get in touch? 31 March 2026 Content that gets quoted #### Should our content answer the questions our customers actually ask? 31 March 2026 ### Bring this one to the call. Thirty minutes with a founder. We look at your site, two competitors and what Google and ChatGPT say about all three beforehand. You leave with three fixes in writing. Book a Growth Review --- Title: What should my SEO retainer deliver? | Fortitude Media URL: https://fortitudemedia.ai/sound-familiar/paying-for-seo-dont-know-what-i-get Summary: Paying an SEO agency every month and unsure what it does? What a retainer should produce, how to check yours in ten minutes, and when to stop. Things founders tell us · No. 5 of 20 · The website ## “I'm paying for SEO. I don't know what I get.” The direct debit goes out every month. A report arrives with green arrows and a rising graph. You could not name one thing that changed on your website last quarter. Book a Growth Review ### How do I know if my SEO agency is doing anything? Ask for a dated list of what was changed on the website last month, and why. Real search work leaves evidence: pages rewritten, faults fixed, coverage earned on other sites. An agency that can show only rankings and traffic charts is reporting on the weather. The test that matters is whether enquiries from people who did not already know the firm are rising. Causes ### Why you cannot see what you are buying. #### The report measures what is easy to push up Rankings for phrases nobody buys from, impressions, a visibility score invented by the agency's own software. All can rise for a year while enquiries stay flat. #### You are paying for a retainer, not for work Many SEO contracts name a monthly fee and no deliverables. The hours go to whichever client complains loudest. If you never ask what was done, the true answer some months is not much. #### The target moved and the contract did not Your retainer was probably scoped for a list of ten links on Google. Buyers now also ask an AI assistant and get two or three names. Work aimed only at the old results page misses part of how you are found. The cost ### What it is costing you. Do the sum with your own numbers Multiply the monthly fee by the months you have paid it. For example, £1,500 a month for three years is £54,000. Now count the clients you can trace to a search by someone who had never heard of you. Divide the first number by the second for your cost per client. If you cannot count them at all, that is the finding. Do it yourself ### Check it yourself in ten minutes. You need last month's report, your contract and a Google Search Console login. - 1 Email the agency for a dated list of every change made to your site in the past three months. - 2 Read the contract for deliverables. If it lists only a fee and a notice period, you have your explanation. - 3 Open Search Console yourself, not the agency's dashboard. Compare clicks from searches without your company name against the same months last year. - 4 Pick the three keywords the report is proudest of. Ask your sales lead whether a buyer has ever used those words. If you would rather we did it ### What we would do in the first 90 days. Whoever does the work, you should be able to see it. - Week 1 We agree the 40 to 150 questions your buyers ask and record, every week, who ChatGPT, Claude, Gemini and Google name for each. That replaces the keyword list. - Weeks 2 to 8 Pages are written for those questions, answer first, with prices, timelines, locations and credentials marked up so machines can read them. - Weeks 9 to 13 At least one improvement a week, dated and logged with the reason. On the first Monday of each month, a one-page report. This is Website That Sells: build £5,000 to £10,000, run from £999 a month, live in 4 to 8 weeks, all excluding VAT. How Website That Sells works Questions founders ask ### The short answers. Should I sack my SEO agency? Not on the strength of this page. Run the check first. Some agencies do solid work and report it badly. A dated change list will show that. If the list is thin or never arrives, give notice. A month's pause while you decide is unlikely to undo work that was real. Is SEO dead now that people ask AI assistants? No. Google still sends buyers to websites, and AI assistants read many of the same pages. The prize has changed: being named in the answer, rather than sitting third in a list. The work overlaps: clear pages that answer real questions. What would it cost to do this in-house? The 2026 UK median for a content and search manager is £42,055, before employer National Insurance, pension, software and recruiter fees. Website That Sells runs from £999 a month after a build of £5,000 to £10,000, plus VAT. Either route can work. Sound familiar? ### Founders who said this also said these. We rebuilt it two years ago and nothing changed. A rebuild changes how a site looks. Enquiries depend on what it says and who sees it. If the new site carried over the old words, answered no more buyer questions than before, and was left untouched after launch, there was no reason for the result to change. Websites that produce enquiries are a monthly job, not a project with an end date. Website That Sells I can't tell what marketing made us. In B2B, with long sales cycles and several people in each purchase, precise attribution is not available and chasing it wastes money. A simpler discipline works. Record the source of every first conversation, ask each new buyer how they found you, and review the answers monthly against spend. That shows which activity starts conversations and which starts none. Growth Team programme All 20 things founders tell us From Insights ### Three things worth reading first. Getting named by AI assistants #### Do we still need SEO now that buyers ask ChatGPT instead of Google? 22 April 2026 How buying is changing #### We rank first on Google. Why does ChatGPT not mention us? 31 March 2026 Costs, returns and the business case #### How do we choose a firm to help us get recommended by ChatGPT? 31 March 2026 ### Bring this one to the call. Thirty minutes with a founder. We look at your site, two competitors and what Google and ChatGPT say about all three beforehand. You leave with three fixes in writing. Book a Growth Review --- Title: Why is our sales pipeline feast or famine? | Fortitude Media URL: https://fortitudemedia.ai/sound-familiar/pipeline-is-feast-or-famine Summary: Pipeline is feast or famine when selling stops every time delivery gets busy. Why it happens, what it costs, and how to put a floor under it. Things founders tell us · No. 11 of 20 · Pipeline and outreach ## “Pipeline is feast or famine.” You have a strong quarter, everyone is flat out delivering, and nobody sells. Then the work finishes and you look up to an empty diary. Book a Growth Review ### Why is our sales pipeline feast or famine, and how do we even it out? Pipeline swings between feast and famine when selling only happens in the gaps between delivery. The people who sell are the people who deliver, so prospecting stops whenever the firm is busy, and the hole appears one sales cycle later. The fix is a small, fixed amount of outbound activity every week that carries on however busy delivery gets. Causes ### What is really going on. #### Selling is what you do when you are quiet Prospecting starts when the diary empties and stops when work lands. With a three-month sales cycle, the effort you make in a panic pays off just as you become too busy to repeat it. #### Referrals keep their own timetable Referrals are the best leads you will ever get, and you control none of the timing. Two in one month and none for the next four is normal. That is weather, not a system. #### Nobody owns the first conversation Ask who is responsible for starting ten new conversations this month. In most founder-led firms the answer is you, when you have time. A job with no owner and no weekly number does not get done. The cost ### What it is costing you. Do the sum with your own numbers Look at new business won in each of your last eight quarters. Take the best and the worst and work out the difference. For example, if a good quarter brings in £300,000 and a bad one £90,000, the gap is £210,000. Then add what the famine does: people you carry with nothing to bill, and the discounts you give when you need the work. Do it yourself ### Check it yourself in ten minutes. You need your order book and your calendar, nothing else. - 1 List new business won by month for the last 24 months. Mark the months where nothing was signed. - 2 Next to each month, write how many first conversations with new prospects took place. If you cannot find the number, that is a finding. - 3 Work out your sales cycle: the typical gap between first conversation and signature. Shift the conversations row forward by that much and compare it with the wins. - 4 Count the first conversations booked for the next 30 days. That is your revenue one sales cycle from now. If you would rather we did it ### What we would do in the first 90 days. The aim is a floor under the pipeline: a fixed number of new conversations every month, whatever delivery is doing. - Week 1 A free 30-minute Growth Review with a founder. Then we agree, in writing, what a good-fit firm looks like and what counts as a qualified meeting. - Weeks 2 to 4 Month one is warm-up. We build the list and research each firm, then each person. A person writes and approves every message. - Months 2 and 3 80 to 120 researched approaches a month, every month, including the months you are flat out. A one-page report on the first Monday shows meetings held and pipeline accepted by sales. This is Pipeline That Fills: set-up £1,500, management £1,000 to £1,500 a month, and £500 to £1,000 per meeting held, all excluding VAT. No meeting, no meeting fee. How Pipeline That Fills works Proof it works · Helm Club The club had the members. It needed the calls. 1 in 5 new members this year came through an AI recommendation Read the story Questions founders ask ### The short answers. Can't we just block out a day a week for selling? You can, and some firms manage it. It tends to last until the first deadline week, because client work always shouts louder than prospecting. If you try it, track one number weekly: first conversations started. When that drops for a fortnight, the discipline has gone. What does it cost? Set-up is £1,500. Management is £1,000 to £1,500 a month, and you pay £500 to £1,000 for each qualified meeting that is held. No meeting, no meeting fee. All prices exclude VAT. The first step, a 30-minute Growth Review with a founder, is free. What if it doesn't work? Month one is warm-up, so judge it on months two and three. If fewer than six qualified meetings are held in those two months combined, month four's management fee is waived. You can stop after any step, and the list is yours to keep. Sound familiar? ### Founders who said this also said these. We've outgrown word of mouth. Word of mouth stops being enough when the revenue you need grows faster than your network. Referrals depend on how many people know your work, and that number rises slowly. Keep them, because they convert best, and add one channel you control: researched approaches to firms that fit, backed by a website that holds up when they look you up. Pipeline That Fills We've been at £5m for three years. A founder-led B2B firm usually stalls when the things that built it, referrals and the founder's own selling, reach their ceiling. The network has been worked, the founder's diary is full, and each year's new business only replaces the clients that left. Growth tends to restart when strangers can find the firm, trust it before the first call, and are approached on purpose. Growth Team programme All 20 things founders tell us From Insights ### Three things worth reading first. Costs, returns and the business case #### What is it costing us not to be recommended by ChatGPT? 31 March 2026 Costs, returns and the business case #### How can we tell which of our new leads came from ChatGPT? 31 March 2026 Websites that bring work in #### How do we measure the return we get from our website? 31 March 2026 ### Bring this one to the call. Thirty minutes with a founder. We look at your site, two competitors and what Google and ChatGPT say about all three beforehand. You leave with three fixes in writing. Book a Growth Review --- Title: Why has my business stopped growing? | Fortitude Media URL: https://fortitudemedia.ai/sound-familiar/revenue-has-been-flat-for-years Summary: Revenue flat for years at the same figure. Why founder-led B2B firms plateau, how to check the cause in ten minutes, and what restarts growth. Things founders tell us · No. 17 of 20 · Running growth at all ## “We've been at £5m for three years.” The top line has barely moved for three year-ends running. The team is busy and clients are happy. But every year you write a plan that says six, and every year you land back at five. Book a Growth Review ### Why has my business stopped growing? A founder-led B2B firm usually stalls when the things that built it, referrals and the founder's own selling, reach their ceiling. The network has been worked, the founder's diary is full, and each year's new business only replaces the clients that left. Growth tends to restart when strangers can find the firm, trust it before the first call, and are approached on purpose. Causes ### Why firms stick at the same number. #### You are replacing, not adding You probably won a respectable amount of new work each year. It roughly matched what finished, shrank or left. A firm can sell hard and stand still for a long time that way. #### Referrals do not scale with headcount Your network produces about the same number of introductions whether you employ twenty people or forty. It got you to £5m, and has no particular reason to get you to £8m. #### Nothing sells while you are delivering When you are busy, nobody is starting conversations. A buyer who has never heard of you asks an AI assistant who to use and gets two or three names. If yours is not among them, you never learn the enquiry existed. The cost ### What it is costing you. Do the sum with your own numbers Flat revenue is not flat profit, because salaries and rent went up anyway. Now compound the rate your plan assumed over the years you have stood still. For example, a £5m firm growing at 15% a year would reach about £7.6m after three years. Add the shortfall across all three years and it comes to nearly £5m, a whole year's turnover that did not happen. Do it yourself ### Check it yourself in ten minutes. All of this is in your accounts system and your memory. - 1 For each of the last three years, list clients won and lost with their annual value. See whether wins simply cover losses. - 2 For every client won in the last twelve months, write down how they first heard of you. - 3 Count how many came from someone neither you nor your team already knew. - 4 Ask ChatGPT the question your buyer would ask about your service and region. Note who is named. If you would rather we did it ### What we would do in the first 90 days. First find the tight constraint. Fixing the wrong one is how another three years go by. - Week 1 Growth Review with a founder. We check your site, two competitors, and what Google and ChatGPT say about all three, then send three fixes in writing. - Weeks 2 to 6 Site pages are rewritten around the questions your buyers ask, answer first. The outreach list is researched in two passes, the firm and then the person. - Days 45 to 90 After a warm-up month, outreach runs at 80 to 120 researched approaches a month, each written and approved by a person. Three expert pieces a month go out in your voice. The Growth Team programme covers all three services from £2,500 a month (most clients pay £5,000 to £8,000), all prices exclude VAT, and you can stop after any step. How Growth Team programme works Questions founders ask ### The short answers. How long before the revenue line moves? Longer than anyone selling to you would like. A rebuilt site goes live in 4 to 8 weeks. Outreach spends month one warming up, and meetings are judged across months two and three. Movement in AI answers takes months, not weeks. Then add your own sales cycle on top. Would hiring my own team be better? It can be, at a larger size. On 2026 UK medians, a marketing manager, a content and search manager, a marketing executive and a sales development rep cost £203,991 a year with employer NI, pension and software. A £6,500-a-month programme is £78,000 a year. What if marketing is not the problem? Then do not buy marketing. If clients leave as fast as they arrive, or you cannot deliver more work than you have, new enquiries will not help. Fix retention or capacity first. The free Growth Review is a reasonable place to test which problem you have. Sound familiar? ### Founders who said this also said these. If I took a month off, sales would stop. Separate the parts of selling that need the founder from the parts the founder merely happens to do. Finding prospects, first approaches, follow-up and proving the firm's credibility can all run without you once written down and owned by someone. Closing the largest deals may stay with you for years. The aim is conversations that keep arriving while you are away. Growth Team programme We've outgrown word of mouth. Word of mouth stops being enough when the revenue you need grows faster than your network. Referrals depend on how many people know your work, and that number rises slowly. Keep them, because they convert best, and add one channel you control: researched approaches to firms that fit, backed by a website that holds up when they look you up. Pipeline That Fills All 20 things founders tell us From Insights ### Three things worth reading first. Costs, returns and the business case #### What is it costing us not to be recommended by ChatGPT? 31 March 2026 How buying is changing #### Can we wait a year before doing anything about AI search? 22 April 2026 Costs, returns and the business case #### How do I get my board to approve budget for AI search? 22 April 2026 ### Bring this one to the call. Thirty minutes with a founder. We look at your site, two competitors and what Google and ChatGPT say about all three beforehand. You leave with three fixes in writing. Book a Growth Review --- Title: When word of mouth is not enough to grow | Fortitude Media URL: https://fortitudemedia.ai/sound-familiar/we-have-outgrown-word-of-mouth Summary: Outgrown word of mouth? Referrals don't grow with your revenue target. How to work out the gap and add a channel you control without losing referrals. Things founders tell us · No. 13 of 20 · Pipeline and outreach ## “We've outgrown word of mouth.” Referrals built the firm, and they still close faster than anything else. But the number you need each year has gone up and the number that arrive has not. Book a Growth Review ### What do you do when word of mouth isn't enough to grow the business any more? Word of mouth stops being enough when the revenue you need grows faster than your network. Referrals depend on how many people know your work, and that number rises slowly. Keep them, because they convert best, and add one channel you control: researched approaches to firms that fit, backed by a website that holds up when they look you up. Causes ### Why referrals run out. #### Your target went up and your network didn't At £1m, five referrals was a good year. At £5m you need several times that just to replace the clients who leave, and the same few dozen people are doing the recommending. #### Referrals bring more of what you already have People recommend you for what they watched you do. That keeps you in the same sector, with the same size of client, even when you are trying to move up. #### Even referred buyers check first A recommendation used to be enough. Now the person checks your website, your LinkedIn and what an AI assistant says about you before replying. If those fall short, you never hear about the referral you lost. The cost ### What it is costing you. Do the sum with your own numbers Write down next year's revenue target and subtract what existing clients will bring. Divide the gap by your average new client value. For example, a £1m gap at £50,000 a client is 20 new clients. Now count how many referrals became clients in each of the last three years. If the answer is about eight, hoping is the current plan for the other 12. Do it yourself ### Check it yourself in ten minutes. This one is a counting exercise. - 1 List every new client from the last three years with the source next to each: referral, existing contact, inbound, outbound, other. - 2 For the referrals, write down who made the introduction. Count the different names. A short list is a fragile channel. - 3 Do the sum above for next year and compare the clients you need with your yearly referral average. - 4 Search your firm's name plus the word reviews in Google, then ask ChatGPT about your firm. Read the results as if you had never heard of yourself. If you would rather we did it ### What we would do in the first 90 days. Referrals stay. We add a second channel beside them that you can turn up or down. - Week 1 A Growth Review with a founder. Beforehand we check your site, two competitors, and what Google and ChatGPT say about all three. Afterwards: three fixes in writing. - Weeks 2 to 4 Your best referred clients become the pattern: sector, size, what was happening when they bought. That sets the agreed criteria for the list. Month one is warm-up. - Months 2 and 3 80 to 120 researched approaches a month to firms that look like your best clients and have never heard of you. We stop after four unanswered emails. The list and conversations are yours to keep. That is Pipeline That Fills: £1,500 set-up, £1,000 to £1,500 a month management, £500 to £1,000 per meeting held, no meeting, no meeting fee, all excluding VAT. How Pipeline That Fills works Proof it works · Helm Club The club had the members. It needed the calls. 1 in 5 new members this year came through an AI recommendation Read the story Questions founders ask ### The short answers. Can't I just ask clients for more referrals? Yes, and you should. Ask when a project has just landed well, and ask for a named introduction, not a general mention. It will not change the limit, which is that your clients know a fixed number of people who might need you this year. Isn't a cold approach much weaker than a referral? It is, and we would not pretend otherwise. The point is a number you control. At 80 to 120 approaches a month, not 800, each one can be researched properly: the firm first, then the person, with a reason to talk this month. Referrals keep arriving alongside. How long before it produces anything? Month one is warm-up: list, research, first messages. Judge it on months two and three. If fewer than six qualified meetings are held in those two months combined, month four's management fee is waived. After that, your own sales cycle sets the pace. Sound familiar? ### Founders who said this also said these. I am still our best salesperson. A founder usually outsells everyone because they carry authority, know every past project and can change the offer in the room. None of that transfers by hiring a salesperson. What does transfer is the work before the meeting: finding the right firms, researching them and opening the conversation. Hand that over first and keep the closing meetings. Pipeline That Fills We're the best-kept secret in our industry. A reputation that lives only in clients' heads does not travel. To be known beyond your network, the evidence has to exist in public: expert articles under named people, coverage in the trade press your buyers read, reviews, accurate listings and a visible founder. Buyers research before they call, and can only shortlist firms they can find. Reputation That Wins All 20 things founders tell us From Insights ### Three things worth reading first. How buying is changing #### What should my website do now that visitors arrive pre-sold by AI? 31 March 2026 Costs, returns and the business case #### How do we find out whether ChatGPT is recommending our firm? 31 March 2026 Websites that bring work in #### How do I know whether our website is actually working? 31 March 2026 ### Bring this one to the call. Thirty minutes with a founder. We look at your site, two competitors and what Google and ChatGPT say about all three beforehand. You leave with three fixes in writing. Book a Growth Review --- Title: We rebuilt the website. Why no change? | Fortitude Media URL: https://fortitudemedia.ai/sound-familiar/we-rebuilt-the-website-and-nothing-changed Summary: A new website that produced no new enquiries. Why redesigns so often fail founder-led B2B firms, what yours cost you, and what to do instead. Things founders tell us · No. 3 of 20 · The website ## “We rebuilt it two years ago and nothing changed.” It took longer than planned, cost more than the quote, and everyone agreed it looked better. Then the phone rang exactly as often as before. Now someone says 'website' in a board meeting and your jaw tightens. Book a Growth Review ### Why did our new website make no difference to enquiries? A rebuild changes how a site looks. Enquiries depend on what it says and who sees it. If the new site carried over the old words, answered no more buyer questions than before, and was left untouched after launch, there was no reason for the result to change. Websites that produce enquiries are a monthly job, not a project with an end date. Causes ### Why the rebuild changed nothing. #### You bought a design project The brief was about look, feel and brand, and the agency delivered that. Nobody was asked what a buyer needs to read before picking up the phone. #### The old copy moved into the new templates Content is usually last on the project plan and first to slip. Someone pasted the old service pages into the new layout the week before launch. Same words, better font. #### Launch day was the last day anyone touched it The agency moved to its next client. A site that never changes cannot learn from what buyers search for. Two years on it is the same site, only older. The cost ### What it is costing you. Do the sum with your own numbers The rebuild fee and the internal hours are gone. The running cost matters more. Take the enquiries you expected the new site to bring each month, multiply by your close rate and your average client value, then by 24 months. For example, two enquiries a month, one in four closing, at £15,000 each, is £180,000 over two years. Use your own figures. Do it yourself ### Check it yourself in ten minutes. You need your analytics login and the old site, which web.archive.org probably still has. - 1 Put an old service page next to the same page today. Highlight every sentence that is substantially unchanged. - 2 Check when each main page was last edited. Count the pages untouched since launch. - 3 In Google Search Console, list the searches that bring people in. Remove your company name and see what is left. - 4 Ask whoever takes sales calls for the three questions they answered most this month. See whether the site answers any. If you would rather we did it ### What we would do in the first 90 days. The last project failed on words and neglect, so that is where the time goes. - Week 1 We read the site as a buyer would, compare it with two competitors, and collect the questions your sales conversations keep hearing. - Weeks 2 to 8 We write the missing pages and rewrite the weak ones: question as heading, answer first, facts marked up, proof and one next step on each. If parts of what you already paid for can be kept, we will say so. - Weeks 9 to 13 At least one dated, logged improvement every week, driven by search data and what sales hears. This is Website That Sells: build £5,000 to £10,000, run from £999 a month, live in 4 to 8 weeks, all excluding VAT. How Website That Sells works Proof it works · arum A 28-year-old firm's new website. Built as if it were our own. 310 pages rebuilt across three regions, in weeks Read the story Questions founders ask ### The short answers. Are you telling me to pay for another website? You would be paying for different work: finding buyer questions, writing the answers, marking up the facts, then the monthly running, which is where your last project stopped. We look at your site before the free Growth Review, so ask what could be kept before you spend anything. How long before we would see a difference? The build takes 4 to 8 weeks. After that it is gradual. Search data shows which pages are being found, and we adjust weekly. We do not promise enquiry numbers. You get one page on the first Monday of every month and can judge for yourself. Could our marketing person run the monthly improvements? Possibly, and it is cheaper if they can. They need access to search data, a regular line into what sales is hearing, and the authority to change pages without a committee. Take one away and the weekly rhythm tends to stop, which is how the site went stale. Sound familiar? ### Founders who said this also said these. Our website undersells us. A website undersells a firm when it describes the business in general terms and leaves out the specifics buyers judge it by: who it works for, what it costs, how long it takes, and proof it has been done before. Most were written once, by an outsider, and never updated as the firm grew. The fix is specifics, not a redesign. Website That Sells The last agency sent reports, not customers. A marketing agency is working if you can name customers, or at least serious sales conversations, that exist because of what it did. Traffic, impressions, rankings and followers measure activity. They only matter when enquiries from the right buyers move with them. If several months in nobody can trace one qualified conversation to the work, the reports describe effort, not results. Growth Team programme All 20 things founders tell us From Insights ### Three things worth reading first. Websites that bring work in #### Should we redesign our website or keep improving the one we have? 31 March 2026 Websites that bring work in #### What does it cost us to leave our website unchanged for years? 31 March 2026 Websites that bring work in #### What does leaving our website until later actually cost us? 31 March 2026 ### Bring this one to the call. Thirty minutes with a founder. We look at your site, two competitors and what Google and ChatGPT say about all three beforehand. You leave with three fixes in writing. Book a Growth Review --- ## Guides Title: How do we move beyond founder-led sales, and when should we hire our first salesperson? URL: https://fortitudemedia.ai/insights/beyond-founder-led-sales-first-sales-hire Author: Ross Williams Published: 2026-10-06 Last reviewed: 2026-10-06 Topic: Meetings and pipeline Short answer: Hire your first salesperson when you can teach someone how you win and there is a steady source of conversations for them to work on. Before the hire, write down your pitch, define the buyer, build a pipeline source that is not your personal network and use a CRM yourself. Then hire, handing over research, first approaches and follow-ups first, while you keep closing calls, key clients and the firm's public voice. You still win most of the new work. Clients want to meet you, you know every past project, and you can change the terms in the room. It works, until your diary becomes the limit on growth: deals wait for a gap in your week, and when delivery gets busy, new business stops. The obvious answer is to hire a salesperson. It is also where many founders lose a year. [Entrepreneur Plus, writing in August 2026](https://entrepreneurplus.co.uk/how-to-hire-a-salesperson-when-to-make-your-startups-first-sales-hire/), puts it bluntly: "Most founders hire their first salesperson too early, before they've proven a sales process exists to hand over." Even in a working sales team, [only about 66% of UK SDRs hit quota](https://salesrecruituk.com/salary-guide/sdr-bdr-salary-uk-2026/), according to Sales Recruit UK's June 2026 salary guide, citing RepVue data. This guide is for founders who still do most of the selling. It covers the signs you have reached the ceiling, what has to exist before a first sales hire can work, why first hires fail, the order that works, and what you should keep doing yourself. It is written from a founder's point of view: Ross Williams built and ran a subscription business, without outside investment, for 21 years. ## Key takeaways - Founder-led sales has hit its ceiling when deals wait for your diary, new work comes only from your network, and sales takes more than half your week. - First sales hires usually fail because they are hired to create demand that does not exist yet, with a pitch that lives only in the founder's head. - Four things should exist before the hire: a pitch someone else can repeat, a written definition of the buyer, a pipeline source that is not your personal network, and a CRM you use yourself. - The order that works is to write down how you sell, build a pipeline source, then hire, handing over the work before the meeting first. - A UK SaaS SDR earns about £33,000 base and £49,000 on target (Sales Recruit UK, June 2026), and new SDRs typically take three to six months to reach full productivity. - The founder should keep the meetings where authority matters, the conversations that teach you about the market, and the public voice of the firm. ## What are the signs that founder-led sales has hit its ceiling? The clearest sign is that growth now depends on your calendar. If you cannot take a month off without new business stopping, the firm has a sales process with one person in it. Look for these: - **Deals wait for you.** Proposals sit until you have time to write them, and first meetings are booked weeks out. - **Pipeline follows delivery.** When you are busy with clients, new enquiries dry up, and when projects finish there is a scramble. We describe this pattern in [pipeline is feast or famine](/sound-familiar/pipeline-is-feast-or-famine). - **Every client comes through your network.** Referrals and introductions still convert best, but there is no second source. - **Selling takes more than half your week.** Entrepreneur Plus suggests that when selling "consistently eats more than half your working week, and you're the bottleneck stopping deals from closing rather than the person driving them, that's your cue to bring someone in". - **Nobody else can explain the pitch.** Your best lines, client stories and answers to objections are in your head, not written down. If most of these are true, read [I am still our best salesperson](/sound-familiar/i-am-still-our-best-salesperson) and [if I stop selling, growth stops](/sound-familiar/if-i-stop-selling-growth-stops). Both describe the same bottleneck from different sides. ## Why do first sales hires fail? Most fail because they are hired to create demand that does not exist yet. The founder wins work through relationships and reputation; the new salesperson arrives with neither, a pitch nobody has written down, and a target based on the founder's results. [Techstars mentor Steve Walsh](https://www.techstars.com/blog/advice/founders-your-first-sales-hire-is-probably-a-mistake) wrote in March 2026 that first sales hires struggle "not because they are bad at their job, but because there is no clear playbook yet." [Bright Reach](https://brightreach.co.uk/blog/when-should-a-founder-hire-the-first-sales-reps-and-a-vp-of-sales/) advises against hiring sales reps if "all your customers are from personal networks" or if "you haven't identified repeatable lead generation mechanisms". Both describe the same gap. The founder's network is not transferable, so the hire needs a source of conversations that does not depend on it. The averages make this expensive. Sales Recruit UK puts quota attainment for UK SDRs at about 66%. [ORRJO](https://orrjo.com/blog-outsourced-sdr-vs-in-house) wrote in January 2026 that a new SDR "typically takes three to six months to reach full productivity" and that "the average SDR tenure is 14 months." On those averages, a hire who takes six months to ramp gives you well under a year at full speed. A hire with no pipeline source spends much of that time looking for one. There is a quieter cost too. Walsh warns that "the salesperson becomes the buffer between the customer and the founder." In a founder-led firm, sales conversations are where you learn what buyers want, which objections matter and what to build next. Hand them all over too early and you lose that. ## What has to exist before the first sales hire? Four things. Without them, the hire is being asked to invent a sales process as well as run one. | What must exist | How you know it exists | What happens without it | |---|---|---| | A pitch someone else can repeat | It is written down: the problem, the proof, the stories, the answers to objections | The hire improvises, results vary and you step back in | | A defined buyer | You can describe the sectors, company sizes, roles and trigger events that make a good client | The hire chases anyone who will take a call | | A pipeline source that is not your network | New conversations arrive each month without your introductions | The hire has nothing to work on, or burns your contacts | | A CRM that is used | Every open deal, next step and lost reason is in it, including yours | Nobody can see what is working, and handover is guesswork | Entrepreneur Plus suggests hiring "once you've personally closed 10 to 20 customers"; Bright Reach gives a range of 5 to 20, depending on the business model. The number matters less than being able to explain how each one was won. Walsh's test for a first sales leader works for any first sales hire: the right time "is not when you are tired of selling. It is when you have enough pattern recognition to teach someone else how to do it." ## In what order should we move beyond founder-led sales? Write down how you sell, build a pipeline source that does not depend on you, then hire. Hiring first means asking the new person to do the first two steps as well. ### Step 1: Write down how you sell Record your next ten first meetings, with permission, or write notes straight after each one. Capture the questions buyers ask, the stories that land, the objections and your answers, and the reasons you win and lose. Turn it into a short document a new person could read in an hour. ### Step 2: Define the buyer you want more of Look at your best ten clients. Write down what they have in common: sector, size, the role of the person who bought, and what was happening in their business when they did. This becomes the brief for every approach and every hire. ### Step 3: Build a pipeline source that is not your network Pick one channel you can run every month without your introductions. In our experience, researched outbound is the fastest to prove for most founder-led B2B firms: a person researches each firm and writes each approach, so you learn within weeks whether the pitch works on strangers. Buyers also decide early. The [6sense 2025 Buyer Experience Report](https://6sense.com/newsroom/the-timeline-for-influencing-b2b-buyers-is-shrinking-insights-from-6senses-2025-buyer-experience-report/) found buying groups had ranked preferred vendors before first contact in 94% of cases, so your website and reputation need to hold up when a prospect checks you. If you use an agency, our guide to [pay-per-meeting pricing](/insights/pay-per-meeting-cost-uk) explains what to put in the contract. ### Step 4: Put the CRM in place and use it yourself Choose a simple CRM and log your own deals in it for at least a quarter before anyone joins. If the founder does not use it, nobody will. ### Step 5: Hire, and hand over the work before the meeting first Give the new person the research, the first approaches, the follow-ups and the booking. Keep the first meetings and closing calls yourself at the start, and hand those over one segment at a time as the hire proves they can run them. ### Step 6: Review at 90 days and at six months At 90 days, check activity and early meetings against the brief. At six months, check whether meetings are turning into proposals without you in the room. Given the three to six month ramp, judging a hire on revenue in the first quarter is usually too early. This table sums up the stages: | Stage | What the founder does | What to add | |---|---|---| | Founder sells everything | Finds, meets and closes every client | Notes on every meeting, a first draft of the pitch | | Founder sells, process written | Still closes, but works from the written pitch and buyer definition | A CRM, used for every deal | | Pipeline source added | Takes the meetings a researched outbound programme or a better website brings in | Monthly reporting on meetings held and proposals | | First hire | Keeps closing calls and key relationships; coaches the hire | An SDR or account executive working from the playbook | | Small team | Sets direction, handles senior deals and the firm's public voice | A sales lead once there is a team to lead | ## SDR, account executive or outsourced: which should come first? It depends on which part of the work is the bottleneck. If you have more meetings than you can take, you need someone to run deals. If you have too few meetings, you need someone to open conversations. | Option | What they do | What they need from you | Fits when | |---|---|---|---| | SDR (sales development representative) | Researches prospects, opens conversations, books meetings | A defined buyer, a written pitch and your time for the meetings | You can still close, but first meetings are scarce | | Account executive | Runs deals from first meeting to signed contract | A pipeline to work and a proven process | You have more good meetings than you can handle | | Outsourced outbound | Researches and opens conversations for you | A definition of a qualified meeting and your time for the meetings | You want to prove a pipeline source before you hire | On pay, Sales Recruit UK puts a UK SaaS SDR at "£33K base / £49K OTE" on the UK average, and notes that B2B roles have the widest entry band, from £22,000 to £48,000 base. Entrepreneur Plus puts a first account executive at a UK start-up at around £30,000 to £40,000 base with £50,000 to £70,000 on-target earnings at the smaller end. On top of salary, employers pay [National Insurance at 15% on earnings above £5,000 a year](https://www.gov.uk/guidance/rates-and-thresholds-for-employers-2026-to-2027) in 2026 to 2027 and a [minimum 3% workplace pension contribution](https://www.gov.uk/workplace-pensions/what-you-your-employer-and-the-government-pay) on qualifying earnings, before tools, data, recruitment and your management time. Entrepreneur Plus also warns that "bringing in an ex-VP of Sales for your very first hire often backfires". A first hire needs to be comfortable doing the work, not managing it. Outsourcing is a way to build the pipeline source in Step 3 before you commit to a salary. It works best when you still take the meetings, because that keeps you close to what buyers say. ## What should the founder keep doing? Keep the parts where your authority and judgement matter, and hand over the parts that only need time. **Keep:** first meetings with senior buyers until the hire is ready; closing calls and pricing decisions; conversations with existing clients, which is where you learn what to build next; and the public voice of the firm, on LinkedIn and in the trade press. Our guide to [whether founder LinkedIn posts get you recommended by ChatGPT](/insights/founder-linkedin-chatgpt-recommendations) covers the last one. **Hand over:** research into firms and people, first approaches, follow-ups, booking, CRM updates and meeting preparation. These take most of the hours and need the least of your authority. ## What to do this month 1. Track your own week for a fortnight and note how many hours go on selling, and on which parts. 2. Write down your pitch, your five best client stories and your answers to the ten objections you hear most. 3. Describe your ideal buyer in one paragraph: sector, size, role and trigger. 4. Put every open deal into a CRM, with a next step and a date. 5. Choose one pipeline source that does not depend on your network, and set a 90-day test with a clear measure. ## Frequently asked questions ### When should I hire my first salesperson? When you can teach someone else how you win, and there is a steady source of conversations for them to work on. Entrepreneur Plus suggests after you have personally closed 10 to 20 clients, and when selling takes more than half your week. Hiring because you are tired of selling, before those conditions exist, is the mistake Techstars warns against. ### Should our first sales hire be senior or junior? Usually someone hands-on who is comfortable doing the work from scratch. Entrepreneur Plus warns that an ex-VP of Sales as a first hire often backfires, because they are several steps removed from early-stage selling. Hire a sales leader when there is a team to lead. ### How much does a first salesperson cost in the UK? Sales Recruit UK puts a UK SaaS SDR at about £33,000 base and £49,000 on target in June 2026. Entrepreneur Plus puts a first account executive at around £30,000 to £40,000 base. Add employer National Insurance at 15% above £5,000 a year, pension, tools, recruitment and your management time. ### How long before a new salesperson pays for themselves? Expect three to six months before a new SDR reaches full productivity, according to ORRJO. With average SDR tenure around 14 months, the months at full speed are limited, which is why the pitch, the buyer definition and a pipeline source need to be ready on day one. ### Can we outsource sales instead of hiring? You can outsource the work of opening conversations, which lets you prove a pipeline source before you hire. Closing usually stays with the founder or an in-house hire, because buyers want to meet the people who will do the work. ### What should the founder never hand over? The relationships with your most important clients, pricing and the firm's public voice. Keep taking some sales conversations even after the team grows. As Walsh puts it, "Every sales conversation is a product conversation." If you want a pipeline source running before you make the first hire, [Pipeline That Fills](/pipeline-that-fills) sets out how we research and write outreach for founder-led firms, or book a free [Growth Review](/growth-review) for three written fixes. ## Sources - [Techstars, Founders: your first sales hire is probably a mistake (March 2026)](https://www.techstars.com/blog/advice/founders-your-first-sales-hire-is-probably-a-mistake) - [Entrepreneur Plus, How to hire a salesperson: when to make your startup's first sales hire (August 2026)](https://entrepreneurplus.co.uk/how-to-hire-a-salesperson-when-to-make-your-startups-first-sales-hire/) - [Bright Reach, When should a founder hire the first sales reps and a VP of Sales? (April 2025)](https://brightreach.co.uk/blog/when-should-a-founder-hire-the-first-sales-reps-and-a-vp-of-sales/) - [Sales Recruit UK, SDR and BDR salary UK 2026 (June 2026)](https://salesrecruituk.com/salary-guide/sdr-bdr-salary-uk-2026/) - [ORRJO, Outsourced SDR vs in-house (January 2026)](https://orrjo.com/blog-outsourced-sdr-vs-in-house) - [6sense, 2025 Buyer Experience Report (November 2025)](https://6sense.com/newsroom/the-timeline-for-influencing-b2b-buyers-is-shrinking-insights-from-6senses-2025-buyer-experience-report/) - [GOV.UK, Rates and thresholds for employers 2026 to 2027 (September 2026)](https://www.gov.uk/guidance/rates-and-thresholds-for-employers-2026-to-2027) - [GOV.UK, Workplace pensions: what you, your employer and the government pay (checked September 2026)](https://www.gov.uk/workplace-pensions/what-you-your-employer-and-the-government-pay) --- Title: How does a B2B fintech get recommended when buyers ask ChatGPT? URL: https://fortitudemedia.ai/insights/ai-optimisation-fintech Author: Ashley Kenerson Published: 2026-10-01 Last reviewed: 2026-10-27 Topic: Sector guides Short answer: ChatGPT and other assistants name fintechs they can match to a buyer's use case and confirm elsewhere. For a B2B fintech that means a Financial Services Register entry whose name, reference number and status match your website, security certifications stated in text, listings in the directories and marketplaces you integrate with, reviews on sites such as G2 and fintech trade press. Fix those details first, especially after an acquisition, then publish a page for each buyer use case. Business buyers of financial software now do part of their due diligence in a chat window. [G2's April 2026 survey](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html) of 1,076 software buyers found that 51% start their research in an AI chatbot, up from 29% in 2025, and 45% said citations of review sites were the most confidence-inspiring signal in an AI answer. A finance director choosing a payments provider, or a compliance lead choosing a KYC platform, asks ChatGPT or Copilot for a shortlist, then asks whether each name is regulated, secure and used by firms like theirs. They also check the answer. The [TrustRadius 2026 B2B Buying Disconnect Report](https://www.prnewswire.com/news-releases/trustradius-2026-b2b-buying-disconnect-report-reveals-ai-has-changed-how-buyers-research-but-not-what-they-trust-302825792.html) (July 2026) found that 63% of technology buyers used AI during their purchase and 94% of those fact-checked it at least sometimes. In fintech the facts they check sit on the FCA's Financial Services Register, Companies House, partner directories, review sites and the trade press, and a fintech that has merged, rebranded or restructured gives those sources more room to disagree. When they do, an assistant hedges or names a rival whose story is easier to confirm. This guide is for the founder of a UK B2B fintech: payments, lending platforms, regtech, wealth and treasury tools or open banking. ## Key takeaways - Business buyers ask AI assistants for a fintech shortlist, then ask about regulation, security and integrations before they book a demo. - Assistants name fintechs whose use case, customers and credentials they can confirm in independent sources, and hedge when those sources disagree. - Your Financial Services Register entry, website, Companies House record and profiles must give the same name, reference number, status and principal. - If you are not authorised by the FCA yourself, say so plainly and explain what you are. - After an acquisition, a redirect is not enough, as ID-Pal found when AI answers kept recommending NorthRow, the firm it had bought, as a rival. - Fix the entity details first, then publish one page per buyer use case, then track a fixed set of buyer questions every month. ## What do business buyers ask ChatGPT about fintech providers? They ask who belongs on the shortlist, then whether each name checks out. The first kind of question describes a use case; the second names a provider and asks about regulation, security or integrations. | Segment | Shortlist question | Check question | |---|---|---| | Payments | "Which payment providers suit a UK B2B firm collecting invoice payments by bank transfer?" | "Is this provider authorised by the FCA, and for which payment services?" | | Lending platforms | "Which loan origination platforms do UK SME lenders use?" | "Who owns this platform, and which lenders use it?" | | Regtech | "Which KYB and KYC providers do UK wealth managers use?" | "Does this provider integrate with Salesforce, and where is our data stored?" | | Wealth and treasury | "What treasury management software suits a UK mid-market finance team?" | "Is this tool ISO 27001 certified?" | | Open banking | "Which open banking providers can check income for affordability decisions?" | "Is this provider enrolled in the Open Banking Directory?" | Shortlist questions are answered from pages that describe the use case in the buyer's words, where most fintech websites describe the technology. Check questions are answered from the Register, security pages, marketplace listings and reviews. Answers change from run to run, so ask these yourself in a clean session and note who appears. ## How does ChatGPT decide which fintechs to name? It names providers it can match to the buyer's situation and confirm from sources other than their own website. For a B2B fintech, the evidence is: - **A stated use case and customer**: "client onboarding checks for UK fund administrators", not "an identity orchestration layer". - **Regulatory status that matches the Register.** - **Security credentials in text**, with the certifying body, scope and date. - **Partner and integration listings** that confirm what you connect to. - **Reviews** from the kind of firm the buyer runs, and **trade press** that has written about you. Our guide to [why ChatGPT sounds unsure about a firm](/insights/ai-cross-references-business-sources) explains how assistants compare sources, and our [guide for B2B tech companies](/insights/ai-optimisation-technology-companies) covers documentation, pricing and comparison pages. ## How does an assistant check whether a fintech is FCA authorised? It can only check what it can find: your entry on the Financial Services Register, and pages that repeat your firm name, firm reference number (FRN) and status. If your website does not give the same details in text, there is nothing to match, and a careful answer hedges or sends the buyer to the Register. The FCA describes the [Financial Services Register](https://www.fca.org.uk/firms/financial-services-register) as "a public record of firms, individuals and other bodies that are, or have been, authorised by us or the PRA". Its consumer tool, the [FCA Firm Checker](https://www.fca.org.uk/consumers/fca-firm-checker), leaves out "products and services firms only offer to other firms or professionals", which appear on the full Register. For a B2B fintech, the full Register is the record that counts. The FCA tells anyone checking a firm to [check that the FRN and contact details they have been given match](https://www.fca.org.uk/consumers/avoid-scams-unauthorised-firms) the details it holds, because "some firms pretend to be authorised firms". A compliance lead and an assistant reading the same pages should reach the same answer: | Where you appear | What must agree | |---|---| | Financial Services Register | Legal name, trading names, FRN, status, permissions, contact details | | Website footer and legal page | The same name, trading names, FRN and status, plus your principal if you have one | | Companies House | Company name, number and registered office | | LinkedIn, G2 and marketplace profiles | Brand name, with the legal name and FRN in the description | The FCA asks firms to keep details such as "your firm's name, contact information and permissions" correct through its Connect system, and many firms must confirm them each year ([FCA, August 2026](https://www.fca.org.uk/firms/firm-information)). Treat the Register as the master record. ### What if we are not regulated ourselves? Many regtech, treasury and data firms sell to regulated firms without holding FCA permissions. Say so in one sentence, for example: "We are not authorised by the FCA; our clients are regulated firms." Never imply authorisation you do not hold. An assistant that finds no entry and no explanation may guess. ### What if we are an appointed representative? The FCA's Register page explains that appointed representatives carry out activities "on behalf of another firm, known as their principal", and that the principal "is responsible for that business". Name your principal and its FRN on your website, in the Register's words. ## What happens after an acquisition or a rebrand? Assistants may keep treating the old name as a separate company, sometimes as your competitor, until something tells them otherwise. [ID-Pal](/work/id-pal), which verifies identities for fund administrators, wealth managers, law firms and accountants, had bought NorthRow, a UK business verification platform, and folded it into id-pal.com. "The redirect from northrow.com was technically correct, but nothing on either site told Google or the AI engines that NorthRow and ID-Pal were the same company." The fix: "About four hours of structured data work declaring NorthRow part of ID-Pal on both domains. Live in week one, with no copy changes and no risk to the redirect." The result: "AI answers still recommending NorthRow as a rival: six in May, one in September 2026." If you have bought a firm or renamed a product, check that every old address redirects to its closest new page, that structured data on both domains declares the relationship, that one plain page explains what the old brand is now, and that every profile you control uses the current name. Our guide to [why ChatGPT describes a firm wrongly](/insights/ai-conflicting-information-online) covers the wider clean-up. ## Which security details should a fintech publish? Publish the facts a security questionnaire asks for, in text, on one page. Buyers ask assistants whether a provider is certified, and an assistant can only answer from what it can read. State each certification with its name, level, certifying body, scope and date: ISO 27001, a SOC 2 report if you have one, and Cyber Essentials or Cyber Essentials Plus. Say where customer data is stored, how it is encrypted and how often you commission penetration tests. The reports can stay behind an NDA; the facts about them should be public. Our guide to [awards and accreditations](/insights/awards-accreditations-ai-trust) explains which credentials carry weight. ## Which directories, marketplaces and review sites matter for a fintech? The ones that confirm who you work with and what you connect to. Outside listings are how an assistant checks a claim you make on your own site. | Listing | Who it matters for | What to get right | |---|---|---| | [Open Banking Directory](https://www.openbanking.org.uk/regulated-providers/) | Account information and payment initiation providers | Open Banking Limited lists companies "directly enrolled in the Open Banking Directory and authorised to provide secure open banking-enabled services" | | Salesforce AppExchange, Microsoft AppSource, Xero App Store | Fintechs that integrate with them | One listing per integration, using your website's product name and use case | | Banking, scheme and technology partner pages | Payments, cards and embedded finance | Ask each partner to name you and describe the work | | G2, Capterra, Gartner Peer Insights, TrustRadius | Software for finance, risk and compliance teams | Reviews from the firms you want to be named for | | AltFi, Finextra, The Paypers | Fintech buyers and investors | Coverage of results, data and people, not only funding news | Being described is not the same as being recommended. When we first recorded answers for ID-Pal: "Across 36 recorded AI answers to the questions ID-Pal's buyers actually ask, ID-Pal was named three times, and only for its brand or its Salesforce integration. AI knew who it was, and had no reason to recommend it." Being recommended for a use case needs pages written for that use case. ## What can a fintech publish without breaking the financial promotion rules? Most B2B fintech content explains how a product or process works, which is what assistants quote. Whether a page is a financial promotion depends on what it says and who it reaches, so agree a sign-off route with your compliance lead before you publish at volume. The FCA's [financial promotions page](https://www.fca.org.uk/firms/financial-promotions-adverts) says promotions "can take the form of a website, Facebook post, tweet, etc." and "must be clear, fair and not misleading regardless of the media type". The products it lists include loans, investments and "payment services and e-money". Lending and investment platforms need the most care, and a page written for businesses can still reach sole traders. Keep guides general and dated, avoid claims about approval rates or savings you cannot evidence, and put anything that invites a reader to take up a regulated product through your promotion checks. ## What should a fintech fix first? The facts an assistant uses to confirm you, then the pages it quotes, then the measurement. ### Step 1: Make your regulatory and company details agree Put your Register entry, Companies House record, website footer and main profiles in one sheet and correct every difference, starting with the Register. ### Step 2: Connect every old name to the current one List every brand, product and company name you have used or acquired. Redirect, declare the relationship in structured data and publish one page that explains it. ### Step 3: Publish one page for each buyer use case ID-Pal's version: "Five industry pages rebuilt: fund administration, wealth management, legal, accountancy and estate agency. Each opens with the question a compliance lead types, names the regulator, and answers inside the first ninety words." And the result: "Gemini now names ID-Pal for UK wealth managers. Claude names it for UK fund administrators, on a page rebuilt in August, and links NorthRow to ID-Pal in the same answer." ### Step 4: Track a fixed set of buyer questions Ask 20 to 50 buyer questions several times a month in ChatGPT, Gemini, Claude, Copilot and Perplexity, in a clean session, and count a win only when it repeats. Our guide to [tracking whether ChatGPT recommends you](/insights/track-ai-recommending-your-business) sets this up. ## What to do this month 1. **Check your Register entry against your website** and correct every difference in name, FRN, status, principal and contact details. 2. **Add a plain regulatory sentence to your footer**: who you are, whether you are authorised and, if relevant, your principal. 3. **List every name you have traded under or acquired** and check that each points to your current name. 4. **Publish or update your security page** with each certification, its scope and date. 5. **Write the first use-case page** for the segment you most want to win. 6. **Agree a content sign-off route** with your compliance lead. ## Frequently asked questions ### Do we need to be FCA authorised to be recommended by ChatGPT? No. Many B2B fintechs, especially in regtech and treasury software, sell to regulated firms without holding FCA permissions. What matters is that your status is stated plainly and matches the public record. Implying authorisation you do not hold is the fastest way to lose a buyer. ### How do I check whether a fintech provider is FCA authorised? Search the Financial Services Register at register.fca.org.uk. Check that the firm reference number and contact details match the ones you have been given, and use the FCA's contact details rather than the firm's. For services sold only to businesses, use the full Register rather than the Firm Checker. ### How long before ChatGPT starts naming us? Assistants that search the web live, such as ChatGPT search, Perplexity, Copilot and Google's AI features, can reflect a new page or corrected listing within weeks. Answers drawn from a model's own memory change only when it is retrained, which takes months. The dated example: "UK buyer questions where AI names ID-Pal: none in May, two in August, four in September 2026." ### Should we publish our SOC 2 report or ISO 27001 certificate? Publish the facts about them: which report or certificate, its scope, the audit or certification firm and the date. The documents can stay behind an NDA. An assistant can then answer "is this provider certified?" correctly, and a buyer knows what to ask for. ### Does G2 matter for a fintech that sells to banks and large firms? It matters if your buyers read it: in G2's April 2026 survey, 45% of software buyers called review-site citations the most confidence-inspiring signal in an AI answer. Buyers at larger firms may also look at Gartner Peer Insights, analyst coverage and references. Ask your last ten customers where they read reviews, and start there. ### Is a blog post about our product a financial promotion? It can be. The FCA says promotions can take the form of a website or a social post, so the format does not decide it; what the content says and who it reaches does. Keep guides general and factual, and send anything that invites a reader to take up a regulated product through your promotion checks. Buyers of financial products look for reasons to say no, and they now ask an assistant first. Our [fintech page](/for/fintech) sets out how we make the evidence easy to find, and a free [Growth Review](/growth-review) is a 30-minute call with a founder that checks what Google and ChatGPT say about your firm and two competitors, with three fixes in writing. ## Sources - [G2, New G2 research: half of B2B software buyers now start their research with AI chatbots (April 2026)](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html) - [TrustRadius, 2026 B2B Buying Disconnect Report (July 2026)](https://www.prnewswire.com/news-releases/trustradius-2026-b2b-buying-disconnect-report-reveals-ai-has-changed-how-buyers-research-but-not-what-they-trust-302825792.html) - [FCA, Financial Services Register (February 2026)](https://www.fca.org.uk/firms/financial-services-register) - [FCA, FCA Firm Checker (April 2026)](https://www.fca.org.uk/consumers/fca-firm-checker) - [FCA, Protect yourself from scams: check the firm (January 2026)](https://www.fca.org.uk/consumers/avoid-scams-unauthorised-firms) - [FCA, Keeping your firm's information up to date (August 2026)](https://www.fca.org.uk/firms/firm-information) - [FCA, Financial promotions and adverts (May 2024)](https://www.fca.org.uk/firms/financial-promotions-adverts) - [Open Banking Limited, Regulated providers (September 2026)](https://www.openbanking.org.uk/regulated-providers/) --- Title: Should the founder post on LinkedIn, and does it help ChatGPT recommend the firm? URL: https://fortitudemedia.ai/insights/founder-linkedin-chatgpt-recommendations Author: Ross Williams Published: 2026-10-01 Last reviewed: 2026-10-01 Topic: Press, reviews and being trusted Short answer: Yes, if the founder shares practical expertise and keeps it up. Semrush found LinkedIn in about 11% of AI answers in early 2026, and in ChatGPT search and Google AI Mode 59% of LinkedIn citations came from individuals rather than company pages. Cited posts share knowledge or advice, are original, and often have only 15 to 25 reactions. It helps the firm when the founder's profile, website and public records describe the firm the same way. Most founders we speak to already suspect they should be posting on LinkedIn. What has changed is the second half of the question. Buyers now ask ChatGPT, Google's AI Mode and Perplexity who to hire, and founders want to know whether a steady run of posts makes their firm more likely to be named. The short version is yes, with conditions. LinkedIn has become one of the sources AI assistants quote most. A [Semrush study published in March 2026](https://www.semrush.com/blog/linkedin-ai-visibility-study/) of 325,000 prompts found LinkedIn in about 11% of AI answers on average: 14.3% in ChatGPT search, 13.5% in Google AI Mode and 5.3% in Perplexity. Semrush placed LinkedIn second for citations on all three, ahead of Wikipedia and YouTube. Not everything on LinkedIn gets cited, though. Assistants pick up practical, original writing from people who post regularly, and pass over most of the rest. ## Key takeaways - LinkedIn is one of the most cited sources in AI answers: Semrush found it in about 11% of responses across ChatGPT search, Google AI Mode and Perplexity in early 2026. - In ChatGPT search and Google AI Mode, 59% of LinkedIn citations came from individual members rather than company pages, so the founder's own profile matters. - Assistants cite knowledge: 54 to 64% of cited posts shared knowledge or practical advice, and reshares made up only about 5% of citations. - A large following is not required. The median cited post had 15 to 25 reactions, and people with fewer than 500 followers were at least as likely to be cited. - Two posts a week and one article a month, kept up for a year, does more than a burst of daily posting that stops after three weeks. - Posting helps the firm only when the founder's profile, the website and public records describe the firm the same way, so an assistant can connect the person to the company. - A ghostwriter is fine if the thinking is the founder's and the founder approves every post. ## Does LinkedIn content actually appear in AI answers? Yes. Three large 2026 studies reached the same broad conclusion, though they measured different things. - **Semrush (March 2026)** analysed 89,000 LinkedIn URLs cited across ChatGPT search, Google AI Mode and Perplexity in January and February 2026. LinkedIn appeared in about 11% of responses and ranked second for citations on all three. - **[Profound (March 2026)](https://www.tryprofound.com/blog/linkedin-is-the-most-cited-domain-for-professional-queries-in-ai-search)** looked at about 1.4 million citations across six AI platforms between November 2025 and February 2026 and found LinkedIn was the most cited domain for professional queries. In ChatGPT, LinkedIn climbed from about eleventh to fifth among cited domains over those three months. - **Peec AI**, in an analysis of 30 million sources [reported by Search Engine Land in March 2026](https://searchengineland.com/ai-search-engines-cite-reddit-youtube-and-linkedin-most-study-473138), ranked LinkedIn third overall behind Reddit and YouTube, and found Perplexity leaned on Reddit, LinkedIn and G2 for B2B queries. All three come from companies that sell AI visibility tools, and none is specific to the UK, so treat the percentages as directional. The pattern is consistent: for professional questions, LinkedIn is one of the first places assistants look. ## Does the founder's profile get cited, or only the company page? Mostly the people. Semrush found that in ChatGPT search and Google AI Mode, individual members accounted for 59% of LinkedIn citations. Perplexity was the exception: company pages made up 59% of its LinkedIn citations. That fits what gets cited: individuals write the explanations and how-to posts that answer questions, while company pages mostly publish announcements. So the founder's profile is where effort pays back first. Keep the company page accurate and active as well: it is the firm's own record on LinkedIn, and Perplexity cites company pages heavily. ## Does posting help ChatGPT recommend the firm, or just the founder? It helps the firm when the assistant can connect the founder to it. A cited post puts the founder's name and expertise in front of the assistant. Whether the firm benefits depends on everything else saying the same thing: the founder's headline and current role, the firm's website and team page, and its public record at Companies House. Assistants cross-check what they find. When a founder's LinkedIn says one thing, the website another and the company page a third, the assistant has less reason to name the firm with confidence; we explain why in [why ChatGPT sounds unsure about a firm](/insights/ai-cross-references-business-sources). When the facts line up, each cited post becomes one more piece of evidence that this person, at this firm, knows the subject. LinkedIn is also one input among several. [Muck Rack's May 2026 analysis](https://www.globenewswire.com/news-release/2026/05/07/3290268/0/en/generative-pulse-earned-media-consistently-drives-ai-citations-holding-at-84.html) found earned media, meaning coverage others chose to give you, made up 84% of links cited by AI tools. Press coverage, reviews and your own website still carry weight; see [how ChatGPT weighs earned, owned and paid sources](/insights/earned-owned-paid-authority-signals). People count too. In the [2025 Edelman-LinkedIn B2B Thought Leadership Impact Report](https://www.edelman.com/expertise/Business-Marketing/2025-b2b-thought-leadership-report), a survey of 1,934 executives (US respondents), 64% of "hidden" decision makers, the finance, procurement and legal people who shape a purchase, said they trust thought leadership more than marketing materials when assessing a supplier, and 95% said strong thought leadership makes them more receptive to sales and marketing outreach. ## What should a founder post? Post what you would tell a good prospect across a table: how a problem works, what you see going wrong, what you would do first. Semrush found 54 to 64% of cited posts shared knowledge or practical advice, and 95% of cited posts were original rather than reshares. Specifics matter more than polish. [Scrunch's 2026 analysis of about 4,000 LinkedIn posts and whether ChatGPT cited them](https://scrunch.com/blog/linkedin-posts-robots-cant-resist-what-data-says-about-chatgpt-citations) found posts with specific technical details were 77% more likely to be cited, and naming specific companies, people, products, frameworks and tools also helped. The same study found a post's reaction count had near-zero power to predict whether ChatGPT cited it. Our guide to [what makes content read as expert](/insights/what-makes-expert-quality-content) goes further on this. | Post type | What it is for | Typical length | How often | |---|---|---|---| | Answer post | Answers one question buyers ask, in plain steps. The type assistants cite most. | 50 to 300 words | Weekly | | LinkedIn article | A fuller answer with headings, a direct first paragraph and a link to the guide on your website. Articles were 50 to 66% of LinkedIn citations in the Semrush study. | 500 to 2,000 words | Monthly | | Market observation | What you are seeing in your sector this quarter, with an example you can stand behind. | 100 to 300 words | Every two weeks | | Lesson learned | Something your firm got wrong and changed. Builds trust with people; useful to assistants when it teaches a method. | 150 to 300 words | Monthly | | Client result | A named result, with the client's written permission. Proof for buyers. | 100 to 250 words | When you have one | | Company news or reshare | Hires, awards, launches. Useful for people who already know you, rarely cited. | Short | Sparingly | Lengths for answer posts and articles follow Semrush (posts of 50 to 299 words and articles of 500 to 2,000 words were cited most); the others are our guide. ## How often should a founder post? Often enough to be a regular voice, and no more often than you can keep up for a year. Semrush found three-quarters of cited post authors had published more than five posts in the previous four weeks, which suggests assistants mostly cite regular posters. For most founders, two posts a week and one article a month fits around the day job and clears that mark. Holding it for twelve months beats posting daily for three weeks and stopping. The reasons it stops are usually the same: no system for capturing ideas, and nobody holding the founder to it. Our page on [founders who know they should post but never do](/sound-familiar/i-should-post-on-linkedin) covers both. You do not need a big audience to be cited. The median cited post in the Semrush study had only 15 to 25 reactions and no more than one comment, and people with fewer than 500 followers were just as likely to be cited as those with more. ## How should the founder's profile be set up? Treat the profile as the page a buyer, and an assistant, reads to decide who you are. Five fixes cover most of it. ### Step 1: Rewrite the headline to say who you help Replace the bare job title with a plain line on who you help and with what. For example: "Co-founder of a payroll software firm for UK accountancy practices". ### Step 2: Open the About section with facts The first two lines should name the firm, what it does, who for and where, in the same words your website uses. Opinions and story can follow. ### Step 3: Link your current role to the company page When you add your current role, choose the firm's LinkedIn page so the logo appears and the two are connected. Check the firm's name matches the website and the name registered at Companies House. ### Step 4: Use the Featured section Pin your best article, the website guide it links to, and a case study or talk. These are the first things a visitor sees after the headline. ### Step 5: Make posts visible to everyone Assistants can only read what is public. When String Nguyen [tested whether ChatGPT could find her profile (July 2026)](https://blog.thetrustedvoice.co/linkedin-profile-chatgpt-search/), she found it could see her public profile and posts set to "Anyone", and only a simplified public version of the profile. Check your visibility settings and post to "Anyone". ## Can someone else write the founder's posts? Yes, as long as the thinking is the founder's and the founder approves every word. Speechwriters have worked this way for decades. In practice a writer interviews the founder for half an hour or so, turns what they said into drafts, and the founder corrects and approves them. The founder then replies to comments personally. What does not work is a writer inventing opinions or stories, or a tool producing generic posts in the founder's name. Those posts read like everyone else's, and the evidence above suggests assistants pass over them in favour of specific, original, practical content. We cover keeping your voice in [outsourcing content without losing your voice](/insights/outsourcing-content-brand-voice), and the interview method in [our best thinking is in our heads](/sound-familiar/our-best-thinking-is-in-our-heads). ## What should a founder avoid? - **Engagement pods and bought reactions.** They inflate a number that, on Scrunch's evidence, does not predict citations anyway, and buyers can often tell. - **Generic AI posts.** If a post could have been written by anyone in your sector, it will be treated that way. Use AI to tidy a draft, not to think for you. - **Decorated Unicode text.** Scrunch found posts using Unicode formatting, the bold and italic letters made from special characters, were 58% less likely to be cited by ChatGPT. - **Automation that posts, comments or connects for you.** It puts the account at risk, and automated replies rarely convince anyone. - **Posting only about the firm.** A feed of news and reshares gives assistants nothing to quote. ## Is posting the same as using LinkedIn to find clients? No. Posting builds a reputation that buyers and assistants can find. Outreach starts a conversation with a specific buyer who has not found you yet. The two work together: a prospect who receives a message usually checks the sender's profile, and a profile full of useful posts gives them a reason to reply. Outreach done well is researched and personal. Our work for [Helm Club](/work/helm-club), a founder membership community, is an example: "Around 2,000 well-matched founders approached a month across three LinkedIn accounts, each checked against their live profile first. The first message asks a question. No pitch, no link, and every reply gets a same-day answer from a person." That is what [Pipeline That Fills](/pipeline-that-fills) does. If the idea of LinkedIn outreach makes you wince, read [I will not send LinkedIn spam](/sound-familiar/i-will-not-send-linkedin-spam). ## What to do this month 1. Rewrite the founder's headline and the first two lines of the About section so they match how your website describes the firm. 2. Link the founder's current role to the company page, and check the firm's name matches Companies House and the website. 3. List ten questions buyers ask in first meetings. Each one is a post, and the best three are articles. 4. Book a 30-minute recorded conversation every fortnight to capture the founder's answers, whoever writes them up. 5. Publish two posts a week and one article a month for the next three months, with every post set to "Anyone". 6. At the end of the month, note which posts drew comments from buyers and which topics to repeat. ## Frequently asked questions ### Can ChatGPT read LinkedIn posts? ChatGPT search can find and cite public LinkedIn content: in the Semrush study, LinkedIn appeared in 14.3% of ChatGPT search responses. It can only read what is public, so the profile must be public and posts set to "Anyone". ### How long before posting makes a difference? Assistants that search the web live, such as ChatGPT search, Perplexity, Copilot and Google's AI features, can pick up a new post or article within weeks. Answers drawn from a model's own memory change only when the model is retrained, which takes months. In our experience, buyers start mentioning posts in first meetings after a few months of steady posting. ### Do I need a lot of followers? No. Semrush found the median cited post had 15 to 25 reactions, and people with fewer than 500 followers were at least as likely to be cited as those with more. Relevance and specifics matter more than audience size. ### Should we post from the company page instead? Do both, with different jobs. The founder's profile carries the expertise, which is what ChatGPT search and Google AI Mode cite most; the company page carries the firm's facts and news, which Perplexity cites more. ### Should the founder start a LinkedIn newsletter? Only once regular posts and monthly articles are running. A newsletter is a series of articles sent to subscribers, so the same rules apply. None of the studies above measured newsletters separately, so there is no evidence yet that the format itself helps citations. ### Is it all right to use ChatGPT to write LinkedIn posts? Use it to tidy, shorten or restructure what the founder has said. The ideas have to come from the founder. Cited posts are original and specific, and a tool cannot know what the founder saw this month. Read every post aloud before it goes out: if it does not sound like the founder, rewrite it. If you want this run for you, with the founder's profile rewritten, posts drafted from short interviews and every word approved by the founder, that is part of [Reputation That Wins](/reputation-that-wins). If you would rather start with a view of where you stand, book a free [Growth Review](/growth-review): 30 minutes with David Adams and three fixes in writing. ## Sources - [Semrush, We analyzed 89K LinkedIn URLs cited in AI search (March 2026)](https://www.semrush.com/blog/linkedin-ai-visibility-study/) - [Profound, LinkedIn is the most-cited domain for professional queries in AI search (March 2026)](https://www.tryprofound.com/blog/linkedin-is-the-most-cited-domain-for-professional-queries-in-ai-search) - [Search Engine Land, AI search engines cite Reddit, YouTube and LinkedIn most, Peec AI study (March 2026)](https://searchengineland.com/ai-search-engines-cite-reddit-youtube-and-linkedin-most-study-473138) - [Muck Rack, What is AI reading? Earned media drives AI citations (May 2026)](https://www.globenewswire.com/news-release/2026/05/07/3290268/0/en/generative-pulse-earned-media-consistently-drives-ai-citations-holding-at-84.html) - [Edelman and LinkedIn, 2025 B2B Thought Leadership Impact Report (June 2025)](https://www.edelman.com/expertise/Business-Marketing/2025-b2b-thought-leadership-report) - [Scrunch, LinkedIn posts that robots can't resist: what the data says about ChatGPT citations (2026)](https://scrunch.com/blog/linkedin-posts-robots-cant-resist-what-data-says-about-chatgpt-citations) - [The Trusted Voice, Does my LinkedIn profile show up in ChatGPT? I tested mine (July 2026)](https://blog.thetrustedvoice.co/linkedin-profile-chatgpt-search/) --- Title: What does pay-per-meeting lead generation cost in the UK, and what counts as a meeting? URL: https://fortitudemedia.ai/insights/pay-per-meeting-cost-uk Author: David Adams Published: 2026-09-29 Last reviewed: 2026-09-29 Topic: Meetings and pipeline Short answer: UK agencies charge £150 to £500 per qualified meeting, and retainer-only programmes run up to £8,000 or more a month (Supernova, 2026). The price only means something once the contract defines a meeting: booked, held, or held with someone who can buy. Cheap per-meeting pricing tends to lower the qualification bar, so pay for meetings held, write your definition of qualified into the contract, and keep the data. You have two quotes on the desk. One agency wants a monthly retainer. The other will charge only for the meetings it books, and the price per meeting looks modest. Paying only for results feels like the safer bet, and sometimes it is. The market price is easy to find. [Supernova's 2026 guide to UK and US lead generation pricing](https://www.supernova-ai.com/post/outsourced-lead-generation-pricing-in-2026-what-uk-and-usa-b2b-companies-actually-pay) puts pay-per-meeting lead generation at £150 to £500 per qualified meeting, with retainer-only programmes running up to £8,000 or more a month. What is harder to find is what the agency means by "a meeting". A booked slot with a junior employee who never turns up and a held conversation with the person who signs the contract can both be called a meeting, and at a low price per meeting the agency is often paid the same for either. This guide covers what UK agencies charge, what counts as a meeting, why cheap per-meeting prices lower the bar, and the six clauses a per-meeting contract needs. Our own fees are on [what it costs](/what-it-costs). ## Key takeaways - UK pay-per-meeting lead generation costs £150 to £500 per qualified meeting, according to Supernova's 2026 pricing guide, and retainer-only programmes run up to £8,000 or more a month. - A price per meeting means nothing until the contract defines a meeting: booked, held, or held with someone who can buy. - Paying per meeting booked leaves the no-show risk with you; paying per meeting held moves it to the agency. - Agencies paid purely on volume often lower their qualification bar, so write your definition of a qualified meeting into the contract before you sign. - Hybrid models, with set-up, a monthly management fee and a fee per meeting held, pay for the research and still tie most of the reward to meetings that happen. - Every contract should cover the definition of qualified, a no-show rule, a dispute and replacement window, data ownership, GDPR and PECR warranties, and a short first term. ## What does pay-per-meeting lead generation cost in the UK? Between £150 and £500 per qualified meeting, according to [Supernova](https://www.supernova-ai.com/post/outsourced-lead-generation-pricing-in-2026-what-uk-and-usa-b2b-companies-actually-pay), in a guide published in May 2026 and updated in August 2026. Where a quote falls in that range depends on how senior the buyer is, how large your market is and how tightly the meeting is defined. The same guide gives UK figures for other models. Retainer-only programmes run from the low thousands of pounds to £8,000 or more a month, depending on volume and sector. Supernova's entry tier, priced in the low thousands a month, aims at 5 to 10 qualified meetings; its mid tier, at up to £5,000 a month, aims at 10 to 20. Some agencies add a one-off set-up fee, which Supernova puts at anything from nothing to a few thousand pounds. [ORRJO's roundup of UK lead generation firms](https://orrjo.com/blog-best-b2b-lead-generation-companies-uk), updated in August 2026, lists starting prices from about £1,000 to about £4,500 a month. In the US, [Leads at Scale](https://leadsatscale.com/insights/pay-per-appointment-vs-retainer-which-pricing-model-delivers-the-best-roi/) put pay per appointment at $50 to $500 in July 2025. | Pricing model | What agencies charge | Source and date | |---|---|---| | Pay per qualified meeting (UK) | £150 to £500 a meeting | Supernova, May 2026, updated August 2026 | | Retainer only (UK) | Low thousands to £8,000 or more a month | Supernova, 2026 | | Starting prices at named UK firms | About £1,000 to about £4,500 a month, many custom | ORRJO, August 2026 | | One-off set-up fee (UK) | Nothing to a few thousand pounds | Supernova, 2026 | | Pay per appointment (US, for comparison) | $50 to $500 an appointment | Leads at Scale, July 2025 | One UK agency, [Lead Conneqt](https://www.leadconneqt.com/blog/b2b-lead-generation-cost-uk), declined to publish ranges in June 2026, arguing that "an honest range is too wide to be useful". Much of that spread comes from what each agency counts as a meeting. ## What counts as a meeting? Whatever your contract says. There are three common definitions, and a price per meeting only makes sense against the one you are buying. | Definition | What it means | What can go wrong | |---|---|---| | Booked | A prospect accepted a calendar invitation | No-shows, the wrong person, a polite yes to end a call | | Held | The call or meeting took place | Held with someone who cannot buy, or who did not know what the call was for | | Qualified and held | It took place, with a firm that fits your agreed criteria and a person who owns or materially influences the decision, who knowingly agreed to talk about your offer | Disagreements about whether the criteria were met, which a written definition settles | Leads at Scale defines a qualified appointment as one where "the lead meets your agreed-upon criteria and has expressed genuine interest". The important words are "agreed-upon criteria". If nothing is agreed in writing, almost any meeting qualifies. Lead Conneqt makes the same point from the buyer's side: the first question to ask an agency is "what counts as a lead in your pricing", before you ask how much. ## Why does a cheap price per meeting lower the qualification bar? Because the agency's costs are per approach and its income is per meeting. When the price per meeting is low, the only way to make a margin is to book more meetings from the same effort, and the easiest way to do that is to widen what counts. Supernova says it plainly: agencies on "pure pay-per-performance models often lower their qualification bar to hit volume", and buyers should "define 'qualified' in writing before signing". Example: suppose an agency charges £200 a meeting. If it takes 150 researched approaches to book one meeting with a finance director at a firm of the right size, the research and writing may cost the agency more than £200 before it has earned anything. If a meeting with anyone at a loosely matching firm takes 40 approaches, the same effort books several times as many meetings. The fee rewards the second route. These are round, illustrative numbers, not published figures. You see the result in the diary: calls with junior staff and people who did not know what the call was about. One reply on a [UK Business Forums thread](https://www.ukbusinessforums.co.uk/threads/are-there-any-lead-generation-companies-you%E2%80%99d-actually-recommend.429454/) from July 2025, written by a telemarketing provider, described lead generation firms "offering volume over quality, relying heavily on scraped data, untargeted email blasts or outsourced telemarketing". If this sounds like your last agency, our page on [meetings with people who cannot say yes](/sound-familiar/meetings-with-people-who-cannot-say-yes) sets out why it happens. ## Which pricing model carries the least risk for us? Paying per meeting held, with a written definition of qualified, carries the least risk of paying for nothing. A hybrid, which adds a set-up fee and a management fee, costs more up front but removes the pressure on the agency to cut the research that makes meetings worth having. | Model | How you pay | Who carries the no-show risk | What the agency is rewarded for | Best when | |---|---|---|---|---| | Retainer | A fixed monthly fee | You | Activity, whatever it produces | You trust the team and want control of the method | | Pay per meeting booked | A fee for each accepted invitation | You | Volume of bookings | The offer is simple and the market is large | | Pay per meeting held | A fee for each meeting that happens | The agency | Meetings that take place | You want to pay only for real conversations | | Hybrid | Set-up and management, plus a fee per meeting held | Shared, mostly the agency | Research and meetings that happen | The market is small and every approach matters | We use a hybrid for our own outbound: a fee per meeting held, on top of set-up and management, and no meeting means no meeting fee. Research is the expensive part of outreach that works, and a model that pays nothing for it tends to get less of it. Our fees are on [what it costs](/what-it-costs), and our comparison with a [lead generation agency](/vs/lead-generation-agency) explains where each approach fits. ## What should a pay-per-meeting contract include? Six clauses decide whether a price per meeting means what you think. Get each in writing. ### Step 1: Write down what qualified means List the sectors, sizes, locations, exclusions and roles that count. Our own definition has four parts: a firm that matches the agreed sector, size, geography and exclusions; a person who owns or materially influences the decision; someone who knowingly agreed to a conversation about your offer; and a meeting that took place. ### Step 2: Pay for meetings held, and set a no-show rule If you pay per meeting booked, agree what happens when the prospect does not turn up. Better still, pay only for meetings held, so a no-show costs you nothing and the agency confirms each booking. ### Step 3: Agree how disputes and replacements work Agree a short window to flag a meeting that missed the definition, and whether the agency replaces or refunds it. Without one, every dispute becomes a negotiation. ### Step 4: Make sure you own the data and the conversations The prospect list, the replies and the notes should be yours when the contract ends. The same forum reply advised buyers to "avoid long lock-in contracts or setups where the provider owns the data". Some agencies also keep the sending domains and mailboxes, so ask. ### Step 5: Get GDPR and PECR warranties The agency contacts people in your name. Under the [ICO's business-to-business guidance](https://ico.org.uk/for-organisations/direct-marketing-and-privacy-and-electronic-communications/business-to-business-marketing/), you can email people at companies and LLPs without prior consent if you identify yourself and offer an opt-out, but sole traders and some partnerships are treated as individuals. Since 5 February 2026, PECR fines can reach £17.5 million or 4% of global turnover, up from £500,000 ([Clifford Chance, February 2026](https://www.cliffordchance.com/insights/resources/blogs/talking-tech/en/articles/2026/02/key-aspects-of-the-data--use-and-access--act-take-effect.html)). The contract should warrant compliance and cover you if the agency breaks the rules. Our guide to [whether cold emailing is legal in the UK](/insights/cold-email-law-uk-b2b) covers the rules in full. ### Step 6: Start with a short term and a review A long lock-in removes the pressure to deliver. [ORRJO](https://orrjo.com/blog-outsourced-sdr-vs-in-house) suggests looking for "three-month initial terms with rolling renewals", and its August 2026 roundup of UK firms warns that "Some agencies lock you into 12-month contracts with no performance guarantees". Agree what will be reviewed at the end of the first term: meetings held, meetings that met the definition and opportunities created. ## How do we compare a per-meeting quote with a retainer? Divide what you will spend by the number of qualified meetings held, and compare that figure, not the headline price. Example: Agency A charges a retainer of £4,000 a month and holds 8 qualified meetings, which is £500 per qualified meeting held. Agency B charges £200 per meeting booked and books 20 meetings, of which 13 are held and 5 meet your definition. You spend £4,000, which is £800 per qualified meeting held. The cheaper price per meeting turned out to be the more expensive option. These are round, illustrative figures. Then set that cost against what a new client is worth to you. If the meetings turn into proposals, a higher cost per qualified meeting can be excellent value. If they rarely do, no price per meeting is low enough. ## What does careful outreach look like in practice? It looks slower per approach and better per conversation. For [Helm Club](/work/helm-club), a founder membership community, the work was "Around 2,000 well-matched founders approached a month across three LinkedIn accounts, each checked against their live profile first. The first message asks a question. No pitch, no link, and every reply gets a same-day answer from a person." What was counted was "Qualified calls a week, reported on one page every Monday and checked against Helm's own member records rather than a platform dashboard." The result: "Between 40 and 50 per cent of founders who took a call went on to join." Counting against the client's own records, rather than bookings in a dashboard, is what keeps a definition of qualified honest. ## When is pay-per-meeting the right choice? When the offer is simple, the market is large and the average sale is small enough that volume matters. Then a booked-meeting model with a clear no-show rule can be good value. When your market is a few thousand firms and you cannot afford to annoy them, pay for research and care instead. Researched outreach, where a person writes and approves every message, reaches buyers before any AI assistant is asked who to hire. ## What to do this month 1. Write a one-paragraph definition of a qualified meeting for your firm: sectors, sizes, locations, exclusions and roles. 2. Work out what a new client is worth to you, so you know what you could pay per qualified meeting held. 3. Ask every agency on your shortlist to price against your definition, per meeting held. 4. Ask each one for a sample meeting brief and a sample monthly report that shows meetings booked, held and qualified as separate numbers. 5. Check the contract for data ownership, GDPR and PECR warranties, the no-show rule and the length of the first term. ## Frequently asked questions ### Is pay-per-appointment lead generation worth it? It can be, if the contract defines a qualified meeting and you pay for meetings held. Without that, a low price per appointment often buys calls with people who cannot buy. For senior buyers in a small market, expect prices at the top of the £150 to £500 range or above it. ### Should we pay for no-shows? No. Pay for meetings held, or agree that a no-show is replaced at no cost. If an agency insists on charging for bookings, ask what share of its bookings were held last quarter. ### How many meetings a month should an agency deliver? It depends on your market and your definition. Supernova's tiers aim at 5 to 10 qualified meetings a month at entry level, 10 to 20 in the mid tier and 20 or more at the top. Fewer meetings that meet a tight definition are usually worth more than a full diary of loose ones. ### Who owns the data when the contract ends? You should. Write into the contract that the prospect list, replies, notes and any sending domains bought for your campaign pass to you when it ends. Some agencies keep the data and the mailboxes, so check before you sign. ### Is it legal for an agency to cold email UK businesses for us? Yes, for companies and LLPs, if the sender identifies itself and offers an opt-out, under the ICO's guidance. Sole traders and some partnerships need consent. You remain responsible for outreach sent in your name, so get warranties in the contract. ### Can an agency guarantee meetings? An agency can guarantee what it controls: the effort, the reporting and a fee that is only due when a qualified meeting is held. Be wary of guarantees on booked meetings with no definition attached. If you would rather pay for meetings that happen, with people who can say yes, [Pipeline That Fills](/pipeline-that-fills) sets out how we research, write and report outreach, and [what it costs](/what-it-costs) lists our fees. ## Sources - [Supernova, Outsourced lead generation pricing in 2026: what UK and USA B2B companies actually pay (May 2026, updated August 2026)](https://www.supernova-ai.com/post/outsourced-lead-generation-pricing-in-2026-what-uk-and-usa-b2b-companies-actually-pay) - [Leads at Scale, Pay per appointment vs retainer: which pricing model delivers the best ROI (July 2025)](https://leadsatscale.com/insights/pay-per-appointment-vs-retainer-which-pricing-model-delivers-the-best-roi/) - [Lead Conneqt, B2B lead generation cost in the UK (June 2026)](https://www.leadconneqt.com/blog/b2b-lead-generation-cost-uk) - [ORRJO, Best B2B lead generation companies in the UK (updated August 2026)](https://orrjo.com/blog-best-b2b-lead-generation-companies-uk) - [ORRJO, Outsourced SDR vs in-house (January 2026)](https://orrjo.com/blog-outsourced-sdr-vs-in-house) - [UK Business Forums, Are there any lead generation companies you would actually recommend? (July 2025)](https://www.ukbusinessforums.co.uk/threads/are-there-any-lead-generation-companies-you%E2%80%99d-actually-recommend.429454/) - [ICO, Business-to-business marketing (May 2025)](https://ico.org.uk/for-organisations/direct-marketing-and-privacy-and-electronic-communications/business-to-business-marketing/) - [Clifford Chance, Key aspects of the Data (Use and Access) Act take effect (February 2026)](https://www.cliffordchance.com/insights/resources/blogs/talking-tech/en/articles/2026/02/key-aspects-of-the-data--use-and-access--act-take-effect.html) --- Title: Is cold emailing other businesses legal in the UK? URL: https://fortitudemedia.ai/insights/cold-email-law-uk-b2b Author: David Adams Published: 2026-09-24 Last reviewed: 2026-09-24 Topic: Meetings and pipeline Short answer: Yes, for most businesses. Under PECR you can send marketing emails to companies, LLPs and Scottish partnerships without consent, as long as you identify yourself and give a working way to opt out. Sole traders and ordinary partnerships count as individuals and need consent. UK GDPR also applies to named contacts, so you need a lawful basis, usually legitimate interests, and must say where you got their details. Since 5 February 2026, PECR fines can reach £17.5 million or 4% of global turnover. Most founders who ask this have heard two opposite answers. One adviser says any unsolicited email breaks GDPR. A lead generation agency says business email is exempt from everything. Neither is right, and the gap between them is where firms get into trouble. The short version: you can email people at UK companies and LLPs without asking first, as long as you say who you are, give them a working way to opt out and have a fair reason to contact them. Sole traders and most ordinary partnerships are different, because the law treats them like consumers. The stakes also went up this year. Since 5 February 2026, fines under the Privacy and Electronic Communications Regulations (PECR) can reach £17.5 million or 4% of global annual turnover, up from £500,000, under the Data (Use and Access) Act 2025 ([Clifford Chance, February 2026](https://www.cliffordchance.com/insights/resources/blogs/talking-tech/en/articles/2026/02/key-aspects-of-the-data--use-and-access--act-take-effect.html)). This guide sets out the rules in plain English, based on the [ICO's business-to-business marketing guidance](https://ico.org.uk/for-organisations/direct-marketing-and-privacy-and-electronic-communications/business-to-business-marketing/) and the changes the new Act brought in. It is not legal advice: if your situation is unusual, or you plan to email at scale, take advice from a data protection lawyer. ## Key takeaways - Under PECR you can send marketing emails to companies, LLPs and Scottish partnerships without consent, provided you identify yourself and give a valid address for opting out. - Sole traders and ordinary partnerships in England, Wales and Northern Ireland count as individuals, so you need their consent or the soft opt-in, which only covers people whose details you collected while selling to them or negotiating a sale. - An email to a named person, such as jane.smith@company.com, involves personal data, so UK GDPR applies as well, and most firms rely on legitimate interests as their lawful basis. - The Data (Use and Access) Act 2025 raised the maximum PECR fine to £17.5 million or 4% of global annual turnover from 5 February 2026. - Bought and scraped lists are not banned outright, but you must tell people where you got their details by your first message at the latest, and Gmail tells senders not to buy addresses at all. - When someone objects, stop at once and add them to a suppression list rather than deleting them, so they are never contacted again. ## Is cold emailing a business legal in the UK? Yes, for most businesses. Two sets of rules apply, and a compliant email has to pass both. PECR decides whether you may send marketing by email at all. It draws its line by the type of recipient, not by whether the message feels like business to business. UK GDPR decides whether you may use a person's details, and it applies whenever the address or your records identify an individual. The ICO is plain about this: "If you are processing personal data when sending marketing by electronic mail to another business, you need to comply with the UK GDPR." It gives jane.smith@company.com as an example of an address that identifies a person. So "B2B is exempt from GDPR" is wrong, and so is "you need consent to email anyone". For a named contact at a limited company: no consent under PECR, legitimate interests under UK GDPR, and your identity plus an opt-out in every message. ## Who counts as a corporate subscriber, and who counts as an individual? The ICO calls the recipients you may email without consent "corporate subscribers". Everyone else is an "individual subscriber", even if they run a business. | Recipient | How PECR treats them | What you need before emailing | |---|---|---| | Limited company or plc | Corporate subscriber | No consent. Identify yourself and give a valid opt-out address. | | Limited liability partnership (LLP) | Corporate subscriber | No consent. Identify yourself and give a valid opt-out address. | | Scottish partnership | Corporate subscriber | No consent. Identify yourself and give a valid opt-out address. | | Some government bodies | Corporate subscriber | No consent. Identify yourself and give a valid opt-out address. | | Sole trader | Individual subscriber | Consent, or the soft opt-in | | Ordinary partnership in England, Wales or Northern Ireland | Individual subscriber | Consent, or the soft opt-in | | Other unincorporated bodies of individuals | Individual subscriber | Consent, or the soft opt-in | The soft opt-in rarely helps with cold email. It covers people whose details you collected during a sale or the negotiation of a sale, for your own similar products and services, where you gave them a clear chance to opt out. A prospect you have never dealt with is not covered. Check before you send. The ICO warns: "If you are unsure whether the contact details belong to an individual subscriber or a corporate subscriber this puts you at risk of breaching PECR." In practice that means confirming "Ltd", "plc" or "LLP" on Companies House or the firm's own website before a contact goes on your list, and leaving out anyone you cannot confirm. Freelancers, consultants and small professional practices are the usual trap. ## What is a legitimate interests assessment, in plain terms? It is a short written record showing you thought about the person on the receiving end before you emailed them. Where PECR does not require consent, the ICO says "in many cases it is likely that legitimate interests will be the appropriate lawful basis", and it describes three tests: identify the interest, show the processing is necessary, and balance it against the person's interests, rights and freedoms. Write your answers down once for each target market. One page is usually enough. ### Step 1: Name the interest Say what you want to achieve, for example offering a compliance service to finance directors at UK fund administrators. The Data (Use and Access) Act 2025 now says that direct marketing can be a legitimate interest. The ICO groups this with changes that make the law easier to follow "without materially changing how you can use personal information", so it does not remove the next two tests. ### Step 2: Show that email to this person is necessary Explain why emailing this person is a reasonable way to reach that goal. Relevance does most of the work. A message to the person who owns the problem, about that problem, is easy to justify. The same message sent to everyone with a job title is not. ### Step 3: Balance it against the person's interests Ask whether someone in that role would reasonably expect to hear from a firm like yours at their work address, whether you are using only work contact details, and whether opting out is easy. In our view, if the answers are yes, the balance usually favours you. If you are emailing personal addresses, or writing to people about decisions outside their role, it does not. ## What changed with the Data (Use and Access) Act 2025? The basic line between corporate and individual subscribers did not move. Three things did. | Change | When | What it means for your outbound | |---|---|---| | Maximum PECR fine raised to £17.5 million or 4% of global annual turnover, from £500,000 | 5 February 2026 | A breach of the email marketing rules now carries penalties on the same scale as UK GDPR. | | Direct marketing named as something that can be a legitimate interest | In force; the ICO confirmed on 19 June 2026 that all the Act's data protection provisions had commenced | Clearer footing for legitimate interests, but you still need the necessity and balancing tests. | | A duty to handle data protection complaints | 19 June 2026 | You must help people complain, for example with an electronic complaints form, acknowledge complaints within 30 days and respond without undue delay. | One more thing to watch: the ICO's own B2B page now opens with the note that "Due to changes made by the Data (Use and Access) Act, this guidance is under review and may be subject to change." Check it again before any large campaign. ## Can we buy a list or use scraped contacts? You can, but it moves the risk onto you, and it rarely pays. PECR does not ban bought lists of corporate subscribers. UK GDPR still applies to every named contact on them, and the ICO's B2B guidance says that if you buy a list of business contacts, "you must also tell people that you have obtained their data". Its [guidance on the right to be informed](https://ico.org.uk/for-organisations/uk-gdpr-guidance-and-resources/individual-rights/individual-rights/right-to-be-informed/) sets the deadline: if you use the data to contact someone, you must give them privacy information "at the latest, when the first communication takes place", including where the data came from. There are three further problems: - **Bought lists mix in sole traders and partnerships**, who need consent. You inherit the seller's mistakes. - **Scraping LinkedIn breaks LinkedIn's rules.** Its help centre says it does not permit software that scrapes the site, and that members who use such tools "risk having their accounts restricted or shut down" ([LinkedIn Help](https://www.linkedin.com/help/linkedin/answer/a1341387)). - **Mailbox providers treat bought addresses as a warning sign.** Google's [email sender guidelines](https://support.google.com/a/answer/81126?hl=en) say: "Don't purchase email addresses from other companies." If you use a list, ask the supplier in writing where each record came from, when it was collected, what people were told and whether sole traders were removed. Better still, build the list yourself, one checked firm and person at a time. ## What must a compliant first email contain? Keep it short, but these elements have to be there. | Include | Avoid | |---|---| | Your real name, your firm's name and a way to contact you | A disguised sender or a "From" line that hides who you are | | A working opt-out, such as "reply 'no thanks' and we will not contact you again" | An opt-out that needs a login or a long form, or that does not work | | A line on where you got their details, with a link to your privacy notice | Silence about where the data came from | | One clear reason you are writing to this person, in their role | A generic pitch that could go to anyone | | A plain subject line that matches the message | "Re:" or "Fwd:" on a first email, which Google's guidelines also warn against | | Work contact details only | Personal email addresses | The first three rows are legal requirements for most cold emails to named people. The rest are what keep your legitimate interests assessment honest. ## What should we do when someone objects? Stop, record it and never contact them again. Under UK GDPR the right to object to direct marketing is absolute; the ICO says "there are no grounds for you to refuse". For corporate subscribers under PECR, the ICO says "you should comply with a corporate subscriber's opt-out request" as well. Do not simply delete them. The ICO advises adding the business or contact to a "do not contact" or suppression list "instead of simply deleting all record of them", because a deleted contact can reappear in the next list you build or buy. Share that list with anyone who sends on your behalf, and check every new list against it before sending. If someone complains about how you used their data, the Act now requires you to acknowledge the complaint within 30 days and deal with it without undue delay. A named owner and a simple complaints route in your privacy notice cover it. ## What records should we keep? | Record | Why it matters | |---|---| | Your legitimate interests assessment | Shows you considered the recipient before sending | | The source and date for each contact | Needed for privacy information and for answering any complaint | | How you screened out sole traders and partnerships | Shows you applied PECR's line | | The privacy notice in use at the time | Shows what people were told | | A dated suppression list | Proves opt-outs were honoured | | A complaints log | Meets the 30-day acknowledgement duty | | Your agency's written terms on data and compliance | Your firm's name is on the emails, whoever sends them | ## What does this mean for how you run outbound? The law and good selling point the same way. A message written by a person, to someone checked against a clear brief, with an honest opt-out, is easy to justify and rarely draws complaints. A tool firing templated emails at a bought list is where most of the risk sits. That is how we run [Pipeline That Fills](/pipeline-that-fills): researched outbound where a person writes and approves every message. It reaches buyers before any AI assistant is asked who to hire. If you have avoided outbound because it feels like spam, [that worry is common](/sound-familiar/i-will-not-send-linkedin-spam), and it is a good instinct to keep. If you are weighing up outside help, see [how we compare with a lead generation agency](/vs/lead-generation-agency). ## What to do this month 1. Go through your current prospect list and mark each contact as company, LLP, sole trader or partnership, removing anyone you cannot confirm. 2. Write a one-page legitimate interests assessment for your main target market. 3. Add a line to your email template saying where you found the person's details, with a link to your privacy notice. 4. Test your own opt-out: reply "no thanks" to one of your emails and check the address lands on your suppression list. 5. Name one person to own complaints, so any complaint is acknowledged within 30 days. 6. If an agency sends for you, ask in writing where its data comes from and how opt-outs reach your suppression list. ## Frequently asked questions ### Is cold emailing illegal under GDPR? No. UK GDPR does not ban cold email; it requires a lawful basis for using a person's details, and legitimate interests usually fits business outreach. You must also tell people where you got their details and let them object. PECR then decides whether you need consent, which depends on whether the recipient is a corporate subscriber. ### Can we email generic addresses like info@ or sales@? Yes, if the business is a corporate subscriber, and you still identify yourself and offer an opt-out. A generic address that does not identify a person is usually not personal data. Generic inboxes rarely reach the person who decides, though. ### Do we need consent to email a sole trader? Yes, unless the soft opt-in applies, which means you collected their details while selling to them or negotiating a sale. The ICO treats sole traders and ordinary partnerships in England, Wales and Northern Ireland as individual subscribers. Many freelancers, consultants and small practices fall into this group, so check before they go on a list. ### Who carries the risk if an agency sends the emails for us? Your firm's name is on every message, so treat the agency's compliance as your own risk. Ask where its data comes from, how it screens out sole traders and how opt-outs reach your suppression list, and put the answers in the contract. If the agency cannot answer clearly, that tells you what you need to know. ### Do the same rules apply when we email businesses abroad? No. This guide covers UK rules. Other countries set their own rules for email marketing, and some are stricter about business email, so take local advice before emailing prospects outside the UK. If you want outbound that stays on the right side of these rules and still fills the diary, see [Pipeline That Fills](/pipeline-that-fills), or [book a free Growth Review](/growth-review) for three fixes in writing across your website, reputation and pipeline. ## Sources - [ICO, Business-to-business marketing (May 2025)](https://ico.org.uk/for-organisations/direct-marketing-and-privacy-and-electronic-communications/business-to-business-marketing/) - [Clifford Chance, Key aspects of the Data (Use and Access) Act take effect (February 2026)](https://www.cliffordchance.com/insights/resources/blogs/talking-tech/en/articles/2026/02/key-aspects-of-the-data--use-and-access--act-take-effect.html) - [ICO, The Data (Use and Access) Act 2025: what does it mean for organisations? (June 2026)](https://ico.org.uk/about-the-ico/what-we-do/legislation-we-cover/data-use-and-access-act-2025/the-data-use-and-access-act-2025-what-does-it-mean-for-organisations/) - [ICO, Right to be informed (January 2025)](https://ico.org.uk/for-organisations/uk-gdpr-guidance-and-resources/individual-rights/individual-rights/right-to-be-informed/) - [Google, Email sender guidelines (checked September 2026)](https://support.google.com/a/answer/81126?hl=en) - [LinkedIn Help, Prohibited software and extensions (checked September 2026)](https://www.linkedin.com/help/linkedin/answer/a1341387) --- Title: What is GEO, and is it just SEO? URL: https://fortitudemedia.ai/insights/what-is-geo-generative-engine-optimisation Author: David Adams Published: 2026-09-24 Last reviewed: 2026-09-24 Topic: Getting named by AI assistants Short answer: GEO (generative engine optimisation) means work aimed at getting your firm named and cited in answers from ChatGPT, Gemini, Copilot and Google's AI features; AEO (answer engine optimisation) means much the same. Google's July 2026 guide names both and calls them still SEO, with no special files or llms.txt needed. What changes is the goal of being named, the weight of press coverage and reviews, and measuring with repeated buyer questions. UK searchers mostly call it AI SEO. Somebody has probably offered you "GEO" in the last few months. It tends to arrive as a new line on an agency proposal, or in a LinkedIn post announcing that SEO is finished. The word sounds technical, the fee often sits above the SEO line it replaces, and nobody explains what will change on your website on Monday morning. GEO stands for generative engine optimisation: work aimed at getting your firm named, quoted and linked inside the answers that AI assistants write, such as ChatGPT, Gemini, Microsoft Copilot, Claude, Perplexity and Google's AI Overviews. AEO, answer engine optimisation, is a close cousin that puts the stress on being the single direct answer a system gives. Most UK searchers use neither label. Over the 12 months to 24 September 2026, "AI SEO" drew about nine times the UK search interest of "generative engine optimization" on [Google Trends](https://trends.google.com/trends/explore?date=today%2012-m&geo=GB&q=AI%20SEO,generative%20engine%20optimization,answer%20engine%20optimization,AI%20visibility,LLM%20SEO), and GEO drew more interest than AEO. Google has now given its view. Its 2026 guide to generative AI search names both terms and says that, from Google Search's perspective, optimising for AI answers is still SEO ([Google Search Central, updated July 2026](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide)). That holds for Google and only partly for everyone else. This guide explains where the term came from, what Google says, what does change when the goal is being named rather than ranked, and how to judge an agency that sells GEO. ## Key takeaways - GEO (generative engine optimisation) means work aimed at getting a firm named and cited in AI-written answers; AEO (answer engine optimisation) means much the same, with more stress on being the one direct answer. - The term comes from a research paper by Princeton and IIT Delhi researchers, first posted in November 2023 and presented at KDD 2024, which found that adding quotations, statistics and cited sources made a page more visible in generated answers. - Google's own guide, updated in July 2026, names GEO and AEO and calls them still SEO: no special files, no llms.txt, no chunking and no separate pages for every variation of a question. - What does change is the goal (being named in an answer), the reach of each question (Google fans one question out into several searches), the weight of independent sources such as press coverage and reviews, and the way results are measured. - UK searchers mostly call this work "AI SEO", so the label on a proposal tells you little about the work behind it. - Judge any GEO offer by what it will change on your site and beyond it, how it will measure named answers over time, and whether its claims match Google's published guidance. ## What does GEO actually mean? GEO is the practice of shaping your website and your wider reputation so that AI assistants mention your firm when a buyer asks a question you can answer. The target is a sentence in an answer, such as "firms worth speaking to include...", rather than a position in a list of links. The phrase comes from a paper called [GEO: Generative Engine Optimization](https://arxiv.org/abs/2311.09735) by Pranjal Aggarwal and colleagues at Princeton and IIT Delhi, first posted in November 2023 and accepted at KDD 2024, a major data science conference. The authors described generative engines as systems that search the web and then summarise several sources with a large language model. They tested simple changes to web pages and measured how visible each page became in the generated answer. Three changes helped most. Adding quotations from relevant people raised visibility by about 41%, adding statistics by about 34% and citing credible sources by about 30%. Keyword stuffing lowered visibility by about 8% on the main measure. Pages that ranked lower in the underlying search results gained the most, which is encouraging for smaller firms up against large sites. Two cautions. The tests ran on a research engine and a benchmark built in 2023, so the figures are directional and say nothing specific about ChatGPT or Google in 2026. And the changes that worked are the ones a good editor would make anyway: evidence, named sources and clear writing. ## What is AEO, and is it different from GEO? AEO, answer engine optimisation, means making your content the answer a system gives, rather than one link among several. The analyst firm Forrester uses the term in its January 2026 buyer research, telling vendors they will need to move from "driving traffic through search engine optimization to driving visibility through answer engine optimization" ([Forrester, January 2026](https://www.forrester.com/blogs/b2b_buyers_make_zero_click_buying_number_one/)). In practice agencies use the two labels for the same work. Here is how the main terms compare. | Term | Stands for | Where you will see it | How UK searchers use it | |---|---|---|---| | AI SEO | SEO for AI search | Agency sites, founder conversations | The most searched label: about nine times "generative engine optimization" over 12 months (Google Trends, UK) | | GEO | Generative engine optimisation | The 2023 research paper, agency proposals, Google's 2026 guide | Well behind AI SEO, but ahead of AEO | | AEO | Answer engine optimisation | Forrester research, Google's 2026 guide | Behind GEO | | SEO | Search engine optimisation | Everywhere | Still far more searched than any AI label | Google Trends shows relative interest, not search volumes, so read the table as a ranking. The practical point: if you write about this subject for UK buyers, use the words they type and explain the acronyms once. ## Is GEO just SEO? What does Google say? For Google, yes. Google's guide, "Optimizing your website for generative AI features on Google Search", published in May 2026 and last updated on 10 July 2026, names AEO and GEO and says: "From Google Search's perspective, optimizing for generative AI search is optimizing for the search experience, and thus still SEO" ([Google Search Central](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide)). The same guide lists things you do not need to do for Google's AI features: - You do not need new machine-readable files, AI text files, llms.txt, special markup or Markdown copies of your pages. - You do not need to break content into tiny chunks for AI to understand it. - You do not need to write in a special way just for AI search. - Chasing inauthentic mentions across the web is not as helpful as it may seem. - Structured data is not required, and there is no special schema markup for AI. It also warns that creating separate content for every possible variation of how people might search breaks Google's scaled content abuse spam policy. And it says the content that counts most is "non-commodity": first-hand expert experience rather than the common-knowledge lists anyone could write. Google speaks only for Google. ChatGPT search, Copilot, Claude and Perplexity run their own systems. ChatGPT search, for example, can only include your site if you allow OpenAI's search crawler, OAI-SearchBot ([OpenAI Help Center](https://help.openai.com/en/articles/9237897-chatgpt-search)). Most of the groundwork is shared, which is why our guide on [whether you still need SEO](/insights/seo-vs-llm-optimisation) says to keep the technical basics and build on them. ## What actually changes when the goal is being named? Four things change, and they are the fair case for treating this as more than a new name. **A named answer is not a ranked list.** Ten links give ten firms a chance. An assistant's answer usually names a handful, and there is no page two. A firm can rank well and still be left out, which our guide on [getting ChatGPT to recommend your firm](/insights/what-is-llm-optimisation) explains in detail. **One question becomes several searches.** Google's AI features use query fan-out: the model runs a set of related searches at the same time to build its answer. That is one reason Ahrefs found in March 2026 that only 38% of AI Overview citations came from pages ranking in the top ten, down from 76% in July 2025 ([Ahrefs, March 2026](https://ahrefs.com/blog/ai-overview-citations-top-10)). Ranking still helps. Covering the follow-up questions a buyer would ask matters as well. **Other people's words count for more.** Muck Rack's May 2026 analysis of more than 25 million links cited by ChatGPT, Claude and Gemini found that 84% came from earned media, meaning coverage written by someone else, and 0.3% from paid content ([Muck Rack, May 2026](https://www.globenewswire.com/news-release/2026/05/07/3290268/0/en/generative-pulse-earned-media-consistently-drives-ai-citations-holding-at-84.html)). Reviews matter too: in [G2's April 2026 survey](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html) of 1,076 software buyers, 45% said citations of review sites were the most confidence-inspiring signal in an AI answer. **Measurement works differently.** There is no rank to check, and answers vary from run to run. The useful measure is a fixed list of 20 to 50 buyer questions, asked several times a month in a clean session, with a win counted only when it repeats. Our method for ID-Pal reads: "The same buyer questions run every month in ChatGPT, Gemini and Claude, in a fresh chat. A win only counts when it repeats, and losses are reported too." The result so far: "UK buyer questions where AI names ID-Pal: none in May, two in August, four in September 2026" ([ID-Pal case study](/work/id-pal)). Here is the whole picture in one table. | Area | Classic SEO | Warning signs in a GEO proposal | What actually changes | |---|---|---|---| | Goal | Rank for a keyword | A promise to "rank you in ChatGPT" | Be named and cited in answers to real buyer questions | | Technical access | Googlebot can crawl and index the site | Special AI files and markup as the main fix | Also let AI search crawlers in (OAI-SearchBot, Bingbot and others) and keep key text in the page's HTML | | Content | Pages built around keywords | A new page for every variation of a question | Fewer, deeper pages that answer a question and its follow-ups with evidence | | Authority | Links to your site | Mentions placed anywhere that will take them | Earned coverage, reviews and the same facts about you on independent sites | | Measurement | Rankings and clicks | One "visibility score" with no questions behind it | Repeated buyer questions, counted only when a result repeats | | Timescale | Months | Guaranteed results within weeks | Weeks for assistants that search live, months for answers from a model's memory | ## Why do you keep hearing GEO if Google says it is SEO? Because a new label helps sell a new line on an invoice. That is not always cynical. In our experience, many SEO programmes left out press coverage, reviews and question-led writing, and a firm that adds those has changed its work, whatever it calls it. The label matters far less than the list of changes behind it. ## How do you judge an agency selling GEO? Ask for specifics. A credible provider can show you what the assistants say about you today and tell you exactly what it will change. These five steps take about an hour per proposal. ### Step 1: Ask to see today's answers Ask the agency to run ten of your buyer questions in ChatGPT, Gemini, Copilot and Google's AI Mode before you sign, and to show you the answers, including who is named instead of you. If they cannot, they are guessing. ### Step 2: Ask what will change, and where Every change should sit in one of three places: your website (structure, pages, facts), your content (questions answered, evidence added) or other people's sites (coverage, reviews, directories). A proposal that promises "AI optimisation" without naming a single page is a warning sign. ### Step 3: Check their claims against Google's guide If the plan leans on llms.txt files, splitting pages into chunks, special AI schema or a new page for every variation of a question, it runs against Google's published guidance. Some of it may help with other assistants, so ask them for the evidence. ### Step 4: Ask how they will measure Look for a fixed list of buyer questions, repeated runs, losses reported alongside wins, and a monthly report you can read in five minutes. Our guide on [tracking whether ChatGPT recommends you](/insights/track-ai-recommending-your-business) sets out the method. ### Step 5: Ask what they will guarantee Nobody can guarantee what an AI assistant will say. A provider can guarantee effort, reporting and outcomes it controls. Be wary of anyone who guarantees a place in ChatGPT's answers. Our guide on [comparing AI search providers](/insights/how-to-compare-ai-optimisation-providers) has a longer list of questions, and [what AI search optimisation costs in the UK](/insights/ai-search-optimisation-cost-uk) covers budgets. ## What to do this month 1. Write down the 20 buyer questions that lead to your services, in your buyers' own words. 2. Run them in ChatGPT, Gemini, Copilot and Google, in a clean session, and note who is named and which sources are cited. 3. Check that your robots.txt and your CDN or firewall let OAI-SearchBot, Bingbot and Googlebot in. 4. Pick the three questions you most want to win and make sure one page on your site answers each directly, with evidence, in its first third. 5. Read any GEO or AI SEO proposal on your desk against the five steps above. ## Frequently asked questions ### Is GEO a real discipline or a rebrand of SEO? Both, depending on who sells it. The foundations are SEO, and Google says so. What is new is the goal of being named in an answer, the weight of press coverage and reviews, and measuring with repeated questions instead of rankings. Judge the work behind the label. ### What does AEO stand for? Answer engine optimisation. It means making your content the direct answer an engine gives. Today agencies use it almost interchangeably with GEO, and Google's 2026 guide names both. ### Do we need an llms.txt file for GEO? Not for Google. Google's guide says its search ignores llms.txt, so the file will neither help nor harm you there. It is cheap and harmless to publish one for other tools, but it is not a ranking factor, and a plan built around it is a weak plan. ### Does GEO replace SEO? No. Google's AI features, ChatGPT search and Copilot all start from a search index, so pages that search engines cannot find or trust rarely appear in answers. Keep the SEO basics, then add what makes a firm quotable: clear answers, evidence and coverage on other sites. ### How long does GEO take to work? Assistants that search the web live, such as ChatGPT search, Perplexity, Copilot and Google's AI features, can reflect a new page or article within weeks. Answers drawn from a model's own memory change only when the model is retrained, which takes months. Plan for both. ### Can a GEO agency guarantee we will be recommended by ChatGPT? No. Nobody controls what an AI assistant says, and answers vary from run to run. A provider can fairly guarantee the work it does, its reporting and outcomes it controls. Treat a guaranteed place in ChatGPT's answers as a reason to ask more questions. ### Is structured data part of GEO? It helps search engines read your facts correctly, so keep it accurate. Google says its AI features need no special markup, and OpenAI and Anthropic have not said they read it. Treat it as housekeeping. If you want to know what the assistants say about your firm today and which three changes would matter most, book a free [Growth Review](/growth-review). The ongoing work of earning mentions sits within [Reputation That Wins](/reputation-that-wins). ## Sources - [Aggarwal and others, GEO: Generative Engine Optimization, KDD 2024 (June 2024)](https://arxiv.org/abs/2311.09735) - [Google Search Central, Optimizing your website for generative AI features on Google Search (July 2026)](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide) - [Google Trends, UK search interest in AI SEO and related terms, 12 months to 24 September 2026 (September 2026)](https://trends.google.com/trends/explore?date=today%2012-m&geo=GB&q=AI%20SEO,generative%20engine%20optimization,answer%20engine%20optimization,AI%20visibility,LLM%20SEO) - [Ahrefs, share of AI Overview citations from top-ten pages (March 2026)](https://ahrefs.com/blog/ai-overview-citations-top-10) - [Muck Rack, What is AI reading? (May 2026)](https://www.globenewswire.com/news-release/2026/05/07/3290268/0/en/generative-pulse-earned-media-consistently-drives-ai-citations-holding-at-84.html) - [Forrester, B2B buyers make zero-click buying number one (January 2026)](https://www.forrester.com/blogs/b2b_buyers_make_zero_click_buying_number_one/) - [G2, The Answer Economy research on software buyers (April 2026)](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html) - [OpenAI Help Center, ChatGPT search (accessed September 2026)](https://help.openai.com/en/articles/9237897-chatgpt-search) --- Title: Can we outsource our content without losing our voice? URL: https://fortitudemedia.ai/insights/outsourcing-content-brand-voice Author: Ross Williams Published: 2026-04-22 Last reviewed: 2026-09-24 Topic: Content that gets quoted Short answer: Yes, if you give the writer what they cannot invent: your opinions, your real examples and a named expert who reviews for substance. Write a three to five page voice guide, brief each piece with the buyer's question and your two-sentence answer, record a twenty-minute call, and have the expert review before the editor. AI tools can transcribe and draft, but the angle, examples and final check must stay human. Budget two revision rounds for the first few pieces. You know the content needs writing. You also know you are not going to write it, because you have a business to run. So you hire a freelancer or an agency, and six weeks later you get a draft that is grammatically perfect, says nothing you would say, and could have been published by any competitor with the logo swapped. That draft is worse than no draft. Buyers notice, and it gives AI assistants nothing to quote. In a [Princeton and IIT Delhi study presented at KDD 2024](https://arxiv.org/abs/2311.09735), adding quotations from relevant people raised a source's visibility in AI-generated answers by about 41%, adding statistics by about 34% and citing sources by about 30%. It was tested on a research engine, so treat it as directional. Google's [guide to its generative AI features](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide) (updated July 2026) says the same in plainer words: commodity content "is often based on common knowledge, which could originate from anyone, and typically adds little unique insight for readers". Buyers read it the same way. In the [2025 Edelman and LinkedIn B2B Thought Leadership Impact Report](https://www.edelman.com/expertise/Business-Marketing/2025-b2b-thought-leadership-report), a survey of US executives, 64% of the "hidden" decision makers who shape a purchase said they trust thought leadership more than marketing materials when judging what a supplier can do. You do not have to write everything yourself. You do have to give whoever writes for you three things they cannot invent: what you think, what you have seen, and a named expert who checks the draft. Here is how, in a few hours a month. ## Key takeaways - Outsourced content fails when the writer gets a topic but not your opinions, your examples or your experience. - A three to five page voice guide, written by a partner, is the most useful document you can hand any writer, human or AI. - Brief each piece with the buyer's question, your answer in two sentences, where you disagree with the usual advice and a real example. - AI tools can transcribe the briefing call, draft and tidy, but the angle, the examples and the final check must come from someone who knows the subject. - Have the expert review for substance before anyone edits for grammar, because polish cannot fix a piece that missed the point. - The same voice guide and recorded calls can feed a founder's LinkedIn, provided the founder approves every post. ## Why does outsourced content so often sound like nobody? Because the writer was given a topic and nothing else. "Write 1,500 words on cash flow forecasting for construction firms" is a topic, not a brief. A competent writer will read what everyone else has published and produce a tidy summary of it. That is the generic article, and it is what you asked for. What the writer did not have was your view. For example, you may think most construction firms forecast monthly when they should forecast weekly, because you have watched clients run short of cash between month-end reports. You may remember a firm that nearly failed because one main contractor paid late for four months. None of that is on the internet. Unless it reaches the writer, it will not reach the page. Voice is also more than word choice: it is which trade-offs you admit, which examples you reach for and how blunt you are. And incentives differ: a freelancer's job is to get a draft approved, while yours is to be named when a buyer asks an assistant who knows this subject. [What makes ChatGPT treat content as expert rather than marketing](/insights/what-makes-expert-quality-content) covers what that takes: first-hand experience, specific claims and a named author. ## What should a brand voice guide for a B2B firm include? A plain statement of what your firm thinks and how it talks, three to five pages long, written by a partner. It pays back every time a piece is written, whoever or whatever drafts it. - What you believe about your field, in three or four sentences. - How you sound, shown with sentences you like and sentences you would never publish. - How specific you are: numbers, real situations, worked examples. - How you handle trade-offs, including when the answer is "it depends". - Who the reader is: a busy, sceptical managing director with no patience for jargon. - What to avoid: subjects, claims, competitors and words. - Who signs off each subject, so the writer knows whose view to follow. Example: a page like this, with your own sentences in it, tells a writer more than any list of adjectives. | We write | We never write | |---|---| | "Most firms your size do not need a full audit. Here is when you do." | "Our proven approach delivers results for ambitious firms." | | "This works in regulated sectors. In a group where each division competes, it causes friction." | "A one-stop solution for every business." | | "The first month is the hardest, because it waits on your data." | "Seamless onboarding from day one." | | "We do not know yet, and here is how we would find out." | "Industry-leading expertise you can trust." | In our experience, when this document is delegated to a marketing coordinator it comes back as a description of what the firm would like to sound like, which is different from what it thinks. Two hours of a partner's time, once, is the cost. ## How do you brief each piece so it comes back sounding like you? Give the writer the buyer's question, your answer in two sentences, the point where you disagree with the usual advice, a real example and the length. Do not give them an outline: provide the substance and let the writer find the form. A brief that works fits on one page; the examples are illustrative. | Brief field | What to write | Example | |---|---|---| | The question | In the buyer's words | "Should we lease or buy our next fleet of vans?" | | Your answer in two sentences | If you cannot write it, the piece is not ready to commission | "For firms that keep vans beyond five years, buying usually works out cheaper. For firms that replace them sooner, leasing keeps cash free." | | Where you disagree | The standard advice you think is wrong, and why | "Most guides say leasing is always cheaper. Not for firms that keep vans a long time." | | One or two real examples | Anonymised, with the detail that makes them real | "A groundworks firm with twelve vans, bought outright, running them eight years." | | Must cover | Three to five points | Tax treatment, maintenance, resale value | Length should follow the question; see [how long an article should be for ChatGPT and AI Overviews](/insights/long-form-vs-short-form-ai-prefers). Then add the most valuable input of all: a twenty-minute recorded call in which the writer asks you the question and you answer it out loud. In our experience writers get more usable material from twenty minutes of you talking than from any written brief. The recording is the source. The brief is the frame. The same discipline works on service and sector pages. When we rebuilt ID-Pal's industry pages, the case study records: "Five industry pages rebuilt: fund administration, wealth management, legal, accountancy and estate agency. Each opens with the question a compliance lead types, names the regulator, and answers inside the first ninety words." The [ID-Pal case study](/work/id-pal) also reports: "Where a page has been rebuilt, ID-Pal appears. Where it has not yet, it does not, which tells us exactly what to build next." ## Where can AI tools help, and what must stay human? AI tools can do the mechanical steps: transcribing the call, turning a transcript into a first draft, and tidying grammar. The angle, the examples and the final check for accuracy must come from a person who knows the subject, because a model knows what the internet says about your field and nothing about what you think. | Step | Can an AI tool do it? | Who owns it | |---|---|---| | Choosing the question and the two-sentence answer | No | The expert | | Transcribing the briefing call | Yes | Anyone, checking names and figures | | First draft from the transcript and voice guide | Yes, if given both | The writer, who reworks it | | Real examples and figures | No: it will invent plausible ones | The expert | | Review for substance | No | The expert | | Grammar and house style | Yes | The editor | Will Google penalise content drafted this way? Not for using AI as such. Google's [guidance on AI-generated content](https://developers.google.com/search/blog/2023/02/google-search-and-ai-content) (February 2023) says "Appropriate use of AI or automation is not against our guidelines" and "Using AI doesn't give content any special gains. It's just content." The line it draws is purpose and scale: its [scaled content abuse policy](https://developers.google.com/search/blog/2024/03/core-update-spam-policies), announced in March 2024, covers many pages made mainly to manipulate rankings "no matter whether content is produced through automation, human efforts, or some combination of human and automated processes". The bigger risk is sameness: an AI draft made without a voice guide and a transcript is the most generic version of the piece there is, as [our guide to AI-written and keyword-led content](/insights/thought-leadership-vs-keyword-stuffing) shows. ## Who should review the draft, and how many rounds should we expect? The expert first, for substance. The editor second, for polish. Expect two, sometimes three, rounds for the first few pieces, and fewer once the writer knows you. The expert reads the draft and asks: is this accurate against what we see with clients, did the writer use our angle, does it sound like us, and does every claim have an example, a reason or a source? If not, the piece goes back with a note that names the problem plainly, such as "the section on retentions needs to say what happens when the main contractor disputes the final account." Only then does the editor tidy it. Make the expert's part cheap: thirty minutes reading with a pen, saying what is wrong rather than rewriting. It gets much quicker after half a dozen pieces. In our experience, twelve excellent pieces a year, each answering a question a buyer asks, do more for whether a firm gets named than twenty-four that are nearly right. ## Should we hire a freelance writer, a specialist or a content agency? It depends on the piece, and most firms end up using two sources. The cost column is relative, and the revision rounds are what we typically see at the start, not a rule. | Option | Best use | Relative cost per piece | Revision rounds at first | Main risk | |---|---|---|---|---| | Generalist freelance writer | Steady answer pages and how-to guides | Usually the lowest | Two or three | Fluent but generic unless the brief and call are strong | | Sector specialist | Technical and regulated subjects | Higher | One or two | Brings opinions that may not match yours | | Content agency | Volume across formats, with project management | Usually the highest | Two, plus the agency's own reviews | Polished copy that sounds like its other clients | | AI tool plus an in-house editor | Turning recorded calls into drafts quickly | Low per draft; expert and editor time is the real cost | Two or more | Invented figures and sameness | A freelancer suits a firm with someone in-house to brief, chase and edit; an agency suits a firm that wants volume and has nobody to manage writers. Every option needs the same inputs: your voice guide, your briefs and your expert's review. If an agency says it needs none of that, you have your answer. ## How do you ghostwrite a founder's LinkedIn posts without them sounding fake? Use the same inputs as the articles, and have the founder approve every post. Posts sound fake when a writer invents opinions, anecdotes or enthusiasm the founder does not have. They sound like the founder when each one starts from something the founder said out loud that week. LinkedIn matters for AI search as well as for buyers. A [Semrush study of 89,000 LinkedIn URLs](https://www.semrush.com/blog/linkedin-ai-visibility-study/) (March 2026) found LinkedIn in about 11% of AI answers across ChatGPT search, Google AI Mode and Perplexity. In ChatGPT search and AI Mode, 59% of LinkedIn citations came from individual members rather than company pages, and around three quarters of the authors of cited posts had posted more than five times in the previous four weeks. ### Step 1: Record the raw material Fifteen to twenty minutes a week: what happened with clients, what the founder disagreed with, what they would tell a peer. ### Step 2: Draft from the transcript One post per idea, in the founder's own words, drafted by the writer or by an AI tool given the transcript and the voice guide. No invented stories. ### Step 3: Have the founder approve and answer The founder changes anything they would not say in a meeting, approves every post, and answers the comments personally. That is where conversations with buyers start. Founder LinkedIn is part of our [Reputation That Wins](/reputation-that-wins) service. If you keep meaning to post and never do, [this page describes the pattern](/sound-familiar/i-should-post-on-linkedin). ## How do we stop the voice drifting over time? Every three to six months, read the last six or eight pieces in one sitting. Voice drifts a little at a time towards the writer's habits or the last round of feedback, and nobody notices piece by piece. If it has become more cautious, more salesy or more general, update the voice guide and share it with the writer. ## What to do this month 1. Write the voice guide: a partner, two hours, three to five pages, including the sentences you would never publish. 2. Pick the one buying question you most want to be named for and write your two-sentence answer. 3. Brief one piece with the template above and record a twenty-minute call answering the question out loud. 4. Decide in writing which steps AI tools may do, and name the expert who reviews for substance and the editor who polishes. 5. Book a short weekly call to gather the founder's LinkedIn material. 6. When the piece is live, add its question to a fixed list of buyer questions you check every month. ## Frequently asked questions ### Can we use ChatGPT to write articles in our own voice, and will Google penalise it? You can use it to draft, and it gets closer to your voice when you give it the voice guide, the brief and the call transcript. Google says appropriate use of AI is not against its guidelines; what breaks its rules is producing many pages mainly to manipulate rankings, however they are written. Have the expert review every draft and check every figure it adds. ### How much does a B2B content writer cost per article in the UK? It varies widely with specialism, research, interviews and revision rounds, and we have not found a reliable, sourced UK benchmark to quote. Compare quotes like for like: a price per piece should cover reading your voice guide, a briefing call and two revision rounds, and very cheap content usually skips all three. Our own fees are on [what it costs](/what-it-costs). ### Should the article carry the partner's name if a freelancer wrote it? Yes, if the partner supplied the thinking and approved the draft; most professional writing works this way. Google's guidance on [creating helpful content](https://developers.google.com/search/docs/fundamentals/creating-helpful-content) asks whether it is "self-evident to your visitors who authored your content" and encourages accurate bylines. Use the partner's real name, role and a short biography, and keep them consistent on LinkedIn and in the trade press. ### Does the partner really need to be involved, or can marketing handle it? Marketing can run the process, chase the writer and polish the drafts. The partner needs to write or dictate the voice guide once, then spend twenty to thirty minutes per piece on the briefing call and the substance review. Without that, you are paying for content that says what everyone else says. ### How do we know whether the content is working? Keep a fixed list of 20 to 50 buyer questions and ask them in ChatGPT, Gemini, Claude and Copilot several times a month in a fresh session, because answers vary from run to run; count a win only when it repeats. Then listen in sales calls for prospects mentioning something you wrote. [Finding out whether ChatGPT recommends your firm](/insights/track-ai-recommending-your-business) shows how to make this routine. Outsourcing the writing is sensible; the thinking has to stay with you. Help with the voice guide, briefing, founder LinkedIn and review is part of [Reputation That Wins](/reputation-that-wins), or start with a free [Growth Review](/growth-review): 30 minutes with David Adams and three fixes in writing. ## Sources - [arXiv, Aggarwal and others, Princeton and IIT Delhi: source visibility in AI-generated answers, KDD 2024 (June 2024)](https://arxiv.org/abs/2311.09735) - [Google Search Central, Guide to optimizing for generative AI features on Google Search (July 2026)](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide) - [Edelman and LinkedIn, 2025 B2B Thought Leadership Impact Report (June 2025)](https://www.edelman.com/expertise/Business-Marketing/2025-b2b-thought-leadership-report) - [Google Search Central Blog, Google Search's guidance about AI-generated content (February 2023)](https://developers.google.com/search/blog/2023/02/google-search-and-ai-content) - [Google Search Central Blog, What web creators should know about our March 2024 core update and new spam policies (March 2024)](https://developers.google.com/search/blog/2024/03/core-update-spam-policies) - [Google Search Central, Creating helpful, reliable, people-first content (December 2025)](https://developers.google.com/search/docs/fundamentals/creating-helpful-content) - [Semrush, LinkedIn AI visibility study (March 2026)](https://www.semrush.com/blog/linkedin-ai-visibility-study/) --- Title: Can we wait a year before doing anything about AI search? URL: https://fortitudemedia.ai/insights/first-mover-advantage-ai Author: Ross Williams Published: 2026-04-22 Last reviewed: 2026-09-24 Topic: How buying is changing Short answer: You can, but it is rarely the cheapest choice. Buyers shortlist before they contact anyone, and the signals that get a firm named, such as coverage, reviews, links and what a model learned in training, take months to build, so early starters hold more of them. Live search is more open: ChatGPT search, Perplexity, Copilot and Google's AI features read pages when asked, so a clear new page can be cited within weeks. Start small this quarter with one specialism, a fortnightly piece, three case studies and a monthly check. Most founders we talk to are not sceptical about AI search. They accept that buyers now ask ChatGPT, Gemini, Copilot and Claude who to hire. Their question is about timing: is this urgent, or can it sit on next year's plan behind the new hire, the office move and the pricing review? Waiting is not neutral, because buyers settle on a favourite early. The [6sense 2025 Buyer Experience Report](https://6sense.com/newsroom/the-timeline-for-influencing-b2b-buyers-is-shrinking-insights-from-6senses-2025-buyer-experience-report/) (November 2025, more than 4,000 buyers in North America, Europe and Asia Pacific) found that buying groups had already ranked their preferred vendors before first contact in 94% of cases, and bought from that first choice 77% of the time. More of that early research now happens inside an assistant. In [G2's April 2026 survey of 1,076 software buyers](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html), 51% said they now start their research in an AI chatbot, up from 29% a year earlier. G2 surveyed software buyers, so the figure will differ for services, but the direction is clear. So a year of waiting is a year in which shortlists in your field are drawn up without you. Some of the advantage early firms build sticks and some does not. This guide separates the two, shows what an early lead looks like for a real client, and sets out the smallest sensible start, in hours a week. ## Key takeaways - You can wait a year, but buyers form shortlists before they contact anyone, so a year of absence is a year of shortlists drawn up without you. - Some advantages build slowly and lock in, such as what a model learned in training, press coverage, reviews and links. - Live search does not lock in: ChatGPT search, Perplexity, Copilot and Google's AI features fetch pages when a buyer asks, so a clear new page can be cited within weeks. - It is not too late in 2026: a firm that is absent from AI answers today can break in, as ID-Pal did between May and September 2026. - The smallest useful start is one specialism, one substantial piece a fortnight, three case studies with outcomes, consistent public facts and a fixed list of buyer questions checked every month. - That start costs a few hours a week of senior time, which in our view is a smaller bill than the one for catching up later. ## Is it too late to start with AI search in 2026? No. A rival that started last year holds some advantages you will have to work for, but most AI answers are still put together fresh each time a buyer asks. Google's AI Overviews and AI Mode, ChatGPT search, Perplexity and Microsoft Copilot all search the web and build the answer from what they find. Ahrefs found in [March 2026](https://ahrefs.com/blog/ai-overview-citations-top-10) that only 38% of AI Overview citations came from pages in Google's top ten, down from 76% in July 2025, because Google now runs several related searches for one question (it calls this query fan-out). Ranking still helps, but a page that answers the follow-up questions well can be cited without being the top result. Recency also helps a newcomer. [Seer Interactive's July 2026 study](https://www.seerinteractive.com/insights/study-content-recencys-impact-on-ai-visibility-in-2026) found that 75% of pages cited by AI assistants had been updated in the last 12 months. It covered four consumer and finance industries rather than B2B services, but it suggests a well-kept page from this year can compete with an older one. ## What advantage does starting early actually give? Mostly a head start on evidence. Assistants recommend firms they can see described, discussed and vouched for across the web, and that evidence takes time to build. Most of what assistants cite comes from other people. [Muck Rack's May 2026 analysis](https://www.globenewswire.com/news-release/2026/05/07/3290268/0/en/generative-pulse-earned-media-consistently-drives-ai-citations-holding-at-84.html) of more than 25 million links cited by ChatGPT, Claude and Gemini found that 84% were earned media and 0.3% were paid. An [SE Ranking study reported by Search Engine Journal in November 2025](https://www.searchenginejournal.com/new-data-top-factors-influencing-chatgpt-citations/561954/) found that the number of referring domains, the other sites linking to a page, was the strongest predictor of ChatGPT citations. Coverage and links both build up slowly. In our view that creates a loop: being named brings enquiries, enquiries become case studies, case studies earn coverage, and each is evidence for the next answer. No controlled study measures that loop, so treat it as our reading rather than a law. The wider change in buyer behaviour is covered in [how AI is changing the way B2B buyers choose suppliers](/insights/ai-search-change-your-industry); this guide is about timing. ## What locks in, and what does not? Signals that take months to build lock in; answers fetched live do not. The difference tells you where a late start hurts and where it barely matters. | Signal | Does it lock in? | Why | How fast it can change | |---|---|---|---| | What a model learned in training | Yes, until the next retrain | Memory reflects the web at training time | Months | | Press and trade coverage | Yes, it builds up | Earned media is most of what assistants cite | Months, one article at a time | | Reviews on G2, Clutch, Trustpilot and Google | Yes, it builds up | Review sites are cited in B2B answers | Months | | Links and mentions from other sites | Yes, it builds up | Referring domains predict ChatGPT citations | Months | | Live search (ChatGPT search, Perplexity, Copilot, Google's AI features) | No | Pages are fetched when a buyer asks | Weeks | | What your own site says | No | Live search reads the current version | Weeks, once crawled | The first four rows are where an early rival has an edge you cannot buy back quickly. In the same G2 survey, 45% of software buyers said citations of review sites were the most confidence-inspiring signal in an AI answer, and a rival with three years of reviews has something you cannot produce this quarter. Coverage builds the same way, as we set out in [how long PR takes to work and what a year should look like](/insights/compounding-effect-consistent-pr-12-months). The last two rows are where a late start matters least, so put your first effort there: clear pages that answer buyers' questions, consistent facts and crawlers let in, while the slower evidence builds behind them. ## What does an early lead look like in practice? It looks like a channel rivals are not using yet. Helm Club, a founder membership community running chaired peer groups for founders of growing UK businesses, is our clearest example. "1 in 5 new members this year came through an AI recommendation." As our case study puts it: "A channel that did not exist before now brings one in five new members: founders who ask ChatGPT, Gemini or Claude which peer group to join are told about Helm." The work behind it was specific: "We built the structured data AI engines read, then a comparison page for every question a founder asks when weighing up peer groups: Helm against YPO, Vistage and the rest." The timing lesson is in the result: "Every comparison page we have published ranks first for its question on Google and the AI engines. The pages not yet built are the ones where a competitor still writes the answer." That is the first-mover effect in two sentences. For each buyer question, somebody's page is the answer, and a competitor holds the question until someone writes a better one. The full story is in [the Helm Club case study](/work/helm-club). ## How long does it take to get recommended by ChatGPT from scratch? It depends on which part of ChatGPT is answering. When it searches the web live, a new page or a new article can show up in answers within weeks; when it answers from the model's own memory, the change arrives only when the model is retrained, which takes months. The same split applies to Perplexity, Copilot, Gemini, Claude and Google's AI features. Clarity fixes are the quickest wins, such as a service page that says plainly what you do and for whom, because live search reads them straight away. Being named consistently for a competitive specialism takes longer, because it depends on coverage, reviews and links. Two checks come first. OpenAI says a site must allow its crawler, OAI-SearchBot, to appear in [ChatGPT search](https://help.openai.com/en/articles/9237897-chatgpt-search), and Cloudflare has blocked AI crawlers by default on new domains since July 2025. Our guide to [how AI crawlers read your website](/insights/ai-crawlers-vs-google-spiders) covers the rest, and [how often you need to publish](/insights/how-often-publish-content-ai) covers the rhythm. ## Do AI models only learn about my firm when they are retrained? No. Retraining is one route, and it is slow. The other is live retrieval: when ChatGPT search, Perplexity, Copilot or Google's AI features look something up, they read your pages and other people's pages as they stand that day. Evidence you publish this year reaches live search once crawled, and joins the web the next models learn from. A firm that waits a year gives up a year of live answers and leaves the model's picture of its field to whoever published in the meantime. ## What does waiting cost while we are not named? Absence from AI answers sends no signal: you do not lose the pitch, you never hear about it. We set out how to put a figure on that for your own firm in [what it is costing you not to be recommended by ChatGPT](/insights/cost-of-inaction-losing-right-now). ## What is the smallest sensible start? One specialism, done properly, at a rhythm you can keep after the first quarter, which is where most firms stop. In our experience this is the minimum that moves anything. ### Step 1: Choose one specialism Pick the area where you are strongest and buyers ask questions, and agree it with your partners or board. ### Step 2: Rewrite that service page Say plainly what you do, for whom and with what evidence, and put the direct answer in the first paragraph. Kevin Indig's analysis of 1.2 million ChatGPT answers, [reported by Search Engine Land in February 2026](https://searchengineland.com/chatgpt-citations-content-study-469483), found 44% of citations came from the first 30% of a page. ### Step 3: Publish one substantial piece a fortnight Have it written, or at least shaped, by the person who does the work. Answer the questions buyers ask, including the follow-up questions they ask next. ### Step 4: Ask three clients for case studies with outcomes Use numbers wherever the client allows. Case studies are evidence others can cite. ### Step 5: Make your public facts agree Your website, Companies House, LinkedIn, Google Business Profile, directories and review sites should give the same name, description and services. Our guide to [why ChatGPT describes a firm wrongly](/insights/ai-conflicting-information-online) shows how to find and fix the conflicts. ### Step 6: Check a fixed list of buyer questions every month Write down 20 to 50 questions your buyers ask, run each several times a month in a fresh session in ChatGPT, Gemini and Copilot, and count a win only when it repeats, because answers vary from run to run. The method is in [how to find out whether ChatGPT recommends your firm](/insights/track-ai-recommending-your-business). ## What is the minimum budget for the smallest start? Mostly time. Done in-house, the minimum start takes a few hours a week of senior time plus some editing; with outside help, you swap some of those hours for a fee. Spend the first hours on what live search reads within weeks. For example, a firm running the six steps in-house might plan as follows. The figures are round and illustrative, not measured. | Task | Who | Rough time | |---|---|---| | Choose the specialism and rewrite one service page | Founder or partner, plus an editor | Two to three days, once | | One substantial piece a fortnight | The expert for an interview or draft, an editor to finish | About half a day a fortnight | | Three case studies with outcomes | Client lead, plus an editor | A few hours each, over the first quarter | | Consistent public facts | Whoever manages the website and listings | A day, once, then a short check each quarter | | Monthly check of buyer questions | Anyone careful, with a spreadsheet | Two to three hours a month | There are three ways to carry that load. | Route | What you get | What it costs | |---|---|---| | Do it yourself | The six steps above, at your own pace | Senior time, as in the table above | | [Paid Discovery](/discovery), then your own team | One week, every finding under five lenses, ranked, with fixes written so your own team can ship them | A one-off fee, listed on [what it costs](/what-it-costs), then your team's time | | An ongoing programme | We run the publishing, coverage and site work through [Reputation That Wins](/reputation-that-wins) and [Website That Sells](/website-that-sells) | A monthly fee, listed on [what it costs](/what-it-costs) | For the ranges other UK providers charge, see [how much AI search optimisation costs in the UK](/insights/ai-search-optimisation-cost-uk). ## What to do this month 1. Pick the one specialism you will build first and agree it with your partners or board. 2. Rewrite that service page in plain English: what you do, who for, why you and with what evidence, with the answer in the first paragraph. 3. Check that your robots.txt file and any Cloudflare setting let in OAI-SearchBot, Claude-SearchBot, PerplexityBot and Bingbot. 4. Put two publishing dates a month in the diary for the next six months, with a named author for each. 5. Email your three best clients this week and ask for a case study with numbers. 6. Write your list of 20 to 50 buyer questions and run the first check, so you have a baseline to measure against. ## Frequently asked questions ### Our competitors are not doing anything. Does that mean we can wait? Only if you are sure, and you usually cannot be. Firms doing this well do not announce it; the first sign is often a rival appearing in answers you expected to be in. If nobody in your sector has started, being first is the cheapest position to take, because every buyer question is still open. ### If a competitor started a year ago, is it too late for us? No. ID-Pal shows a firm going from absent to named. At the start, "AI could describe ID-Pal. It never recommended it." The result: "UK buyer questions where AI names ID-Pal: none in May, two in August, four in September 2026." The work included "Five industry pages rebuilt: fund administration, wealth management, legal, accountancy and estate agency." The pattern is plain: "Where a page has been rebuilt, ID-Pal appears. Where it has not yet, it does not, which tells us exactly what to build next." Be narrower than the rival, and pick the questions where you can become the best answer. The detail is in [the ID-Pal case study](/work/id-pal). ### Once a competitor is recommended by ChatGPT, can we ever replace them? Yes, or at least join them. Assistants usually name several firms, answers vary from run to run, and live search reads the web again each time, so a named rival holds a strong position rather than a permanent one. The way in is through the questions they answer badly or not at all: a narrower specialism, a sector they do not serve, or a follow-up question nobody has written a good page for. ### Is this just SEO with a new name? Partly. Google's [guide to its generative AI features](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide), last updated in July 2026, says optimising for them is still SEO, and good search work is the foundation. Ranking helps without guaranteeing a mention, though: the Ahrefs figures above show most AI Overview citations now come from outside the top ten. Being named also depends on evidence from others, as we explain in [why ranking first on Google does not mean ChatGPT mentions you](/insights/google-rankings-wont-protect-you). ### Can we do this ourselves or do we need outside help? You can do the minimum start yourself if a partner will write and someone will keep the rhythm. Firms tend to need help with keeping going, winning coverage where they have no relationships, and structuring the site so assistants read it well. Do not let waiting for help become another reason to wait. Waiting a year is a decision, and usually the most expensive one available. If you would like to know where you stand before you decide, [book a free Growth Review](/growth-review): 30 minutes with David Adams, and three fixes in writing afterwards. ## Sources - [6sense, 2025 Buyer Experience Report (November 2025)](https://6sense.com/newsroom/the-timeline-for-influencing-b2b-buyers-is-shrinking-insights-from-6senses-2025-buyer-experience-report/) - [G2, The Answer Economy: half of B2B software buyers now start with AI chatbots (April 2026)](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html) - [Ahrefs, Only 38% of AI Overview citations come from the top ten (March 2026)](https://ahrefs.com/blog/ai-overview-citations-top-10) - [Seer Interactive, Content recency's impact on AI visibility in 2026 (July 2026)](https://www.seerinteractive.com/insights/study-content-recencys-impact-on-ai-visibility-in-2026) - [Muck Rack, What is AI reading? Earned media drives 84% of AI citations (May 2026)](https://www.globenewswire.com/news-release/2026/05/07/3290268/0/en/generative-pulse-earned-media-consistently-drives-ai-citations-holding-at-84.html) - [Search Engine Journal, SE Ranking data on the top factors influencing ChatGPT citations (November 2025)](https://www.searchenginejournal.com/new-data-top-factors-influencing-chatgpt-citations/561954/) - [Search Engine Land, Kevin Indig's study of 1.2 million ChatGPT citations (February 2026)](https://searchengineland.com/chatgpt-citations-content-study-469483) - [Google Search Central, Optimizing your website for generative AI features on Google Search (July 2026)](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide) --- Title: Digital PR or traditional PR: which gets a B2B firm named by ChatGPT? URL: https://fortitudemedia.ai/insights/digital-pr-vs-traditional-pr Author: Ross Williams Published: 2026-04-22 Last reviewed: 2026-09-24 Topic: Press, reviews and being trusted Short answer: Both can help, but only coverage that ends up as readable text online counts with AI assistants. A broadcast slot with no written trace, a print-only feature or a site that blocks AI crawlers leaves ChatGPT little to use, although OpenAI licenses some paywalled UK titles, including The Times and the FT. For most B2B firms, steady coverage in open trade titles buyers read does more than one national headline, and every broadcast should be turned into text. Most founders who have paid a PR agency know the feeling. The coverage report arrives with a quote in a national paper, a slot on regional radio and a place on a list of firms to watch, all adding up to an audience figure in the millions. The board is pleased for a week. Then someone asks what it did for the pipeline, and nobody can say. That question is sharper now because of where buyers start. In [G2's April 2026 survey of 1,076 software buyers](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html), 51% said they now begin their research in an AI chatbot, up from 29% a year earlier. Those assistants build their answers mostly from other people's words: in [Muck Rack's May 2026 analysis](https://www.globenewswire.com/news-release/2026/05/07/3290268/0/en/generative-pulse-earned-media-consistently-drives-ai-citations-holding-at-84.html) of more than 25 million links cited by ChatGPT, Claude and Gemini, earned media made up 84% and paid or advertorial content 0.3%. But they can only use coverage they can read. A radio interview nobody transcribed, or a print feature that never went online, leaves them nothing. So the useful question is which coverage keeps working after the week it appears, and which kind of PR produces more of it. This guide answers that from the side of the desk that pays the invoice. ## Key takeaways - AI assistants can only use coverage they can read as text online, so where a piece lives matters as much as the prestige of the title. - Broadcast interviews and print-only features rarely reach AI answers unless someone publishes a written version, such as a transcript, a captioned video or a summary article. - Paywalls are not a blanket barrier: OpenAI licenses content from News Corp titles, including The Times, and from the Financial Times, so ChatGPT can draw on some paywalled UK coverage. - For most B2B firms, steady coverage in open trade titles that buyers read does more for AI answers than one national headline. - Traditional PR still earns its fee with investors, senior hires and large clients, and should be judged on those goals. - Judge any PR report by named mentions in relevant, readable publications and by whether AI answers about your firm improve, never by advertising value equivalents. ## What is the difference between digital PR and traditional PR? Traditional PR aims for reach and prestige in national titles, print and broadcast; digital PR aims for coverage that lives online and keeps being found, in trade websites, contributed articles, data stories and podcasts with show notes. The overlap is large, and for AI answers the test is simple: once the week is over, is there a page a machine can read that names your firm and says what you do? | | Traditional PR | Digital PR | |---|---|---| | Main aim | Reach and prestige with a broad audience | Lasting, findable coverage in front of a specific audience | | Typical outputs | National and regional press, broadcast interviews, business magazines, set-piece announcements | Trade and business websites, expert comment, contributed articles, data stories, podcasts with show notes | | How it is usually bought | Monthly retainer or campaign fee | Monthly retainer, campaign fee, sometimes a fee per placement | | How it is usually measured | Audience reach, column inches, sometimes advertising value equivalents | Named mentions, links, referral visits, enquiries, answers from AI assistants | | Effect on AI answers | Strong when the piece is online and readable; little or none when it is broadcast only, print only or blocked | Usually direct, because the coverage is open text on sites the assistants can reach | Advertising value equivalents deserve a warning: they price column inches as if they were adverts, which says nothing about whether anyone changed their mind. The industry's own measurement body, AMEC, runs a campaign to eradicate them ([AMEC, Say No to AVEs](https://amecorg.com/say-no-to-aves/)) and promotes its [Barcelona Principles 3.0](https://amecorg.com/barcelona-principles-3-0/) instead. Most national titles publish online and many trade magazines are now mainly websites, so one story can be both. Which kinds of PR win work beyond AI answers is covered in [which kinds of PR actually win work for a B2B firm](/insights/pr-for-b2b-what-works). ## What can AI assistants actually read? They read text on web pages they are allowed to reach, plus content their makers have licensed. Everything else has to be turned into text first. The assistants draw on what their models learned from the public web in training and, for many questions, on a live web search. Both depend on pages software can reach and read, which rules out or weakens several kinds of traditional coverage. - **Broadcast.** A television or radio interview is invisible to an assistant unless someone publishes a transcript, a write-up or a video page with captions and a proper description. - **Print only.** A feature that never appeared online might as well not exist for AI answers. - **Paywalled articles.** This is less clear-cut than it was. OpenAI's May 2024 agreement with News Corp gives it access to current and archived content from titles including The Times, The Sunday Times and The Sun, with permission to display it in response to user questions ([OpenAI](https://openai.com/index/news-corp-and-openai-sign-landmark-multi-year-global-partnership/)). The Financial Times agreed in April 2024 that ChatGPT users would see attributed summaries, quotes and links to FT journalism ([OpenAI](https://openai.com/index/content-partnership-with-financial-times/)). A paywalled title without such a deal may show an assistant little more than the headline and opening lines. Deals are made assistant by assistant, so coverage ChatGPT can use may be closed to Claude or Perplexity. - **Blocked sites.** Some publishers tell AI crawlers, the programs that fetch pages for AI companies, to keep out. OpenAI says a site must allow its OAI-SearchBot crawler to appear in ChatGPT search ([OpenAI Help Center](https://help.openai.com/en/articles/9237897-chatgpt-search)), and Cloudflare has blocked AI crawlers by default on new domains since July 2025. Coverage on a blocked site still helps with human readers and with Google, but contributes less to what assistants can cite. None of this makes national coverage worthless; it means checking where each piece lives online and how much of it can be read. How crawlers reach websites is explained in [how AI crawlers read your website](/insights/ai-crawlers-vs-google-spiders). ## Where does traditional PR still earn its fee? Where the audience is people you need to impress directly. A profile in a national business section helps with investors, senior hires and large clients who read it; a respected broadcast slot gives your managing director confidence and a clip; a big announcement gives your sales team something to send. It has an indirect value too: a national piece can lead to trade coverage, speaking invitations and requests for comment, and those follow-on mentions are often the ones the assistants find. The trouble starts when it is sold on the promise of getting you named in AI answers, then measured by an audience figure that says nothing about that. If your aim is reputation with a specific group of people, judge it on that. ## Why does steady trade coverage often do more for AI answers? Because it is specific, it matches the questions buyers ask, it is usually open to read, and it adds up. For most B2B firms, a mention most months in the titles their buyers read does more than an occasional national headline. ### It is specific A trade article says what you did, for which kind of client and why it worked, which is the sentence an assistant needs. A national piece often mentions you in passing. ### It matches the buyer's question Buyers ask assistants narrow questions: which firms specialise in this, in this region, for this kind of business. Sector titles are organised around exactly those topics, so their pages turn up in the searches the assistants run. The assistants do not only cite big names: Muck Rack found journalism citations spread across more than 20,000 distinct outlets. ### It is usually open Many trade titles earn their money from advertising and events rather than subscriptions, so they publish openly and their archives stay readable for years. ### It adds up One article is a mention. Ten articles over a year, each naming your firm in the same context, form a pattern, and agreement across independent sources is how assistants decide what to believe. Recency matters too: Muck Rack found more than half of journalism citations came from articles published in the past 12 months, with citations dropping sharply after six months. A steady rhythm keeps fresh coverage in view. How that builds over a year is covered in [how long PR takes to work and what a year should look like](/insights/compounding-effect-consistent-pr-12-months). ## Is coverage enough on its own? No. Coverage tells the assistants that others vouch for you, but they still need a clear page of yours that answers the buyer's question. Without it, coverage earns you a description rather than a recommendation. Our work with [ID-Pal](/work/id-pal), an identity verification firm, shows the page side of the job. Before we started, "AI could describe ID-Pal. It never recommended it." The fix began with "About four hours of structured data work declaring NorthRow part of ID-Pal on both domains. Live in week one, with no copy changes and no risk to the redirect." Then "Five industry pages rebuilt: fund administration, wealth management, legal, accountancy and estate agency." The result: "UK buyer questions where AI names ID-Pal: none in May, two in August, four in September 2026." And the pattern is clear: "Where a page has been rebuilt, ID-Pal appears. Where it has not yet, it does not, which tells us exactly what to build next." Good PR points buyers and assistants at pages like those. Why independent coverage and your own pages carry different weight is set out in [whether ChatGPT trusts press coverage more than your website](/insights/earned-owned-paid-authority-signals). ## How do you get both kinds of coverage from one piece of news? Use one good story to build prestige with people and leave a permanent written record the assistants can read. You do not have to choose. ### Step 1: Start with something worth covering Journalists cover news, data and strong opinions, so a client survey or a clear view on a regulatory change will get further than a company announcement. Our guide to [turning your own data into trade press coverage](/insights/leverage-data-research-press-coverage) shows how to build one. ### Step 2: Pitch trade titles alongside the nationals Offer trade titles their own angle, more detail and a named expert. They are often more likely to run the story in full and to name your firm clearly. [How to get trade journalists to call you for comment](/insights/building-relationships-journalists-editors) covers the relationships behind this. ### Step 3: Make sure every broadcast leaves a written trace If your managing director appears on radio, television, a podcast or a webinar, create at least three written versions: - **A YouTube upload with corrected captions and a full description.** With the broadcaster's or host's permission, upload the clip, correct the automatic captions (they often mangle names and technical terms), and write a description that names your firm, the guest and the main points. - **The host's transcript page.** Ask whether the programme publishes a transcript or episode page, and send the correct spelling of names and terms before it goes up. - **A LinkedIn article summarising the interview.** Have the guest write up the key points in their own words, with a link to the recording. Articles made up 50% to 66% of LinkedIn citations in AI answers in [Semrush's March 2026 study](https://www.semrush.com/blog/linkedin-ai-visibility-study/). Our guide to [turning webinars and podcasts into written content](/insights/video-podcast-transcripts-ai-content) explains the rest. ### Step 4: Publish your own version Put the full story on your website, with the data, the context and links to the coverage, so assistants have one complete source the coverage points back to. ### Step 5: Keep the rhythm going One campaign is a moment; a steady rhythm of comment, contributed articles and occasional data stories is a reputation. Plan for something most months. ## How do you judge what your PR is doing? By what it leaves behind and whether that changes how buyers find you. Audience figures tell you how big a room was, not whether anyone remembered your name. A useful monthly report answers four questions: - Which publications named us this month, and are they ones our buyers read? - Is the coverage open to read online, and does it describe what we do accurately? - Did it link to our site, and did any enquiries follow? - When we ask ChatGPT, Gemini, Claude, Copilot and Perplexity the questions our buyers ask, are we named more often, and described better, than three months ago? For the last question, use a fixed list of 20 to 50 buyer questions, each asked several times a month in a fresh chat, and count a win only when it repeats, as set out in [how to find out whether ChatGPT is recommending your firm](/insights/track-ai-recommending-your-business). If your agency cannot answer it, ask them to start or run the check yourself. ## What to do this month 1. Pull your last twelve months of coverage and mark each piece as open online, paywalled, broadcast only or print only. 2. List the five trade titles your buyers read most, note which have covered you, and check whether their robots.txt files let AI crawlers in. 3. For each broadcast or podcast appearance in the last year, create the written trace: a captioned video, a transcript request and a LinkedIn article. 4. Pick one story you could tell with your own data or a clear opinion, and plan a trade version of it. 5. Ask your PR agency to add a fixed set of buyer questions, and the AI answers to them, to its monthly report. ## Frequently asked questions ### Should we stop doing national press? No, but be clear why you are doing it. National coverage helps with investors, senior recruits and large clients, and often leads to trade coverage. For AI answers it is a bonus rather than the plan: make sure it exists online in an open form, and put most of your effort into regular trade coverage. ### Which UK newspapers can ChatGPT actually read? Those that let OpenAI's crawlers in, plus those that license content to OpenAI. The News Corp deal covers The Times, The Sunday Times and The Sun, and the FT has its own agreement. For any other title, check its robots.txt file for OAI-SearchBot, which ChatGPT search needs, and GPTBot, which collects training data; a title that blocks both leaves ChatGPT little to read beyond what other sites say about the article. ### Is a TV or radio interview worth doing if nobody publishes a transcript? Yes for the people who see or hear it, but only half done for AI answers. Assistants read text, so an interview with no written trace adds nothing they can cite. Do the interview, then create the trace yourself: a captioned video with a full description where you have permission, a LinkedIn article from the guest, and a summary page on your site. ### Do press releases on news wire services help us show up in ChatGPT? A little, and mainly in answers about industry trends. Muck Rack found press releases appear almost exclusively in industry trend responses, at a rate 3.5 times higher than in best-of queries, the "who should I hire" questions that matter most. A wire release is also your own words, so it adds less than an editor choosing to cover the same news. Use the wire for genuine announcements, and pitch the story to trade titles as well. ### Is digital PR worth it for a B2B firm with under £5m turnover? Usually yes, if it is narrow. A firm of that size does not need national coverage; it needs its founder quoted and published, most months, in the three to five titles its buyers read, plus one data story a year. Much of that can be done in-house with a few hours a month of the founder's time, and outside help earns its keep when it saves that time and opens doors to editors. ### What does a digital PR agency cost in the UK? It varies widely with scope, seniority and how many titles and stories are involved. Most agencies charge a monthly retainer or a campaign fee, and a few charge per placement, which can reward volume over relevance. Whatever the model, ask what the fee buys each month in named outputs, and how the agency will show its effect on AI answers. Our own prices are on [what it costs](/what-it-costs). ### How long before new coverage changes what assistants say? Assistants that search the web live, such as ChatGPT search, Perplexity, Copilot and Google's AI features, can reflect a new article within weeks. Answers drawn from a model's own memory change only when the model is retrained, which takes months. Steady coverage over several months moves answers more reliably than one big piece. If you want PR judged by who names you, in the trade press and in AI answers, [Reputation That Wins](/reputation-that-wins) is built around that. A free [Growth Review](/growth-review) shows you what the assistants say about your firm today. ## Sources - [Muck Rack, Generative Pulse: earned media consistently drives AI citations (May 2026)](https://www.globenewswire.com/news-release/2026/05/07/3290268/0/en/generative-pulse-earned-media-consistently-drives-ai-citations-holding-at-84.html) - [G2, New G2 research: half of B2B software buyers now start their research with AI chatbots (April 2026)](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html) - [OpenAI, News Corp and OpenAI sign landmark multi-year global partnership (May 2024)](https://openai.com/index/news-corp-and-openai-sign-landmark-multi-year-global-partnership/) - [OpenAI, Content partnership with the Financial Times (April 2024)](https://openai.com/index/content-partnership-with-financial-times/) - [OpenAI Help Center, ChatGPT search (September 2026)](https://help.openai.com/en/articles/9237897-chatgpt-search) - [AMEC, Say No to AVEs (November 2025)](https://amecorg.com/say-no-to-aves/) - [AMEC, Barcelona Principles 3.0 (November 2020)](https://amecorg.com/barcelona-principles-3-0/) - [Semrush, LinkedIn AI visibility study (March 2026)](https://www.semrush.com/blog/linkedin-ai-visibility-study/) --- Title: Do we need schema markup for AI assistants to understand our site? URL: https://fortitudemedia.ai/insights/structured-data-schema-markup-ai Author: David Adams Published: 2026-04-22 Last reviewed: 2026-09-24 Topic: Getting named by AI assistants Short answer: Not strictly, but accurate markup helps. Schema markup is hidden code that states your firm's facts plainly: name, company number, services, people, and who wrote each article and when. Google says its AI features need no special markup and stopped showing FAQ rich results on 7 May 2026; Microsoft says schema helps its AI systems understand content; OpenAI and Anthropic have not said whether they read it. Add Organization, Person, Article and Service markup, keep it matching the page, and do not treat it as a ranking trick. Your web developer has mentioned "schema" in a quote, your agency says the site "needs structured data for AI", and nobody has explained what either means or whether it will make ChatGPT more likely to name your firm. The useful answer is smaller than the sales pitch. Structured data is a short block of code, invisible to visitors, that tells machines plainly what a page is: this is a company, its name is X, it is based in Leeds, it provides these services, this article was written by this person on this date. Schema markup is the shared vocabulary for writing it, published at [schema.org](https://schema.org) and backed by the major search engines. What is on record is narrower than most agencies claim. Google's [guide to AI features](https://developers.google.com/search/docs/appearance/ai-features) (updated December 2025) says there is "no special schema.org structured data that you need to add" to appear in AI Overviews or AI Mode. Microsoft's Bing team says schema helps "search engines and AI systems understand your content". OpenAI and Anthropic have not said whether they read it. And in an [SE Ranking study of 216,524 pages, reported by Search Engine Journal](https://www.searchenginejournal.com/new-data-top-factors-influencing-chatgpt-citations/561954/) in November 2025, pages with FAQ schema averaged 3.6 ChatGPT citations against 4.2 for pages without it. So schema is not a ranking trick. It is how you state your facts in a form machines cannot misread, and that is still worth an afternoon. ## Key takeaways - Schema markup is hidden code that states your firm's facts plainly, so search engines and AI assistants do not have to infer them from the prose. - Google says its AI features need no special markup, Microsoft says schema helps its AI systems understand content, and OpenAI and Anthropic have not said whether they read it. - A B2B firm needs Organization, Person, Article and Service markup, spelled exactly as schema.org spells them. - The sameAs list, pointing to your LinkedIn page, Companies House record and other profiles, is how machines join your website to facts about you elsewhere. - Google stopped showing FAQ rich results on 7 May 2026 and HowTo rich results in 2023, so FAQ markup no longer earns any display in Google. - Markup must match the visible page, and it must sit in the page's HTML, because the main AI crawlers do not run JavaScript. ## What is schema markup, in plain English? It is information written in a fixed, machine-readable format rather than in sentences, usually in a format called JSON-LD in the page's code. Compare two ways of saying the same thing. In prose: "Founded in 2015, our team of thirty has helped over two hundred manufacturing businesses improve their cash position." In structured data: founding date 2015, number of employees 30, area of expertise manufacturing. The prose is better for a human. The structured version tells a machine at once that these are three separate facts and what each one is. ## Does ChatGPT actually read schema markup? Nobody outside OpenAI knows, because OpenAI has not said. What is on record comes from Google and Microsoft. | Company | What it has said about structured data | When | |---|---|---| | Google (AI Overviews, AI Mode) | No special schema is needed for its AI features; markup should match the visible text; it still helps pages qualify for rich results | December 2025 and July 2026 | | Microsoft (Bing, Copilot) | "Schema is a type of code that helps search engines and AI systems understand your content"; Copilot is powered by Bing's index | October 2025 | | OpenAI (ChatGPT) | Nothing public on schema; a site must allow OpenAI's OAI-SearchBot crawler to appear in ChatGPT search | Not stated | | Anthropic (Claude) | Nothing public on schema | Not stated | Google's [AI optimisation guide](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide), last updated in July 2026, calls "overfocusing on structured data" a myth, then adds that "it's a good idea to continue using it as part of your overall SEO strategy". Microsoft's statement comes from a [Bing product manager's guide to AI search answers](https://about.ads.microsoft.com/en/blog/post/october-2025/optimizing-your-content-for-inclusion-in-ai-search-answers) (October 2025). Two practical points follow. ChatGPT search has drawn on Bing: in a small [Seer Interactive test](https://www.seerinteractive.com/insights/87-percent-of-searchgpt-citations-match-bings-top-results) (February 2025), 87% of its citations matched Bing's top results. Markup that helps Bing read you may therefore help ChatGPT indirectly, though that is our inference, not something OpenAI has confirmed. And how the markup reaches the page matters. [Vercel and MERJ found](https://vercel.com/blog/the-rise-of-the-ai-crawler) (December 2024) that GPTBot, ClaudeBot and PerplexityBot fetch pages but do not run JavaScript. Markup injected by a tag manager after the page loads can be read by Googlebot and missed by the AI crawlers, so put it in the HTML the server sends. Our guide to [how AI crawlers read your website](/insights/ai-crawlers-vs-google-spiders) explains what they see. ## Why does structured data help, if nobody requires it? Because it removes doubt about the basic facts, and doubt is what makes an assistant hedge, guess or leave you out. Prose is ambiguous: "an established firm" could mean five years or fifty. Markup states the founding date, the company number and the exact services once, in a standard format. It also makes contradictions visible. If your markup says you serve software companies and the page talks about retail, trust in both falls, which is why Google asks for markup that matches the visible text. [Why ChatGPT describes firms wrongly](/insights/ai-conflicting-information-online) shows what inconsistency costs. When we reviewed [TCC Group](/work/tcc-group), part of what failed was "machine-readable data that declared a name, a logo and nothing else." What markup will not do is get thin content named, raise your Google ranking on its own or fix a site AI crawlers cannot reach. Content, mentions on other sites and clear positioning decide whether you are named; markup makes them easier to read. ## Which schema types does a B2B firm need? Four types do most of the work: Organization, Person, Article and Service. The rest apply only where they describe you. | Type (exact spelling) | Where it goes | What it states | Priority | |---|---|---|---| | Organization | Homepage, referenced from every page | Names, logo, address, company number, sameAs, founders | First | | Person | Each author page, and inside each Article | Name, jobTitle, worksFor, sameAs | First | | Article | Each guide or insight | headline, author, datePublished, dateModified | Second | | Service | Each service page | Name, description, provider, areaServed | Second | | BreadcrumbList | Every page below the homepage | Where the page sits in the site | Third | | SoftwareApplication or Product | Product pages, if you sell software | Name, category, platform | Only if it applies | | A LocalBusiness subtype | Office pages, only if clients visit | Address, opening hours | Only if it applies | | FAQPage | Pages with visible questions and answers | Each question and answer | Optional; no Google display since 7 May 2026 | Spelling matters. Schema.org uses American spellings, so the type is Organization with a z; a hand-typed "Organisation" is not a schema.org type and validators will flag it. SoftwareApplication and BreadcrumbList are single words with two capitals each. ### Organization: the block every firm needs Here is the Organization block on this site's homepage, trimmed (the full version also lists contact details, our three services and all three founders): The @id gives the organisation a permanent address that every other block can point to, so Article and Person blocks say "this organisation" without repeating its details. ### sameAs: how machines join your site to facts elsewhere Schema.org defines [sameAs](https://schema.org/sameAs) as the "URL of a reference Web page that unambiguously indicates the item's identity". Google's [Organization markup documentation](https://developers.google.com/search/docs/appearance/structured-data/organization) says the markup helps it "disambiguate your organization in search results". List only profiles that are yours and kept current: your LinkedIn company page, your Companies House record, a Wikidata entry if one exists and is accurate, Crunchbase, professional body or regulator listings, and review profiles such as Clutch or G2. If two of them disagree on your name or what you do, fix them first; [why ChatGPT sounds unsure about your firm](/insights/ai-cross-references-business-sources) explains why assistants check. ### Person: markup for your authors Person markup ties each guide to a real, checkable person, which supports the signals covered in [what E-E-A-T is and why AI cares about it](/insights/eeat-why-ai-cares-more). It needs a name, jobTitle, worksFor and sameAs. This is the author block for David Adams on the guides he writes: The blocks for [Ashley Kenerson](/team/ashley-kenerson) and [Ross Williams](/team/ross-williams) follow the same pattern with their own jobTitle and LinkedIn profile. A founder who is a company director can also add their Companies House officer page to sameAs, which ties the person to the company on a public register. ### Article: use the right date fields Article markup needs a headline, an author, datePublished and dateModified. There is no "last reviewed" field on Article: [lastReviewed](https://schema.org/lastReviewed) belongs to WebPage. Change dateModified only when the content changes; our guide on [how old is too old for AI assistants](/insights/content-freshness-recency-ai) covers why. This guide's own Article block, simplified: ### Service, not ProfessionalService Add one Service block to each service page, with a name, a description, the provider (your Organization's @id) and where it is offered. Avoid ProfessionalService: schema.org's own page says it ["was deprecated due to confusion with Service"](https://schema.org/ProfessionalService). ### Organization or LocalBusiness if clients never visit? Use Organization. Google recommends a LocalBusiness subtype when "your site is about a local business, for example a restaurant or a physical store". A consultancy or software firm whose clients never walk in is not that. If clients do visit your offices, a specific subtype such as LegalService or AccountingService on each office page is accurate, alongside the Organization block. ## Is FAQ schema still worth adding now Google has dropped FAQ rich results? Not for Google. [Search Engine Journal reported](https://www.searchenginejournal.com/google-drops-faq-rich-results-from-search/574429/) that FAQ rich results stopped appearing on 7 May 2026, and Google removed HowTo rich results in September 2023, as its [Search Central blog](https://developers.google.com/search/blog/2023/08/howto-faq-changes) recorded. FAQPage is still a valid type and existing markup does no harm, so there is no need to strip it out. This site still carries FAQPage and HowTo markup on its guides; it costs nothing to keep, and we do not expect it to earn anything in Google. The SE Ranking figures above suggest FAQ markup does nothing for ChatGPT citations either. What matters is the visible questions and answers, which give readers and assistants a clean passage to quote. ## Can schema markup get my company a Google knowledge panel? Not on its own. Google's [knowledge panel help page](https://support.google.com/knowledgepanel/answer/9163198?hl=en) says panels "are automatically generated, and information that appears in a knowledge panel comes from various sources across the web". Organization markup can shape what a panel shows: Google says some properties affect "which logo is shown in Search results and your knowledge panel". Whether you get one depends on Google recognising your firm as a distinct entity from sources beyond your site, such as registers, press and directories; a consistent sameAs list helps join those up. If a panel appears, claim it so you can suggest corrections. ## What does this look like in practice? The clearest case we have is [ID-Pal](/work/id-pal), an identity verification firm that had bought NorthRow, a UK business verification platform, and folded it into id-pal.com. "The redirect from northrow.com was technically correct, but nothing on either site told Google or the AI engines that NorthRow and ID-Pal were the same company." The fix: "About four hours of structured data work declaring NorthRow part of ID-Pal on both domains. Live in week one, with no copy changes and no risk to the redirect." Industry pages were rebuilt alongside it, so the markup was one part of the work. The case study reports: "AI answers still recommending NorthRow as a rival: six in May, one in September 2026." And "Claude names it for UK fund administrators, on a page rebuilt in August, and links NorthRow to ID-Pal in the same answer." ## How do we add structured data without a developer? Most firms can do a first pass in an afternoon with the tools already in their content system. You need a developer only if your site has no way to add code. ### Step 1: Check what you already have View your homepage's source (not the inspector, which shows the page after scripts run) and search for "ld+json". Paste the address into the Schema Markup Validator at validator.schema.org, which reads every type, and Google's Rich Results Test, which shows the types that can earn a Google result. ### Step 2: Write the Organization block Use a schema generator or copy the structure above. Check every field against the visible site, and spell every type the schema.org way. ### Step 3: Add it to the site's HTML WordPress plugins such as Yoast, Rank Math and Schema Pro need no code; Webflow, HubSpot and most others let you paste code into a page's head. Then view the source again to confirm the block is there. If your system makes this hard, see [how your website platform affects what ChatGPT can read](/insights/cms-choice-affects-ai-visibility). ### Step 4: Add Person, Article and Service markup Give each author page a Person block, add Article markup to your five most important guides, then one Service block per service page and breadcrumbs below the homepage. ### Step 5: Test it and keep it true Run each page through the validator and fix every error, since an invalid block is ignored. Then add a markup check to your quarterly content review: if the markup says two hundred clients and the page now says three hundred, the contradiction costs you. Our [technical checklist for AI crawlers and search](/insights/technical-checklist-ai-ready-website) lists the other checks to run at the same time. ## What to do this month 1. View the source of your homepage, search for "ld+json" and paste what you find into the Schema Markup Validator. 2. Write down the ten facts you want every assistant to get right: name, company number, location, founding date, size, services and key people. 3. Correct the Organization block so it states those facts exactly as the visible page does, with every type spelled the schema.org way. 4. Add a sameAs list with your LinkedIn page, Companies House record and any accurate directory, review or professional body profiles. 5. Add Person markup to each author page and Article markup, with author, datePublished and dateModified, to your five most important guides. ## Frequently asked questions ### What schema should a B2B consultancy put on its homepage? An Organization block with the legal and trading names, logo, description, address, company number, contact point, founders and a sameAs list of your real profiles, plus a WebSite block. Do not use ProfessionalService, which schema.org has deprecated, or LocalBusiness unless clients visit your offices. Services belong on their own pages with a Service block each. ### Which type should we add first? Organization, on the homepage. Then Person for each author and Article for each guide, then Service for each service line. Add SoftwareApplication or Product only if you sell software or a product. FAQPage comes last, if at all, since it no longer earns a Google display. ### Will it improve our Google rankings? Not directly; structured data is not a ranking factor in itself. Google says it helps pages qualify for rich results and helps it tell your organisation apart from others. Think of it as making sure you are read correctly, not as a way to be read first. ### Does markup added by a plugin or tag manager count? Plugin markup that appears in the page source counts. Markup injected by a tag manager after the page loads can be read by Google, which runs JavaScript, but the main AI crawlers do not run it and may never see it. Check the page source to be sure. ### Does every page need structured data? Every page should reference your Organization, and every guide should carry Article markup. Beyond that, add only the types that describe what the page is. Padding a page with types that do not apply makes the site less coherent to the systems reading it. Structured data is never urgent and always worth doing, and once it is in place it makes everything else on the site easier to read. Sites built through [Website That Sells](/website-that-sells) include full structured data on every page; to find out first what AI assistants say about your firm today, [book a free Growth Review](/growth-review). ## Sources - [Google Search Central, AI features and your website (December 2025)](https://developers.google.com/search/docs/appearance/ai-features) - [Google Search Central, Optimizing your website for generative AI features on Google Search (July 2026)](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide) - [Google Search Central, Organization structured data (accessed September 2026)](https://developers.google.com/search/docs/appearance/structured-data/organization) - [Search Engine Journal, Google drops FAQ rich results from Search (May 2026)](https://www.searchenginejournal.com/google-drops-faq-rich-results-from-search/574429/) - [Microsoft Advertising blog, Optimizing your content for inclusion in AI search answers (October 2025)](https://about.ads.microsoft.com/en/blog/post/october-2025/optimizing-your-content-for-inclusion-in-ai-search-answers) - [Search Engine Journal, New data reveals the top 20 factors influencing ChatGPT citations (November 2025)](https://www.searchenginejournal.com/new-data-top-factors-influencing-chatgpt-citations/561954/) - [Seer Interactive, 87% of SearchGPT citations match Bing's top results (February 2025)](https://www.seerinteractive.com/insights/87-percent-of-searchgpt-citations-match-bings-top-results) - [Vercel, The rise of the AI crawler (December 2024)](https://vercel.com/blog/the-rise-of-the-ai-crawler) --- Title: Do we still need SEO now that buyers ask ChatGPT instead of Google? URL: https://fortitudemedia.ai/insights/seo-vs-llm-optimisation Author: David Adams Published: 2026-04-22 Last reviewed: 2026-09-24 Topic: Getting named by AI assistants Short answer: Yes, but not the SEO you bought in 2019. Google says optimising for its AI features is still SEO, and ChatGPT search, Copilot and Google's AI answers all start from a search index, so ranking helps but does not guarantee a mention. Keep the technical basics and structured data, and get your site into Bing. Stop keyword-volume pages and bought links. Add what some agencies call GEO (generative engine optimisation): clear positioning, depth on buyer questions and mentions in publications buyers trust. For a decade the deal was clear. You paid an agency, they chased keywords and links, and a monthly report showed your rankings creeping up. It worked, more or less. Then a prospect told you in a first meeting that they had asked ChatGPT for a shortlist of firms like yours, and you were not on it. Your rankings were fine that week. It made no difference. That is where most founder-led B2B firms are now. Google still sends traffic, so nobody wants to abandon SEO. But buyers increasingly start elsewhere: in [G2's April 2026 survey](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html) of 1,076 software buyers, 51% said they now begin their research in an AI chatbot, up from 29% a year earlier. And Google itself now answers many questions with an AI Overview before anyone sees a link. Rankings still matter, but they no longer settle it. [Ahrefs found in March 2026](https://ahrefs.com/blog/ai-overview-citations-top-10) that only 38% of the pages cited in Google's AI Overviews ranked in the top ten for the search, down from 76% in July 2025, because Google now runs several related searches for one question. Most of what gets a firm named by AI assistants also helps it rank. But a chunk of what agencies still sell as SEO does little for AI answers, and some of it does harm. This guide sorts the two. ## Key takeaways - SEO is not dead: Google says optimising for its AI features is "still SEO", and Google's AI features, Copilot and ChatGPT search all start from a search index. - Ranking helps but does not guarantee a mention, because assistants run several searches per question and cite pages that answer the follow-ups too. - ChatGPT search has drawn heavily on Bing and Copilot runs on Bing's index, so verify your site in Bing Webmaster Tools and submit new pages through IndexNow. - Keep technical hygiene, structured data and clear site structure; stop keyword-volume content, bought links and clickbait titles. - Add what assistants reward: positioning they can repeat, depth on the questions buyers ask, and mentions in publications buyers trust. - Run one programme for Google and AI, measured by rankings and by whether you are named for a fixed list of buyer questions. ## Is SEO dead now that people use ChatGPT? No. Google's own [guide to its generative AI features](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide), last updated in July 2026, says optimising for them "is still SEO", because those features are built on Google's core ranking and quality systems. The link between ranking and being named is real but loose. Google's AI features, ChatGPT search and Copilot all start from a search index (Google's, or Bing's for Copilot and, it appears, ChatGPT search), so a page that ranks is more likely to be read. But Google now issues several related searches for one question, a process it calls query fan-out, which is why the top-ten share of AI Overview citations halved. A firm that ranks first for its main phrase and has nothing on the follow-up questions can still be left out. Our guide to [why ranking first on Google does not get you mentioned by ChatGPT](/insights/google-rankings-wont-protect-you) covers that gap in detail. ## What is the difference between SEO and GEO? SEO gets a page ranked in a list; what some agencies now call GEO (generative engine optimisation) or AEO (answer engine optimisation) tries to get a firm named inside a written answer. We call it being named by AI assistants, and most of the work overlaps. The difference is that a filter became a reader. Google's classic results are a filter: you type a phrase, it returns matching pages in order, and you choose. An AI assistant is a reader: you ask in your own words, it reads what it knows and what it can fetch, and it writes an answer that names two or three firms. A filter can be fed the right words. A reader has to be convinced. | | Classic SEO | Being named by AI assistants (GEO or AEO) | |---|---|---| | What the buyer sees | A ranked list of links | One written answer naming a few firms | | What wins | Relevance and authority for a phrase | Clear, consistent, trusted evidence about the firm | | Where the evidence comes from | Mainly your pages and links to them | Your pages plus what others say about you | | What you measure | Rankings, clicks, enquiries | Whether you are named for buyer questions, and enquiries | Buyers mostly use the plain term. [Google Trends](https://trends.google.com/trends/explore?date=today%2012-m&geo=GB&q=AI%20SEO,generative%20engine%20optimization,answer%20engine%20optimization,AI%20visibility,LLM%20SEO) for the UK over the year to 24 September 2026 shows "AI SEO" drawing about nine times the interest of "generative engine optimization", with GEO ahead of AEO. The label matters less than the work. ## Does ChatGPT use Google or Bing to find websites? OpenAI says ChatGPT search uses "third-party search providers" without naming them, and the evidence points to Bing. In a small [Seer Interactive test](https://www.seerinteractive.com/insights/87-percent-of-searchgpt-citations-match-bings-top-results) in February 2025, 87% of ChatGPT search citations matched Bing's top results, against 56% for Google. Microsoft says Copilot is [powered by Bing's search index](https://about.ads.microsoft.com/en/blog/post/october-2025/optimizing-your-content-for-inclusion-in-ai-search-answers). Most B2B firms have never looked at Bing. Three jobs fix that: - Verify your site in Bing Webmaster Tools, submit your sitemap and check which pages are indexed. - Turn on [IndexNow](https://www.bing.com/indexnow), a free protocol that tells Bing and other participating search engines as soon as a page is added, changed or deleted. Many content systems and plugins support it. - Allow OpenAI's search crawler, OAI-SearchBot, in your robots.txt and your CDN settings; OpenAI's help centre says a site must allow it to appear in ChatGPT search. One qualification on robots.txt, the file that tells crawlers what they may visit: it is a request, not a lock. OpenAI says robots.txt rules may not apply to ChatGPT-User, which fetches a page when a person asks for it, as [Search Engine Journal reported](https://www.searchenginejournal.com/openai-says-robots-txt-may-not-apply-to-chatgpts-fetch-bot/585864/) in August 2026. The bigger risk for most firms is the opposite: blocking the crawlers they want by accident. [How AI crawlers read your website](/insights/ai-crawlers-vs-google-spiders) shows how to check in five minutes. ## Keep, stop, add: how to mark up your agency's invoice Take your agency's last invoice or monthly report and mark each line against this table. | Activity | Effect on Google | Effect on AI answers | Verdict | |---|---|---|---| | Technical fixes: speed, crawlability, sitemaps, one version of each page | Helps | Helps; AI crawlers are simpler than Google's | Keep | | Structured data | Helps with rich results and understanding | Removes doubt about your facts | Keep and extend | | Site structure and internal links | Helps | Helps an assistant build a full picture | Keep | | Bing Webmaster Tools and IndexNow | Neutral | Helps Copilot and ChatGPT search find new pages | Add | | Keyword-volume content (one thin page per phrase) | Weak, and risky at scale | Rarely quoted; dilutes your better pages | Stop | | Bought, swapped or directory links | Discounted or penalised | No value; links count only as a sign of real reputation | Stop | | Clickbait titles and descriptions | Short-term clicks | Marks against you when the page does not deliver | Stop | | Plain positioning on every key page | Helps | The first thing an assistant repeats | Add | | Deep pages on real buyer questions | Helps | The pages assistants quote | Add | | Earned coverage and consistent mentions | Helps through links and brand searches | In our view, the biggest gap in most budgets | Add | ### Keep: technical hygiene, structured data and structure A fast, crawlable site with one clear version of every page is easier for every crawler to read, and the main AI crawlers do not run JavaScript, so key text must be in the page's HTML. Structured data labels your facts; [whether you need schema markup for AI assistants](/insights/structured-data-schema-markup-ai) covers what it does and does not do. Clear sections and internal links let an assistant that lands on one page find your sector pages, team and case studies. ### Stop: keyword content, bought links and clickbait The old model was to find a phrase with search volume, write a page that repeats it, and repeat for a hundred phrases. To a reader that looks like writing aimed at a machine, and it dilutes the pages that might be quoted. Google's July 2026 guide goes further: creating separate pages for every query variation mainly to manipulate AI answers breaks its scaled content abuse policy. [Why ChatGPT ignores sites with hundreds of pages](/insights/ai-penalises-thin-content) covers the damage. Links are more nuanced than "links no longer matter". Search indexes do weigh links, and an [SE Ranking study reported by Search Engine Journal](https://www.searchenginejournal.com/new-data-top-factors-influencing-chatgpt-citations/561954/) in November 2025 found the number of referring domains was the strongest predictor of ChatGPT citations. What counts is links that reflect a real reputation: coverage, partners, professional bodies. Directory submissions, paid networks and guest posts on sites nobody reads carry none of that, and Google has discounted them for years. If your agency reports "links built" as a headline number, cut that line first. [Whether backlinks still matter for ChatGPT](/insights/ai-evaluates-quality-referring-domains) goes further. Titles written to win clicks ("7 secrets every MD gets wrong") now work against you: an assistant reads the title to understand the page, and plain titles that say what the page answers work better for both audiences. ### Add: positioning, depth and outside evidence Read your homepage as a stranger with thirty seconds. Can you say what the firm does, for whom, where and why it is different? If the answer needs inference, an assistant will infer wrongly or leave you out. In our experience this is the most common problem on B2B sites and the cheapest to fix. Then depth. One page that properly answers "how do we move payroll provider mid-year without breaking anything", with the steps, the traps and the timescales, does more than twenty short posts circling the topic. Depth means completeness, written by someone who has done the work. Finally, evidence from outside your own site, which most SEO budgets never touched. Muck Rack's [May 2026 analysis](https://www.globenewswire.com/news-release/2026/05/07/3290268/0/en/generative-pulse-earned-media-consistently-drives-ai-citations-holding-at-84.html) of more than 25 million links cited by ChatGPT, Claude and Gemini found 84% came from earned media and 0.3% from paid content. Trade press, professional bodies, review sites and partners' pages are where that evidence lives. Our guide to [which kinds of PR win work for a B2B firm](/insights/pr-for-b2b-what-works) shows how to do it at a founder-led scale. ## How should the budget shift? Do not run two programmes. Run one and change the mix, roughly in thirds, whatever your budget is today. | Share of budget | What it buys | How often | |---|---|---| | About a third | Technical and structural work: crawler access, Bing, structured data, site structure | Done once, then maintained | | About a third | Fewer, deeper pages written from real expertise on the questions buyers ask | Every month | | About a third | Earned coverage, reviews and consistent mentions on the sites buyers and assistants trust | Every month | Once the technical base is sound, its share shrinks to upkeep and the rest moves to content and coverage. The exact split depends on where you start; for what our own programmes cost, see [what it costs](/what-it-costs). The measure changes too. Alongside rankings, track whether the assistants name you for a fixed list of 20 to 50 questions your buyers ask. Ask each several times a month in a clean session, because answers vary from run to run, and count a win only when it repeats. ## What does this look like for a real firm? [ID-Pal](/work/id-pal) sells identity verification to fund administrators, wealth managers, law firms and accountants. It had bought NorthRow, a UK business verification platform, and folded it into id-pal.com. "Traffic to the pages that bring in enquiries had fallen by around 40 per cent since the NorthRow move, and nobody could say why." And "AI could describe ID-Pal. It never recommended it." "Across 36 recorded AI answers to the questions ID-Pal's buyers actually ask, ID-Pal was named three times, and only for its brand or its Salesforce integration. AI knew who it was, and had no reason to recommend it." What changed was a mix of SEO and AI work. First, "About four hours of structured data work declaring NorthRow part of ID-Pal on both domains." Then "Five industry pages rebuilt: fund administration, wealth management, legal, accountancy and estate agency. Each opens with the question a compliance lead types, names the regulator, and answers inside the first ninety words." The result so far: "UK buyer questions where AI names ID-Pal: none in May, two in August, four in September 2026." And "Where a page has been rebuilt, ID-Pal appears. Where it has not yet, it does not, which tells us exactly what to build next." ## How much of our traffic comes from ChatGPT compared with Google? Your own analytics will tell you, once it is set up to. Since 13 May 2026, Google Analytics 4 has had a default "AI Assistant" channel covering ChatGPT, Gemini, DeepSeek, Copilot and Grok, as [Nice Looking Data documents](https://www.nicelookingdata.com/blog/ga4-ai-traffic-chatgpt-referrals). Claude and Perplexity are not included, and clicks from AI Overviews and AI Mode still count as Organic Search. So compare Organic Search, the AI Assistant channel and referrals from claude.ai and perplexity.ai, remembering that many buyers read an answer and then search your name, which shows up as branded search or direct. Search Console's generative AI reports, available to all sites since August 2026, show impressions in AI Overviews and AI Mode. Our guide on [telling which leads came from ChatGPT](/insights/attribution-ai-era-track-leads) covers the rest, including simply asking every new enquiry how they found you. ## What to do this month 1. Check your robots.txt and your CDN or security settings for blocked AI crawlers, and make sure OAI-SearchBot is allowed. 2. Verify your site in Bing Webmaster Tools, submit your sitemap and turn on IndexNow. 3. Rewrite the first two paragraphs of your homepage and each service page so a stranger could repeat back what you do, for whom and where. 4. Ask your agency for the list of links built in the last year and the list of publications that have mentioned you. If the first is long and the second is empty, change the brief. 5. Choose one buyer question and write the page that answers it completely, with someone who has done the work in the room. 6. Write down 20 to 50 questions your buyers ask and start recording who is named for them each month. ## Frequently asked questions ### Should we stop our SEO agency altogether? Not if they are doing the technical and structural work well. Ask them to drop keyword-volume content and link quotas, add Bing and IndexNow, and report on whether you are named by AI assistants for your key buyer questions. If they cannot or will not, the relationship has run its course. [How we compare with an SEO agency](/vs/seo-agency) sets out where our approach differs. ### Should I hire an SEO agency or a GEO agency in 2026? Hire for the work, not the label. The best providers do both, because the technical base, the content and the outside evidence serve Google and AI assistants alike. Nobody can guarantee what an AI assistant will say, so be wary of anyone guaranteeing a place in ChatGPT's answers; a good provider guarantees effort, reporting and the outcomes it controls. Our guide on [choosing a firm to help you get recommended by ChatGPT](/insights/how-to-compare-ai-optimisation-providers) lists the questions to ask. ### Does Google still matter for a B2B firm? Yes. Google still sends traffic, its AI Overviews and AI Mode draw on the same index, and Gemini leans on Google Search. A firm that ignores Google loses on both fronts. The point is to stop the parts of SEO that only ever worked on the filter. [What Google AI Mode is and how it differs from AI Overviews](/insights/end-of-ten-blue-links) explains Google's newer format. ### How long before changes show up in AI answers? Assistants that search the web live (ChatGPT search, Perplexity, Copilot and Google's AI features) can reflect a clearer page or a new article within weeks. Answers drawn from a model's own memory change only when the model is retrained, which takes months. The ID-Pal case study reports: "UK buyer questions where AI names ID-Pal: none in May, two in August, four in September 2026." ### What is the cheapest thing we can do first? Fix your positioning in words. Rewriting the opening of your homepage and service pages costs a day of a founder's attention and changes what an assistant can say about you. Then check you are not blocking AI crawlers and that Bing has your pages. Everything else builds on those. For the wider method, start with [how to get ChatGPT to recommend your firm](/insights/what-is-llm-optimisation). To see how this plays out for your own firm, [book a free Growth Review](/growth-review): 30 minutes with David Adams and three fixes in writing. If the gap is outside evidence, that is the work of [Reputation That Wins](/reputation-that-wins). ## Sources - [Google Search Central, Optimizing your website for generative AI features on Google Search (July 2026)](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide) - [Ahrefs, AI Overview citations from top-ten pages (March 2026)](https://ahrefs.com/blog/ai-overview-citations-top-10) - [G2 via PR Newswire, Half of B2B software buyers now start their research with AI chatbots (April 2026)](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html) - [Seer Interactive, 87% of SearchGPT citations match Bing's top results (February 2025)](https://www.seerinteractive.com/insights/87-percent-of-searchgpt-citations-match-bings-top-results) - [Search Engine Journal, New data reveals the top 20 factors influencing ChatGPT citations (November 2025)](https://www.searchenginejournal.com/new-data-top-factors-influencing-chatgpt-citations/561954/) - [Muck Rack via GlobeNewswire, Earned media consistently drives AI citations (May 2026)](https://www.globenewswire.com/news-release/2026/05/07/3290268/0/en/generative-pulse-earned-media-consistently-drives-ai-citations-holding-at-84.html) - [Google Trends, UK interest in AI SEO, generative and answer engine optimization (12 months to September 2026)](https://trends.google.com/trends/explore?date=today%2012-m&geo=GB&q=AI%20SEO,generative%20engine%20optimization,answer%20engine%20optimization,AI%20visibility,LLM%20SEO) - [Nice Looking Data, AI traffic in GA4 (July 2026)](https://www.nicelookingdata.com/blog/ga4-ai-traffic-chatgpt-referrals) --- Title: Does our website's speed and accessibility really affect enquiries? URL: https://fortitudemedia.ai/insights/website-accessibility-speed-matter Author: David Adams Published: 2026-04-22 Last reviewed: 2026-09-24 Topic: Websites that bring work in Short answer: Yes. A slow or awkward site loses buyers before they read a word. Google's Core Web Vitals ask for main content within 2.5 seconds, a response to taps within 200 milliseconds and a layout shift of 0.1 or less on 75% of real visits. Aim for WCAG 2.2 AA: clear headings, alt text, contrast and labelled forms help disabled visitors, Google and AI assistants alike. Test with PageSpeed Insights and WAVE, then fix your most important pages first. Speed and accessibility are the parts of a website most founders delegate and forget. So nobody notices when the new video header adds a few seconds to every page, when the chat widget, cookie banner and three tracking tools all load before the headline, or when pale grey text looks elegant on a large monitor and unreadable on a phone in daylight. Buyers notice, though they rarely tell you. A finance director who taps your link between meetings and waits at a blank screen taps the next firm instead. A procurement lead who uses a screen reader cannot get past your enquiry form. None of this shows up in a report as a lost enquiry. It just does not arrive. The problem is common. [WebAIM's 2026 test of the home pages of the top million websites](https://webaim.org/projects/million/) found detectable accessibility failures on 95.9% of them, and low-contrast text on 83.9%. On speed, the [HTTP Archive's 2025 Web Almanac](https://almanac.httparchive.org/en/2025/performance) found that only 48% of mobile websites passed all three of Google's Core Web Vitals. Both are global figures, and the fixes are usually far cheaper than a redesign. ## Key takeaways - A slow or awkward website loses buyers before they read a word and tells the rest that nobody is paying attention. - Google's Core Web Vitals ask for main content within 2.5 seconds, a response to taps within 200 milliseconds and a layout shift score of 0.1 or less, on 75% of real visits. - Oversized images, video headers, heavy themes and too many third-party scripts cause most slow B2B sites, and most can be fixed without a rebuild. - The accessibility standard to aim for is WCAG 2.2 at level AA, now also the international standard ISO/IEC 40500:2025. - The European Accessibility Act has applied since 28 June 2025 to certain services sold to EU consumers, so a UK fintech with EU consumers may be caught, while a purely B2B firm usually is not. - Clear headings, alt text, descriptive links and real text in the page help disabled visitors, Google and AI assistants at the same time. ## Why does website speed affect enquiries? Because the first few seconds decide whether a buyer stays. A buyer comparing three firms opens each site in turn. If yours is still loading while the others are readable, you start the comparison behind, and some buyers never see your homepage at all. Speed also sends a signal: a slow site that jumps around or freezes when tapped suggests a firm that does not pay attention. The effect is strongest on phones, where many buyers first look you up after an AI assistant or a colleague has given them your name. We cover the mobile side in [whether your buyers really look at your website on their phones](/insights/mobile-first-design-b2b), and the other small signals that shape trust in [what makes a buyer trust a website enough to get in touch](/insights/psychology-of-trust-website-conversions). ## What does Google measure, and what counts as fast enough? Google measures page experience with three Core Web Vitals, judged on real visits from Chrome users. Google says good scores, along with other page experience aspects, align with [what its core ranking systems seek to reward](https://developers.google.com/search/docs/appearance/core-web-vitals). The thresholds, from [Google's Search Console help on the Core Web Vitals report](https://support.google.com/webmasters/answer/9205520?hl=en): | Measure | What it means | Good | Needs improvement | Poor | |---|---|---|---|---| | Largest Contentful Paint (LCP) | How long the main headline or image takes to appear | 2.5 seconds or less | Up to 4 seconds | Over 4 seconds | | Interaction to Next Paint (INP) | How quickly the page reacts to a tap, click or key press | 200 milliseconds or less | Up to 500 milliseconds | Over 500 milliseconds | | Cumulative Layout Shift (CLS) | How much the page jumps as it loads | 0.1 or less | Up to 0.25 | Over 0.25 | Each score is judged at the 75th percentile: a page counts as good only if three quarters of real visits meet the threshold. PageSpeed Insights shows this real-user data, which Google calls field data, where there are enough visits, and a simulated test otherwise. Speed is one ranking factor among many: a fast site with weak content will not rank on speed alone, but a slow one gives away an advantage for nothing. ## What is a good INP score, and how do we fix a poor one? A good INP is 200 milliseconds or less, and anything over 500 milliseconds is poor. A poor score usually means too much JavaScript is competing for the browser's attention when someone taps or clicks. On B2B sites the usual culprits are chat widgets, tag managers full of old tags, heatmaps, booking tools, cookie banners and page builders that run large amounts of code on every page. Remove scripts nobody uses, and delay the rest until after the page has loaded or until someone needs them, so the chat widget loads when a visitor clicks "chat". If the theme itself is the cause, the fix is larger. Search Console's Core Web Vitals report shows which pages are poor. ## What usually makes a B2B website slow? A handful of causes account for most slow B2B sites: - Oversized images. The [2025 Web Almanac](https://almanac.httparchive.org/en/2025/performance) found that the largest element on 76% of mobile pages is an image, so resizing and compressing images is often the single biggest win. - Video headers and animations, which load slowly and push your message down the page. - Third-party scripts for analytics, advertising, chat, heatmaps and cookie banners, often including tools nobody uses any more. - Heavy themes and page builders that load code on every page whether it is needed or not. - Several web fonts in several weights, and slow or distant hosting. Where the platform itself is the problem, see [how your choice of CMS affects whether AI assistants can read you](/insights/cms-choice-affects-ai-visibility). ## How big a job is each common fix? Most fixes are small jobs a developer can scope in advance. The sizes below come from our experience of B2B sites: small means hours of developer time, medium a few days, large a project in its own right. Ask for a fixed quote for each item. | Fix | What it involves | What it improves | Size of job | |---|---|---|---| | Compress and resize images | Resize to the size shown, compress, serve modern formats such as WebP | LCP | Small | | Replace a video header | A strong still image with the headline over it | LCP | Small | | Remove unused scripts | List every third-party tool, remove the unused ones, delay the rest | INP and LCP | Small to medium | | Reserve space for images and banners | Set sizes so the layout does not jump | CLS | Small | | Fix text contrast | Darken pale text to WCAG's 4.5 to 1 ratio | Accessibility | Small | | Label forms properly | A visible label on every field and clear error messages | Accessibility and enquiries | Small | | Add alt text | Describe meaningful images, mark decorative ones to be skipped | Accessibility, search and AI | Small to medium | | Make menus and pop-ups keyboard friendly | Visible focus, no traps, pop-ups that close with Escape | Accessibility | Medium | | Better hosting, caching and a CDN | Faster server, page caching, copies of the site closer to visitors | LCP | Medium | | Replace a heavy theme or page builder | Rebuild templates on a lighter base | Everything | Large | ## Which speed fixes make the biggest difference on a WordPress site? In our experience, image optimisation, page caching, fewer plugins and better hosting, roughly in that order. Use an image plugin or image CDN that resizes and compresses automatically, turn on page caching at the host or through a well-supported plugin, and remove unused or duplicate plugins, since each can add code to every page. Page builders are a common cause of poor INP; a lighter theme is a larger but lasting fix. ## Does speed affect what AI assistants say about you? Indirectly, yes. Speed alone will not make ChatGPT or Claude recommend you, but a script-heavy site can stop them reading you at all. [Vercel and MERJ found in December 2024](https://vercel.com/blog/the-rise-of-the-ai-crawler) that GPTBot, ClaudeBot and PerplexityBot fetch JavaScript files but do not run them. If your words only appear after scripts run, those crawlers see an empty page. Googlebot does run JavaScript, which helps Gemini and Google's AI features. The work that makes a site fast, serving real text in the page and cutting unnecessary scripts, is the same work that makes it readable by AI assistants. We explain each crawler in [how AI crawlers read your website](/insights/ai-crawlers-vs-google-spiders). ## What does accessibility mean for a B2B website? It means anyone can use the site: people who use screen readers or navigate by keyboard, people with dyslexia or hearing loss, and people reading a phone in bright sunlight or holding it in one hand on a train. The standard to aim for is WCAG 2.2 at level AA, published by the World Wide Web Consortium on 5 October 2023. The [W3C says](https://www.w3.org/WAI/standards-guidelines/wcag/) it is now also an international standard, ISO/IEC 40500:2025, and encourages everyone to use the latest version. If a developer or an old tender still quotes WCAG 2.0 or an older 2.x release, move the target to 2.2. The checks that matter most on a B2B site: - Text contrast of at least 4.5 to 1 for normal text. - Alt text on meaningful images, and headings used in order. - Everything usable with a keyboard, with a visible focus outline that is not hidden behind a sticky header or cookie banner. - Tap targets of at least 24 by 24 CSS pixels, or enough space around smaller ones, a WCAG 2.2 addition. - Forms with labels, clear error messages and no need to type the same details twice in one process. - Links that say where they go, and captions or transcripts for video. ## Is accessibility a legal requirement for a private B2B firm in the UK? No UK law sets a WCAG level for private companies, but the Equality Act 2010 requires service providers to make reasonable adjustments for disabled people, which is widely understood to include websites. Public sector bodies have specific rules, the Public Sector Bodies (Websites and Mobile Applications) (No. 2) Accessibility Regulations 2018, and [GOV.UK's guidance](https://www.gov.uk/guidance/accessibility-requirements-for-public-sector-websites-and-apps), last updated in September 2024, tells them to meet WCAG 2.2 AA and publish an accessibility statement. Public sector and large corporate buyers often ask suppliers about accessibility in tenders. And the people making buying decisions include disabled people: a site they cannot use is a site they will not buy from. ## Does the European Accessibility Act apply to a UK company that sells to EU customers? It can. The [European Accessibility Act (Directive (EU) 2019/882)](https://eur-lex.europa.eu/legal-content/EN/TXT/HTML/?uri=CELEX:32019L0882) has applied since 28 June 2025 to listed services provided to consumers in the EU, wherever the provider is based. What decides it is whether your service is on the list and whether your customers are consumers. | Your situation | Likely position under the Act | |---|---| | B2B firm selling only to businesses | Usually outside it: a consumer is a person buying outside their trade, business or profession | | UK fintech offering payment, e-money, credit or investment services to EU consumers | Likely caught: consumer banking services are covered | | UK firm selling online to EU consumers | Likely caught: e-commerce services are covered | | Service provider with fewer than 10 staff and turnover or balance sheet of no more than €2 million | Exempt as a microenterprise | A provider that is caught must meet the Act's accessibility requirements and publish information on how its service meets them. The Act sets functional requirements rather than naming a WCAG version; in practice, a website built to WCAG 2.2 AA is in a strong position. Each EU country enforces the Act through its own law, so if you serve EU consumers, take legal advice. ## Why does accessibility help search and AI assistants too? Because machines read a page much as a screen reader does, through the text, headings, alt text and link wording. A clear heading structure gives a screen reader user an outline and a crawler a map, and real text, rather than text inside images, PDFs or scripts, can be read aloud, searched and quoted. This is how we build sites. For [arum](/work/arum), the collections and recoveries consultancy, the rebuild delivered "Proper structure and markup on every page, a plain summary AI assistants can read, facts stored once so two pages never disagree, and every old address redirected so years of rankings carry over." Much of our [technical checklist for an AI-ready website](/insights/technical-checklist-ai-ready-website) overlaps with an accessibility review for the same reason, so do both at once. ## How do you test your own website? With free tools and a real phone; you do not need to be technical. ### Step 1: Run PageSpeed Insights on your key pages Test your homepage, main service page, a case study and your contact page. Check whether the Core Web Vitals assessment passes on mobile and note the main issues. ### Step 2: Run an accessibility checker Use WAVE, from WebAIM, or axe DevTools on the same pages. Automated tools catch only part of the picture, so treat a clean result as a start. ### Step 3: Try the site with only a keyboard Use the Tab key to move through your homepage and enquiry form. If you get stuck or cannot submit the form, neither can some of your buyers. ### Step 4: Use a real phone on mobile data Switch off wifi, load your key pages and time how long it takes to read the headline and find the contact details. ### Step 5: Fix the worst problems on the most important pages first Start with the pages that matter most for enquiries, fix the biggest issues and test again. Our [guide to websites that bring work in](/insights/pillar/website-design-performance) covers what else each key page needs. ## What to do this month 1. Run PageSpeed Insights and WAVE on your five most important pages and save the results as a baseline. 2. Ask your developer to list every third-party script on the site and remove any that nobody uses. 3. Compress the images on your homepage and main service pages, and replace any video header with a still image. 4. Fix poor contrast, add alt text to meaningful images and label every form field. 5. Test your enquiry form using only a keyboard and a phone on mobile data. 6. If you sell to EU consumers or the public sector, check your position and set WCAG 2.2 AA as the target. ## Frequently asked questions ### What PageSpeed score should we aim for? Focus on whether your key pages pass the Core Web Vitals assessment on real-user data, not on the score out of 100. The score comes from a simulated test and moves from run to run. A page that passes on real-user data and feels quick on a phone is in good shape. ### How much does a WCAG 2.2 accessibility audit cost in the UK? We do not quote market prices in this guide, because scope varies so much. Cost depends on how many page templates are tested, whether testing includes manual checks with screen readers and keyboards, whether disabled users are involved, and whether a retest after fixes is included. Ask for quotes scoped by template, such as homepage, service page, article and form, and check the report ties each failure to a WCAG 2.2 success criterion. ### Does a private B2B company need an accessibility statement on its website? UK law does not require one from a private company; the rule applies to public sector bodies. A service caught by the European Accessibility Act must publish information on how it meets the requirements. Even where it is optional, a short, accurate statement of the standard you aim for and how to report a problem helps in tenders. ### Will an accessibility plugin or overlay fix our site? Rarely. Overlays that promise accessibility with one line of code cannot fix missing labels, poor structure or inaccessible forms, and many disabled users and accessibility specialists advise against them. Fix the site itself; most issues on a typical B2B site are small corrections. ### Does our choice of hosting affect speed? Yes. A slow server delays everything that follows, however well the pages are built. Good hosting with caching and a content delivery network is one of the easiest improvements to make. A fast, accessible website helps buyers, Google and AI assistants at once, and it rarely needs a rebuild. To have a site built to these standards from the start, see [Website That Sells](/website-that-sells), or book a free [Growth Review](/growth-review) and we will send three fixes in writing. ## Sources - [WebAIM, The WebAIM Million: 2026 report on the accessibility of the top 1,000,000 home pages (March 2026)](https://webaim.org/projects/million/) - [HTTP Archive, Web Almanac 2025: Performance (January 2026)](https://almanac.httparchive.org/en/2025/performance) - [Google Search Console Help, Core Web Vitals report (accessed September 2026)](https://support.google.com/webmasters/answer/9205520?hl=en) - [Google Search Central, Understanding Core Web Vitals and Google search results (December 2025)](https://developers.google.com/search/docs/appearance/core-web-vitals) - [W3C Web Accessibility Initiative, WCAG 2 Overview (September 2026)](https://www.w3.org/WAI/standards-guidelines/wcag/) - [GOV.UK, Accessibility requirements for public sector websites and apps (September 2024)](https://www.gov.uk/guidance/accessibility-requirements-for-public-sector-websites-and-apps) - [EUR-Lex, Directive (EU) 2019/882 on the accessibility requirements for products and services (April 2019)](https://eur-lex.europa.eu/legal-content/EN/TXT/HTML/?uri=CELEX:32019L0882) - [Vercel, The rise of the AI crawler (December 2024)](https://vercel.com/blog/the-rise-of-the-ai-crawler) --- Title: How do I get my board to approve budget for AI search? URL: https://fortitudemedia.ai/insights/business-case-ai-optimisation-board Author: David Adams Published: 2026-04-22 Last reviewed: 2026-09-24 Topic: Costs, returns and the business case Short answer: Show the board what ChatGPT, Gemini and Copilot say when buyers ask for a firm like yours, and tie it to a number they already watch. Work out the break-even from your own gross profit per client, not fees. Ask for a fixed budget with monthly measures, a twelve-month review and stop conditions agreed in advance. Present the work as an asset judged over two to three years, even though it is booked as an expense. Evidence and a capped downside win approval. You have typed a few of your buyers' questions into ChatGPT and watched it name two competitors and not you. Now you need the board, your co-founders or the chair to agree to spend money on something that will not show up in the pipeline report for months. Boards are not hostile to new ideas, only to vague ones. A paper that shows what buyers see today, ties it to a number the board already watches, and sets out a cost, a measure and a point at which you would stop usually gets approved. Market figures help, but they will not carry the paper on their own. In [G2's April 2026 survey of 1,076 software buyers](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html), 51% said they now start their research in an AI chatbot, up from 29% a year earlier, and [Forrester reported in January 2026](https://www.forrester.com/blogs/b2b_buyers_make_zero_click_buying_number_one/) that 94% of business buyers use AI somewhere in their buying process. Both studies are international and lean towards technology purchases. What wins the vote is what the assistants say about your own market, which you can gather this week. ## Key takeaways - Boards approve specific, evidenced proposals, so show them what ChatGPT, Gemini and Copilot say about your market today rather than forecasts about AI. - Tie the proposal to a number the board already watches, such as enquiries, win rate or revenue from new clients. - Work out the break-even on gross profit, not fees, because a finance director will redo the sum with margin. - Present the work as an asset that keeps working after you stop paying, judged over two to three years, even though most of it is booked as an expense. - Promise measures you can report every month, such as how often you are named for a fixed list of buyer questions, rather than a guaranteed return. - Say in advance what would make you stop, because a defined exit makes a cautious board far more willing to start. ## What does a board need to see before it approves this? Four things: evidence that the problem is real in your market, a plan with a cost, a way to tell whether it is working, and a point at which you would stop. Most proposals about AI search offer only the first, and offer it as a general trend rather than as evidence about your firm. A new office comes with a lease, a hiring plan, a revenue target and a review date; this proposal should look the same. The board's real fear is money spent on activity with no visible result and nobody accountable for it, so answer that fear directly. ## How to build the case, step by step ### Step 1: Show the board what buyers see today Write down 20 questions a buyer would ask when looking for a firm like yours, in their words: "best employment law firm for tech start-ups in Manchester", "who can help us get ISO 27001 certified". Ask each in ChatGPT, Gemini, Copilot, Perplexity, Claude and Google's AI Mode, in a fresh chat and more than once, because answers vary from run to run. Record who is named and how you are described, and add two or three screenshots: nothing persuades a board faster than a competitor named for the work you do best. Being known is not the same as being recommended. ID-Pal, an identity verification firm we work with, had bought the UK verification platform NorthRow and folded it into its own site. Our baseline found this: "Across 36 recorded AI answers to the questions ID-Pal's buyers actually ask, ID-Pal was named three times, and only for its brand or its Salesforce integration. AI knew who it was, and had no reason to recommend it." After the fixes, the same questions were asked every month: | What we tracked for ID-Pal | Result | |---|---| | UK buyer questions where AI names ID-Pal | none in May, two in August, four in September 2026 | | AI answers still recommending NorthRow as a rival | six in May, one in September 2026 | | Accuracy | 6 of 6 correct answers to 'what does ID-Pal do?', every engine, every run | That is evidence a board can follow. The full story is in the [ID-Pal case study](/work/id-pal), and [benchmarking how often you are named against competitors](/insights/benchmarking-ai-visibility-competitors) explains how to make your own comparison fair. ### Step 2: Tie it to a number the board already watches Connect the evidence to a number that comes up at every meeting: new enquiries, proposal volume, win rate or revenue from new clients. If web enquiries have fallen while rankings held, explain why: in a [US study by Pew Research Center (July 2025)](https://www.pewresearch.org/short-reads/2025/07/22/google-users-are-less-likely-to-click-on-links-when-an-ai-summary-appears-in-the-results/), people clicked a traditional result on 8% of visits when a Google AI summary appeared, against 15% when none did. If prospects arrive with a shortlist already formed, or you lost a pitch to a firm named in every answer, say so. The aim is to move from "AI is interesting" to "this affects a number we already worry about". ### Step 3: Do the break-even sum on gross profit Use four figures your finance director already trusts: the fees from a typical new client over its lifetime, your gross margin on that work, your win rate and the proposed annual spend. Do not borrow return multiples from a vendor's slide. Example: a typical new client pays £60,000 in fees over three years and your gross margin is 50%, so each client adds £30,000 of gross profit. The proposal is to spend £40,000 a year, or £120,000 over three years. To break even, the work needs four extra clients across the three years. At a win rate of one in three, that is about twelve extra proposals, or four a year. The board is then judging whether being named by AI assistants will bring four more proposals a year. Done on fees, the same sum seems to need only two clients, and a finance director who redoes it will doubt the rest of the paper. In our experience the first year is when spend runs ahead of return, so show the break-even over the full period. Our guide to [what it costs you not to be recommended](/insights/cost-of-inaction-losing-right-now) sizes the other side of the ledger. ### Step 4: Set out the plan and what you are building Describe the work plainly: a website that says what you do, for whom and where; useful answers to buyers' questions; your experts quoted where your buyers read; and one consistent description of the firm across the web. Name who does each part, and say how much senior time it needs: a monthly interview with two or three experts, with someone else writing, is a far smaller ask than an article a week. Then explain why this belongs with investments. The test is what happens the month after you stop paying. Google Ads and sponsored posts rent attention, and the results stop with the invoice. The work that gets you named builds five things that keep working: - **A library of answers** to the questions your buyers ask, with your view and your evidence. - **A record of what others say about you:** coverage, guest articles, awards and reviews. - **One consistent description** of your services, sectors, locations and people, which matters because [AI assistants cross-check what is said about a firm](/insights/ai-cross-references-business-sources). - **Visible expertise:** named people with real credentials next to your firm's name. - **A baseline and a record** of how often you are named, so you can see the asset grow. None guarantees a recommendation, but the earned parts carry most weight. [Muck Rack's May 2026 analysis](https://www.globenewswire.com/news-release/2026/05/07/3290268/0/en/generative-pulse-earned-media-consistently-drives-ai-citations-holding-at-84.html) of more than 25 million links cited by ChatGPT, Claude and Gemini found 84% came from earned media and 0.3% from paid content. And the asset can bring in business. For Helm Club, a founder membership community we work with, "A channel that did not exist before now brings one in five new members: founders who ask ChatGPT, Gemini or Claude which peer group to join are told about Helm." ([Helm Club case study](/work/helm-club)) The argument fails when the work builds nothing, as [why a cheap AI search package rarely works](/insights/cheap-ai-optimisation-doesnt-work) explains. ### Step 5: Agree what you will report and the milestones Promise measures, not outcomes you do not control. These are the ones we would put in a year-one paper: | Measure | What it tells the board | When it tends to move | |---|---|---| | Share of a fixed list of 20 to 50 buyer questions where you are named, counted only when it repeats | Whether the asset is growing | Within weeks in assistants that search the web live; months in answers from a model's memory | | Accuracy of how assistants describe you | Whether your facts are consistent | In our experience, often first | | Visits from AI assistants in Google Analytics | Supporting evidence | Alongside mentions | | Enquiries that mention an AI assistant, and their pipeline | The business result | Usually later in the year | Set milestones as targets, not promises: baseline and website fixes in the first quarter; more accurate descriptions and mentions for some specialist questions in the second; a rising share of mentions and the first AI-sourced enquiries in the second half; then a formal twelve-month review. Our guide to [building an AI visibility dashboard and board report](/insights/build-ai-visibility-dashboard) has a one-page format you can copy, and [how to tell which leads came from ChatGPT](/insights/attribution-ai-era-track-leads) covers the enquiry figures. ### Step 6: Say what would make you stop Most proposals leave this out, and it wins over cautious boards. For example: if the share of mentions across your fixed questions has not moved after nine months, the approach changes; if it has not moved after twelve, the budget goes back to the board. A defined exit turns an open-ended request into a controlled trial, and separates "this is not working" from "this is taking the time we said it would". ## Where should the budget come from? From whichever line the board expects to produce new clients, without switching off paid search, which brings enquiries quickly while slower work builds. Some of the money may already sit in your SEO budget. Google's own [guide to optimising for its generative AI features](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide) (May 2026, updated July 2026) says the work is still SEO, so a capable SEO provider may already cover part of the website side. SEO budgets rarely cover earned coverage and reviews. We compare the routes in [whether being named by ChatGPT is cheaper than Google Ads](/insights/economics-ai-visibility-vs-advertising). ## What objections should you expect? ### "How do we know our buyers use these tools?" Show your Step 1 evidence and ask sales how many recent prospects mentioned an assistant. Many buyers have one at work: [Microsoft 365 Copilot passed 30 million paid business seats in July 2026](https://www.microsoft.com/en-us/investor/earnings/fy-2026-q4/press-release-webcast), and Forrester found business buyers four times as likely as consumers to use Copilot. ### "We rank well on Google, so isn't that enough?" Ranking helps but does not guarantee a mention. [Ahrefs found in March 2026](https://ahrefs.com/blog/ai-overview-citations-top-10) that 38% of Google AI Overview citations came from top-ten pages, down from 76% in July 2025, because Google now runs several related searches for one question. Covering the follow-up questions matters too, as we explain in [why ranking first on Google does not mean ChatGPT mentions you](/insights/google-rankings-wont-protect-you). ### "Can't we just buy an advert in ChatGPT?" You can buy an advert, not a place in the answer. ChatGPT has shown labelled ads to UK users on its Free and Go plans since 11 August 2026, and [OpenAI says](https://openai.com/index/testing-ads-in-chatgpt/) the ads do not change the answer itself. Being named is still earned. ### "Most of our work comes from referrals." In our experience, referred buyers often check the name they were given with an assistant before they call. If it describes you vaguely or suggests alternatives, some referrals leak away. This often persuades a sceptical co-founder, because it is about keeping work you already win. ### "Can't we wait, or do it in-house?" You can wait, but competitors who start now will have a year of coverage and reviews you cannot buy back quickly; see [whether you can wait a year](/insights/first-mover-advantage-ai). Your team may do much of the work if it has the time, press contacts and technical knowledge; if not, [how to choose a firm to help](/insights/how-to-compare-ai-optimisation-providers) lists the questions to ask. ## How much should you ask for? A sum you can defend line by line, fixed for twelve months, with the break-even beside it. [What AI search optimisation costs in the UK](/insights/ai-search-optimisation-cost-uk) sets out published market ranges, and our own prices are on [what it costs](/what-it-costs). If the board is not ready for a year, ask for a fixed-scope first step such as our [Paid Discovery](/discovery): one week, every finding ranked, and fixes written so your own team can ship them. ## What should the board paper look like? Two pages, in this order: 1. **The situation:** buyers in our market ask AI assistants who to hire, and here is what they are told. 2. **The evidence:** the Step 1 table and two screenshots. 3. **The commercial link:** the board number this affects. 4. **The proposal:** what, by whom, over what period, and the five assets it builds. 5. **The cost and break-even:** the annual figure and extra clients needed, on gross profit. 6. **The measures:** the monthly report and milestones. 7. **The review and exit:** the twelve-month review and stop conditions. 8. **The decision sought:** one sentence. ## What to do this month 1. **Run the questions.** Put 20 buyer questions to six assistants, more than once each, and record who is named and how you are described. 2. **Talk to sales** about prospects who mentioned an AI assistant and pitches lost to firms named in every answer. 3. **Do the break-even sum** on client fees, gross margin, win rate and proposed spend. 4. **Write the two-page paper**, including the measures table and stop conditions. 5. **Pre-brief the most sceptical director**, so their questions are answered in the paper rather than in the room. ## Frequently asked questions ### How long should we ask the board to commit for? Two to three years in principle, with a formal review at twelve months and monthly reporting throughout. A shorter commitment tends to end just as results start, and the review gives the board a firm point to change course. ### What return can we expect in the first twelve months? Nobody can give you a reliable figure. In our experience year one brings more accurate descriptions, mentions for some specialist questions and the first enquiries that mention an assistant, while spend runs ahead of return. Judge year one on the leading measures and the return against the three-year break-even. ### What if the board wants a guaranteed return? Nobody can guarantee what ChatGPT or Gemini will say. A provider can guarantee what it controls: effort, reporting and, for outbound work, meetings held. That is why our outbound service, [Pipeline That Fills](/pipeline-that-fills), carries a guarantee tied to meetings held, while we make no promise about AI answers. Offer the board control instead, and be wary of anyone who guarantees a place in ChatGPT's answers. ### Can we put this spend on the balance sheet as an asset? Usually not. Marketing spend, including content and PR, is normally treated as an expense, and your accountant decides how each item is booked. Calling it an investment is about how you judge it: over a longer period, with early signals, expecting what you build to keep working. It is a management view, not an accounting one. ### What if we try it and nothing changes? Then your stop conditions do their job: change the approach at nine months and stop at twelve if nothing has moved. You will still have a clearer website, useful articles and a record of what buyers see. ### Does this make sense for a firm our size? For most founder-led B2B firms with £1m to £20m+ revenue, the break-even needs only a handful of extra clients over three years. The harder question is whether you can hold your nerve through year one. If you need enquiries next month, fund adverts or outbound first and run this alongside. A board that sees real evidence, a clear cost and a defined exit will usually give you room to try. If you want that evidence gathered before the meeting, [book a free Growth Review](/growth-review): 30 minutes with David Adams, and three fixes in writing afterwards. ## Sources - [G2, New G2 research: half of B2B software buyers now start their research with AI chatbots (April 2026)](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html) - [Forrester, B2B buyers make zero-click buying number one (January 2026)](https://www.forrester.com/blogs/b2b_buyers_make_zero_click_buying_number_one/) - [Pew Research Center, Google users are less likely to click on links when an AI summary appears (July 2025)](https://www.pewresearch.org/short-reads/2025/07/22/google-users-are-less-likely-to-click-on-links-when-an-ai-summary-appears-in-the-results/) - [Muck Rack, What is AI reading? Earned media drives 84% of AI citations (May 2026)](https://www.globenewswire.com/news-release/2026/05/07/3290268/0/en/generative-pulse-earned-media-consistently-drives-ai-citations-holding-at-84.html) - [Ahrefs, AI Overview citations from top-ten pages (March 2026)](https://ahrefs.com/blog/ai-overview-citations-top-10) - [OpenAI, Testing ads in ChatGPT (updated August 2026)](https://openai.com/index/testing-ads-in-chatgpt/) - [Google Search Central, Optimizing your website for generative AI features on Google Search (July 2026)](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide) - [Microsoft, FY26 fourth quarter earnings press release (July 2026)](https://www.microsoft.com/en-us/investor/earnings/fy-2026-q4/press-release-webcast) --- Title: How do we get ChatGPT to recommend our firm when buyers ask? URL: https://fortitudemedia.ai/insights/what-is-llm-optimisation Author: David Adams Published: 2026-04-22 Last reviewed: 2026-09-24 Topic: Getting named by AI assistants Short answer: Give the assistants three things: credible mentions of your firm on other sites, a website that says plainly what you do and for whom, and visible evidence of expertise. Assistants find firms from training data, which changes over months, and live search, which can change within weeks: ChatGPT search and Copilot draw on Bing, Google's AI features on Google. Ranking helps but does not guarantee a mention. Track 20 to 50 buyer questions monthly and count a win only when it repeats. For twenty years the question was "where do we rank on Google?" You paid an agency, they reported positions, and if you were on page one for the right terms the enquiries came. Then a prospect mentioned, almost in passing, that they had asked ChatGPT for a shortlist before they rang anyone, and your firm was not on it. You checked. They were right. That is the change in one sentence. Buyers ask ChatGPT, Gemini, Claude, Copilot and Perplexity who to hire, and the assistants answer with names. The audience is large: OpenAI said on [31 August 2026](https://openai.com/index/expanding-access-to-ai-with-chatgpt-ads/) that ChatGPT has more than a billion weekly users, and [Forrester found in January 2026](https://www.forrester.com/blogs/b2b_buyers_make_zero_click_buying_number_one/) that 94% of business buyers use AI somewhere in their buying process. Being one of the names builds on search but is not the same job. This guide explains what an assistant is doing when it names a firm, the two routes by which your firm gets into an answer, what the assistants look for, why a competitor gets named instead of you, and how to tell whether your work is paying off. ## Key takeaways - An AI assistant answers a buyer with a handful of names, so a firm that is left out is invisible for that question. - Firms get into answers by two routes: what the model learned in training, which changes over months, and live web search, which can change within weeks. - ChatGPT search and Copilot draw on Bing's index, while Gemini, AI Overviews and AI Mode draw on Google's, so ranking in both helps but does not guarantee a mention. - Assistants name the firms they are most confident about: firms mentioned credibly on other sites, with a website that says plainly what they do and visible evidence of expertise. - Nobody can buy a place inside the answer; ChatGPT has shown labelled ads to UK users on its Free and Go plans since 11 August 2026, and OpenAI says the ads do not change the answer. - Measure progress with a fixed list of 20 to 50 buyer questions, asked several times a month, counting a win only when it repeats. ## What is an AI assistant doing when it names a firm? It reads what it already knows and, increasingly, what it can find by searching, then writes an answer naming the firms it is most confident about. ChatGPT, Claude and Gemini are built on large language models: systems trained on vast amounts of text so they can understand a question and answer it in plain language. Perplexity is an answer engine that searches the web for each question and writes the answer with a language model. None of them returns ten links. They read the question, weigh the evidence and write an answer that names specific firms. Ask "which employment law firms should a manufacturing business in the Midlands talk to" and the answer names a few firms; how many varies by assistant, wording and user. If yours is among them, you have a recommendation from an intermediary the buyer trusts. ## Two routes into an answer: memory and live search An assistant can know about your firm because it learned about you in training, or because it found you by searching while it wrote the answer. The two routes work on very different timescales. | Route | How it works | Which assistants | How fast it changes | |---|---|---|---| | Memory (training data) | The model learned about firms from text gathered before its training cut-off | All of them, when they answer without searching | Only when the model is retrained, which takes months | | Live search | The assistant searches an index, reads the top pages and cites them | ChatGPT search (which has drawn on Bing), Copilot (Bing), Gemini, AI Overviews and AI Mode (Google), Perplexity and Claude, which search the web | Within weeks of a new page or article | For live search, your pages have to be in the right index and open to the right crawler: Bing for ChatGPT search and Copilot, Google for Google's AI features, and OpenAI's OAI-SearchBot, Anthropic's Claude-SearchBot and PerplexityBot allowed in. [How AI crawlers read your website](/insights/ai-crawlers-vs-google-spiders) shows how to check. For memory, what matters is how often and how consistently your firm was described across the web before the last training run, which changes only as others write about you. When an assistant does search, a clearer page or a new article can show up in its answers within weeks. ## How is being named different from ranking on Google? Google's classic results make you findable; an AI answer endorses a few firms. Ranking helps, but it does not guarantee a mention. Google's AI Overviews and AI Mode, ChatGPT search and Copilot all start from a search index, so a page that ranks is more likely to be read and cited. But the link is loose. [Ahrefs found in March 2026](https://ahrefs.com/blog/ai-overview-citations-top-10) that only 38% of the pages cited in AI Overviews ranked in the top ten for the search, down from 76% in July 2025, because Google now runs several related searches for one question (query fan-out). An assistant looking at a buyer's question also reads pages that answer the follow-ups: what it costs, how long it takes, who else does it. So covering the follow-up questions matters as much as ranking for the main one. [We rank first on Google, so why does ChatGPT not mention us?](/insights/google-rankings-wont-protect-you) covers that gap, and [our comparison of SEO and AI search](/insights/seo-vs-llm-optimisation) sets out which SEO work to keep, stop and add. ## What does an assistant look for before it names you? Three things: credible mentions of your firm elsewhere, a website that says plainly what you do, and visible evidence of expertise. Assistants are weighing patterns in what they read, and these three patterns decide most of it. ### Credible mentions on other sites An assistant is far more confident naming a firm that independent sources discuss. Muck Rack's [May 2026 analysis](https://www.globenewswire.com/news-release/2026/05/07/3290268/0/en/generative-pulse-earned-media-consistently-drives-ai-citations-holding-at-84.html) of more than 25 million links cited by ChatGPT, Claude and Gemini found 84% came from earned media, meaning coverage by others, and 0.3% from paid content. Review sites count too: in [G2's April 2026 survey](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html) of software buyers, 45% said citations of review sites were the most confidence-inspiring signal in an AI answer. Trade press, professional directories, awards, partner pages and client case studies all tell the assistant that others have checked you. Brand recognition built this way helps with assistants as well as with Google. [Which kinds of PR win work for a B2B firm](/insights/pr-for-b2b-what-works) covers how to earn it without a large budget. ### A website that says plainly what you do If your description could apply to a hundred other firms, your services are three clicks deep and your expertise is implied, the assistant cannot confidently say what you do or who you are for. Say it in the first screen of the homepage and every service page: what you do, for whom, where and why you are different. Structured data, a clear heading hierarchy and a logical site structure make that easier for a machine to read; [whether you need schema markup](/insights/structured-data-schema-markup-ai) explains what it does and does not do. ### Visible evidence of expertise If your people have published considered work on your subject, answered the hard questions and named themselves as the authors, the assistant can see it. Thin, generic content does the opposite. Real examples, specific figures with sources and plain discussion of trade-offs are what separate your pages from the rest. [What E-E-A-T is and why AI cares](/insights/eeat-why-ai-cares-more) goes into how assistants judge this. ## Why does ChatGPT recommend my competitor and not me? Usually because the competitor gives the assistant more to work with on the specific question asked. Work through the likely causes in this order, because each makes the next irrelevant until it is fixed. | Likely cause | How to check | What to do | |---|---|---| | The assistant cannot reach or find your pages | Check robots.txt and CDN settings for blocked AI crawlers; search Bing for your key pages | Allow the search crawlers; verify your site in Bing Webmaster Tools | | Your site does not say plainly what you do | Read your homepage as a stranger for thirty seconds | Rewrite the first screen of the homepage and each service page | | Nobody else writes about you | List every third-party mention: press, directories, reviews, partners | Earn a few credible mentions in the places your buyers read | | The competitor answers the follow-up questions and you do not | Compare their guides and service pages with yours on cost, timing and fit | Write the page that answers each follow-up completely | | Sources disagree about you | Compare your name, services and founding facts across your site, LinkedIn and directories | Make every profile say the same thing | Old search tactics can also count against you: hundreds of near-identical keyword pages read as a site built to game a system. [Why ChatGPT ignores sites with hundreds of pages](/insights/ai-penalises-thin-content) covers the fix, and [why ChatGPT describes a firm wrongly](/insights/ai-conflicting-information-online) covers inconsistent facts. ## Does ChatGPT give different recommendations to different people? Yes. Answers vary from one run to the next, with the exact wording of the question, and with what the assistant knows about the person asking, such as their location or, where it is switched on, ChatGPT's memory of earlier chats. That is why a single test proves little. A founder who asks once and sees their firm named, or not named, has learned very little. What counts is how often you are named across a fixed set of questions asked repeatedly in clean sessions. ## Which matters more for B2B: ChatGPT or Google AI Overviews? Both, in that order, then Copilot and Gemini, then Perplexity and Claude. The reach figures explain why. | Assistant | Reach | Why it matters for B2B | |---|---|---| | ChatGPT | More than a billion weekly users (OpenAI, August 2026) | The standalone assistant with the largest audience | | Google AI Overviews and AI Mode | AI Overviews over 2.5 billion monthly users, AI Mode over 1 billion ([Alphabet, June 2026](https://blog.google/alphabet/investor-presentation-june-2026/)) | Shown inside ordinary Google searches, so buyers see it without choosing to | | Microsoft Copilot | Microsoft 365 Copilot passed 30 million paid business seats ([Microsoft, July 2026](https://www.microsoft.com/en-us/investor/earnings/fy-2026-q4/press-release-webcast)) | Built into the tools many buyers use at work; Forrester found business buyers four times as likely as consumers to use Copilot | | Gemini | Gemini app over 900 million monthly users (Alphabet, June 2026) | Draws on Google Search, so Google work carries over | | Perplexity and Claude | Smaller audiences | Both search the web, so the same work covers them | Because ChatGPT search and Copilot lean on Bing and Google's features lean on Google, one programme covers most of them. [What Google AI Mode is and how it differs from AI Overviews](/insights/end-of-ten-blue-links) covers Google's side, and [which sources ChatGPT cites when a B2B buyer asks who to hire](/insights/ai-optimisation-b2b-vs-b2c) covers the B2B pattern. ## Can we pay to get our firm recommended by ChatGPT? Not inside the answer. ChatGPT has shown labelled ads to UK users on its Free and Go plans [since 11 August 2026](https://openai.com/index/testing-ads-in-chatgpt/), and OpenAI says the ads do not change the answer itself. Plus, Pro, Business, Enterprise and Education plans show no ads. Being named in the answer is still earned, which is why the three things above matter more than any ad budget. Be wary of anyone who says they can guarantee a place in ChatGPT's answers. Nobody can guarantee what an AI assistant will say. A provider can guarantee effort, reporting and outcomes it controls, such as pages published or meetings held. ## How do we know whether it is working? Track a fixed list of 20 to 50 questions your buyers ask, in their words. Put each to ChatGPT, Gemini, Claude, Copilot, Perplexity and Google's AI Mode several times a month in a clean session, because answers vary from run to run, and count a win only when it repeats. Record whether you are named, where in the answer, and how you are described. [How to find out whether ChatGPT recommends your firm](/insights/track-ai-recommending-your-business) shows how to do this by hand, and our [example monthly report](/samples/monthly-report) shows one way to present it. Two of our clients show what the results look like when the method is kept up. [ID-Pal](/work/id-pal), an identity verification firm, started from a familiar place: "AI could describe ID-Pal. It never recommended it." Its measurement: "The same buyer questions run every month in ChatGPT, Gemini and Claude, in a fresh chat. A win only counts when it repeats, and losses are reported too." The result so far: "UK buyer questions where AI names ID-Pal: none in May, two in August, four in September 2026." [Helm Club](/work/helm-club) runs chaired peer groups for founders of growing UK businesses. "A channel that did not exist before now brings one in five new members: founders who ask ChatGPT, Gemini or Claude which peer group to join are told about Helm." Its reporting is kept to what the business can check: "Qualified calls a week, reported on one page every Monday and checked against Helm's own member records rather than a platform dashboard." Alongside the question list, ask every new enquiry how they found you, because many buyers read an answer and then search your name. ## What does an effective programme look like? Three things working together: a website built to be understood, content that shows real expertise, and a steady programme of earned coverage. Any one alone moves slowly. The website is the foundation, the content is the proof, and the coverage is the corroboration that lets an assistant check your claims against someone else's account. You can do this yourself with a plan and discipline; for a guide to each part, see [being named by AI assistants](/insights/pillar/ai-llm-optimisation). ## What to do this month 1. Read your homepage as a stranger and check whether the first screen says what you do, for whom and why you are different. Rewrite it if not. 2. Check your robots.txt and CDN settings let OAI-SearchBot, Claude-SearchBot and PerplexityBot in, and verify your site in Bing Webmaster Tools. 3. List every place your firm is mentioned by a third party: press, directories, reviews, client and partner sites. If the list is short, that is your first gap. 4. Find your three most substantive pieces of published content and your thirty thinnest. Plan to strengthen the first and merge or retire the second. 5. Write down 20 to 50 questions your buyers ask and run them through the main assistants to set a baseline. 6. Decide who owns this, because a programme that is nobody's job stops in month two. ## Frequently asked questions ### How much does it cost to get our business recommended by AI? It depends on how much of the work you already have: a clear website, published expertise and outside coverage. Some firms need only rewritten pages and a few earned mentions; others need a rebuild. [What AI search optimisation costs in the UK](/insights/ai-search-optimisation-cost-uk) sets out the market, and our own prices are on [what it costs](/what-it-costs). ### We are a niche or regional firm. Does this still apply? Yes, and often more strongly. When someone asks for a specialist in a narrow field or a particular region, the assistant still has to name someone, and the pool is smaller. A firm with clear positioning and a few credible third-party mentions can own a niche question in a way that is impossible for a broad one. ### How do we get third-party mentions without a big PR budget? Start with what you already have: client case studies with permission, your professional body's directory, sector awards, trade publications that need expert comment, review profiles and partner websites. Five credible mentions in the places your buyers read do more than fifty in the wrong ones. ### How long before we see a change? Assistants that search the web live (ChatGPT search, Perplexity, Copilot and Google's AI features) can reflect a new page or article within weeks. Answers drawn from a model's memory change only when it is retrained, which takes months. ID-Pal's case study gives a dated example: "UK buyer questions where AI names ID-Pal: none in May, two in August, four in September 2026." ### Do our Google rankings become irrelevant? No. Rankings feed AI Overviews, AI Mode and, through Bing, ChatGPT search and Copilot, so they still help. They just do not guarantee a mention, and the tactics that once produced them cheaply, such as keyword pages and bought links, do nothing for an assistant. The firms that get named are rarely the biggest. They are the ones whose expertise is written down, corroborated and easy to read. To see where you stand today, [book a free Growth Review](/growth-review): 30 minutes with David Adams and three fixes in writing. ## Sources - [OpenAI, A milestone in expanding access to AI (August 2026)](https://openai.com/index/expanding-access-to-ai-with-chatgpt-ads/) - [OpenAI, Testing ads in ChatGPT (updated August 2026)](https://openai.com/index/testing-ads-in-chatgpt/) - [Forrester, B2B buyers make zero-click buying number one (January 2026)](https://www.forrester.com/blogs/b2b_buyers_make_zero_click_buying_number_one/) - [Ahrefs, AI Overview citations from top-ten pages (March 2026)](https://ahrefs.com/blog/ai-overview-citations-top-10) - [Muck Rack via GlobeNewswire, Earned media consistently drives AI citations (May 2026)](https://www.globenewswire.com/news-release/2026/05/07/3290268/0/en/generative-pulse-earned-media-consistently-drives-ai-citations-holding-at-84.html) - [G2 via PR Newswire, Half of B2B software buyers now start their research with AI chatbots (April 2026)](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html) - [Google, Alphabet investor presentation (June 2026)](https://blog.google/alphabet/investor-presentation-june-2026/) - [Microsoft, FY26 fourth quarter earnings (July 2026)](https://www.microsoft.com/en-us/investor/earnings/fy-2026-q4/press-release-webcast) --- Title: How does a B2B tech company get recommended when buyers ask ChatGPT? URL: https://fortitudemedia.ai/insights/ai-optimisation-technology-companies Author: Ashley Kenerson Published: 2026-04-22 Last reviewed: 2026-09-24 Topic: Sector guides Short answer: ChatGPT recommends software it can match to a buyer's situation and check elsewhere: a clear use case and customer, public documentation, integrations, security details and pricing, fair comparison pages and independent reviews on sites such as G2. Most SaaS websites say too little in plain text and hide the rest behind forms or scripts. State who the product is for, open up your documentation and pricing model, publish fair 'alternatives to' pages and get reviewed where your buyers read. Buyers of business software now ask an AI assistant before they ask a salesperson. In [G2's April 2026 survey of 1,076 software buyers](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html), 51% said they start their research in an AI chatbot rather than Google, up from 29% a year earlier, and 69% had chosen a different vendor from the one they first planned because of what a chatbot told them. G2 runs a review site, so treat the figures as directional. For example, the operations director of a housing association has to replace the repairs scheduling system before the contract renews. Before anyone books a demo, she asks ChatGPT for the best repairs scheduling systems for UK housing associations and how they compare. She gets five names and a comparison table. Your product was built for housing associations and several of them use it. It is not in the table. Tech companies often assume they are ahead here because they are technical. In our experience the opposite is common: a JavaScript site most AI crawlers cannot fully read, documentation behind a login, pricing that says "contact us" and a homepage about "the intelligent platform for modern teams". This guide is for the founder of a UK B2B software or SaaS company with £1m to £20m+ in revenue. ## Key takeaways - Half of software buyers now start their research in an AI chatbot, so the first shortlist is often drawn up before anyone visits your website. - AI assistants name products they can match to a buyer's situation: an industry, a company size, the systems already in use and a problem. - Documentation, integrations, security details and pricing hidden behind a login, a form or a script do not exist for the assistant. - Reviews on G2 and similar sites, marketplace listings and forum threads carry more weight than anything you say about yourself. - A fair, dated "alternatives to" page that admits where a rival fits better gets quoted; a table you win on every row reads as an advert. - After a rebrand or an acquisition, tell search engines and assistants plainly that the old and new names are one company, or they will treat them as rivals. ## What do buyers ask ChatGPT when they are choosing software? They ask for shortlists, alternatives, comparisons and reassurance, in the language of their own industry, and the person asking is often not technical. - "What is the best practice management software for a 30-person accountancy firm in the UK?" - "What are the alternatives to our current field service scheduling system?" - "Does this CRM integrate with Xero and Microsoft 365?" - "Is this software ISO 27001 certified, and is our data stored in the UK?" - "How much does field service software cost per user for a team of 40 engineers?" - "Which document management systems are on G-Cloud and hold Cyber Essentials Plus?" The questions name an industry, a size and often a place, so a product described in abstract terms cannot be matched to them. Many ask for alternatives, which assistants answer from review sites and forums as much as from vendor websites. The rest are hygiene questions about integrations, security and price, and if your answer is not public, you may be crossed off by default. Some buyers now let an AI agent build the shortlist, which we cover in [what happens when buyers let an AI agent build the shortlist](/insights/ai-agents-change-b2b-buying). ## How does ChatGPT decide which software to name? It names products it can match to the buyer's situation and confirm from several sources. Four kinds of evidence matter most: - **A clear use case and customer**, in words the buyer would use. - **Readable product detail**: documentation, integrations, security and pricing on public pages. - **Independent opinion**: reviews, marketplace listings, trade press and forum discussion. - **Consistency**: the same description on your website, LinkedIn, review sites, marketplaces and Companies House. Ranking on Google helps but does not guarantee a mention. Google's AI features, ChatGPT search and Microsoft Copilot all start from a search index, yet [Ahrefs found in March 2026](https://ahrefs.com/blog/ai-overview-citations-top-10) that only 38% of AI Overview citations came from pages in Google's top ten, down from 76% in July 2025, because Google now runs several related searches for one question. So rank for your category, and also answer what the buyer asks next: integrations, security, migration and price. Positioning matters for the same reason. "An intelligent operations platform for modern teams" could describe hundreds of products; "repairs scheduling software for UK housing associations" is the sentence a buyer's question will match. Each situation you want to win needs its own page in the buyer's words. ## What has to be public on your website? Everything a buyer checks before a demo, in plain text on public pages: - **Documentation.** Public feature guides and setup articles are some of the most trusted material an assistant can find, because they describe what the product actually does. - **Trial documentation.** What a trial or pilot includes, how long it runs, what data a customer brings and how setup works. - **Integrations.** One page per integration, with a sentence on what it does: "invoices sync with Xero both ways, several times an hour". - **Security.** Certifications such as ISO 27001, SOC 2 or Cyber Essentials Plus, where data is hosted, your sub-processors and how you handle UK GDPR. Then check the site can be read at all. The main AI crawlers, GPTBot, ClaudeBot and PerplexityBot, fetch JavaScript files but do not run them ([Vercel and MERJ, December 2024](https://vercel.com/blog/the-rise-of-the-ai-crawler)), so text that only appears after scripts run is invisible to ChatGPT, Claude and Perplexity. Googlebot does render JavaScript, which helps Gemini and Google's AI features. Our guide to [how AI crawlers read your website](/insights/ai-crawlers-vs-google-spiders) shows how to check this and whether your hosting blocks AI crawlers. ## Should a SaaS company publish its pricing? Yes, at least the model and a starting point. "Contact us" does not stop an assistant answering a cost question. It only means the answer comes from somewhere else: a review site's pricing section, a reseller or an old forum thread. Assistants repeat the unit and number they find, so make both unambiguous: - **Per seat.** Price per user, billing period, minimum seats and whether annual billing changes the rate. - **Usage-based.** Name the unit (per transaction, per 1,000 messages) and give one worked total, so the assistant has a figure to quote rather than a formula. - **"From" pricing.** Say what the "from" price buys and who usually pays more. An assistant will quote the lowest figure it sees as the price. Example: a fictional scheduling product could set out its plans like this, as an HTML table rather than an image. | Plan | Who it suits | How it is charged | Price (excluding VAT) | Term | |---|---|---|---|---| | Starter | Up to 10 users | Per user, per month | £10 per user a month | Monthly | | Growth | 10 to 100 users | Per user, billed annually | £20 per user a month | 12 months | | Enterprise | Over 100 users or several sites | Annual licence | From £20,000 a year | 12 months or longer | | Text reminders | Any plan | Per 1,000 messages | £10 per 1,000 messages | Monthly in arrears | Date the page and keep the same figures on your marketplace listings and review profiles. ## How do you get named when buyers ask for alternatives to a competitor? Publish a fair, dated page for each competitor you meet most often, one that compares options without reading like an advert. Example: an outline for "Alternatives to RivalCo for UK housing associations", where RivalCo stands for your most frequent competitor. 1. **Title and date**: the competitor, the buyer type and "reviewed September 2026". 2. **A two-sentence answer at the top**: who RivalCo suits, who it suits less well, and three to five alternatives, including yours. 3. **A comparison table**: one row per option, including at least one that is not you, covering who it suits, pricing model, main integrations and data hosting. 4. **When RivalCo is the better choice**, said plainly. This is what makes the page believable. 5. **When yours is the better choice**, backed by a customer story, a review or a documented integration. 6. **What switching involves**: data export, timescale, notice periods. 7. **How the facts were gathered**, with the date each was checked and an address for corrections, under a named author. Stick to facts you can evidence. An unsupported claim about a competitor's performance damages your credibility and can create legal risk. ## G2 or Capterra: which review sites matter most? G2 is the review site named most often in the citation studies we can verify, so for most B2B SaaS firms it is the first profile to get right. After that, where your own buyers read reviews matters more than any study. | Review site | What the evidence says | |---|---| | G2 | 1.1% of all ChatGPT citations in [Profound's 2025 study](https://www.tryprofound.com/blog/ai-platform-citation-patterns) of 680 million citations, listed with Wikipedia, Reddit and Forbes among its top sources; prominent for B2B queries in Perplexity in [Peec AI's March 2026 study](https://searchengineland.com/ai-search-engines-cite-reddit-youtube-and-linkedin-most-study-473138) | | Gartner Peer Insights | gartner.com as a whole made up 7.0% of Perplexity's top ten sources in Profound's study | | Capterra, GetApp, Software Advice | Not named in the citation studies we could verify; worth completing if your buyers are smaller firms comparing tools | | TrustRadius | Not named in those studies; its July 2026 survey found 74% of tech buyers use reviews | Buyers look for these sources inside AI answers: in G2's survey, 45% said citations of review sites were the most confidence-inspiring signal in an AI answer. To see which sites Perplexity cites in your category, run three prompts such as "best field service management software UK", "best contract management software for UK law firms" and "alternatives to RivalCo for small teams", and open the sources under each answer. Note which review sites appear, and repeat the check monthly because answers vary. ## Which other outside sources confirm what you say? - **Marketplaces.** Microsoft AppSource, AWS Marketplace, Salesforce AppExchange and the Xero App Store each confirm what you integrate with. - **Public sector frameworks.** G-Cloud 15 went live on 6 August 2026, run by the [Government Commercial Agency](https://www.gca.gov.uk/agreements/RM1557.15), formerly the Crown Commercial Service, and buyers shortlist on its Digital Platform. Name the framework and lot on your site, for example "G-Cloud 15, Lot 2b: Other Software as a Service". - **Security credentials.** Cyber Essentials and Cyber Essentials Plus are the [NCSC's certification schemes](https://www.ncsc.gov.uk/cyberessentials/overview), and current certificates can be looked up on the website of IASME, its delivery partner. Add ISO 27001 and, for security testing firms, CREST accreditation. - **Communities.** Reddit, GitHub and specialist forums. A useful answer from one of your engineers in the right thread beats a sponsored post. - **Trade press.** Computer Weekly, Computing and UKTN and, just as important, the trade press of the industries you sell into. ## Worked example: how ID-Pal went from described to recommended [ID-Pal](/work/id-pal) sells identity verification (KYC and KYB) to fund administrators, wealth managers, law firms and accountants, and had bought NorthRow, a UK business verification platform. "AI could describe ID-Pal. It never recommended it." | Stage | What we recorded | |---|---| | Where it started | "Across 36 recorded AI answers to the questions ID-Pal's buyers actually ask, ID-Pal was named three times, and only for its brand or its Salesforce integration. AI knew who it was, and had no reason to recommend it." | | The cause | "The redirect from northrow.com was technically correct, but nothing on either site told Google or the AI engines that NorthRow and ID-Pal were the same company." | | The first fix | "About four hours of structured data work declaring NorthRow part of ID-Pal on both domains. Live in week one, with no copy changes and no risk to the redirect." | | The pages | "Five industry pages rebuilt: fund administration, wealth management, legal, accountancy and estate agency. Each opens with the question a compliance lead types, names the regulator, and answers inside the first ninety words." | | The result | "UK buyer questions where AI names ID-Pal: none in May, two in August, four in September 2026." | The lesson for any SaaS firm: "Where a page has been rebuilt, ID-Pal appears. Where it has not yet, it does not, which tells us exactly what to build next." ## How do you stop ChatGPT using an old product name after a rebrand? Tell every source, in the same words, that the old name and the new one are the same company. If nothing connects them, assistants treat them as two firms, sometimes as rivals. At ID-Pal: "AI answers still recommending NorthRow as a rival: six in May, one in September 2026." 1. **Redirect page by page**, each old address to its closest new page. 2. **Declare the relationship in structured data** on both domains. 3. **Publish a plain page** saying the old product is now part of the new one, when and what changed. 4. **Update every profile you control**: review sites, marketplaces, LinkedIn, partner directories and Companies House. Assistants that search the web live, such as ChatGPT search, Perplexity, Copilot and Google's AI features, can reflect the change within weeks. Answers drawn from a model's own memory change only when it is retrained, which takes months. ## How do you tell whether it is working? Track a fixed list of 20 to 50 buyer questions: category shortlists, "alternatives to" your main competitors, and a few on security, integrations and price. Ask each several times a month in ChatGPT, Gemini, Claude, Copilot and Perplexity, in a clean session, because answers vary from run to run. Record whether you are named, who is named instead and whether the details are right, and count a win only when it repeats. Add a free-text "how did you hear about us" field to your demo form and read it weekly. ## What to do this month 1. **Check the assistants can read you**: robots.txt, hosting settings and key pages that show their text without scripts. 2. **Write one "who it is for" page per core customer type**, in their words. 3. **Publish your pricing model, a security page and an integrations page** in plain text. 4. **Write one fair "alternatives to" page** for the competitor you meet most. 5. **Ask ten recent customers for reviews** on G2 or wherever your buyers research, straight after a go-live or renewal. 6. **Set up your question list** and run it in clean sessions several times a month. ## Frequently asked questions ### Is gating our whitepapers and case studies hurting our visibility in AI search? Yes, for anything behind the form. An assistant cannot fill in a form, so a gated PDF does not exist for it. Publish case studies as open web pages, and for a whitepaper publish a page with the main findings in plain text. Keep the form for the full download if you must. ### Should we make our documentation public? Most of it. Feature guides, setup articles and integration notes help buyers and give assistants accurate material. Keep anything security-sensitive or customer-specific behind a login. Competitors can usually sign up for a trial anyway. ### How many reviews do we need on G2 or Capterra? There is no magic number; recency and detail matter more than volume. Reviews that mention the reviewer's industry, company size and what the product solved give an assistant something to match. Ask after a successful go-live or renewal. ### Do analyst reports still matter if buyers use ChatGPT? Less than reviews for most firms with £1m to £20m+ in revenue. [TrustRadius's July 2026 survey](https://www.prnewswire.com/news-releases/trustradius-2026-b2b-buying-disconnect-report-reveals-ai-has-changed-how-buyers-research-but-not-what-they-trust-302825792.html) of 1,862 tech buyers found 74% use reviews and only 13% consulted analyst reports. Of those buyers, 63% had used AI in their purchase, and 94% of them fact-checked it at least sometimes. Put reviews, marketplace listings and customer stories first. ### We run an IT services firm or MSP. Does this apply to us? The principles do, but your buyers ask local questions about service, such as which IT support firms in a city work with law firms. State your region and sectors, publish service levels in plain words, show Cyber Essentials and partner designations, and ask clients for Google reviews that mention their sector. Our [guide for management consultancies](/insights/ai-optimisation-management-consultancies) covers the similar challenge of describing a service. The software companies that get named are the ones whose product can be read, checked and matched to a buyer's situation. [Website That Sells](/website-that-sells) rebuilds a site so buyers and assistants can both see what you do and who it is for, and a free [Growth Review](/growth-review) shows you what ChatGPT says about you today. ## Sources - [G2, New G2 research: half of B2B software buyers now start their research with AI chatbots (April 2026)](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html) - [TrustRadius, 2026 B2B Buying Disconnect Report (July 2026)](https://www.prnewswire.com/news-releases/trustradius-2026-b2b-buying-disconnect-report-reveals-ai-has-changed-how-buyers-research-but-not-what-they-trust-302825792.html) - [Profound, AI platform citation patterns (June 2025)](https://www.tryprofound.com/blog/ai-platform-citation-patterns) - [Search Engine Land, AI search engines cite Reddit, YouTube and LinkedIn most: study (March 2026)](https://searchengineland.com/ai-search-engines-cite-reddit-youtube-and-linkedin-most-study-473138) - [Ahrefs, How many AI Overview citations come from the top 10 (March 2026)](https://ahrefs.com/blog/ai-overview-citations-top-10) - [Vercel, The rise of the AI crawler (December 2024)](https://vercel.com/blog/the-rise-of-the-ai-crawler) - [Government Commercial Agency, G-Cloud 15 (September 2026)](https://www.gca.gov.uk/agreements/RM1557.15) - [NCSC, Cyber Essentials overview (September 2026)](https://www.ncsc.gov.uk/cyberessentials/overview) --- Title: How does a financial planning firm get named when clients ask ChatGPT? URL: https://fortitudemedia.ai/insights/ai-optimisation-financial-advisory Author: Ashley Kenerson Published: 2026-04-22 Last reviewed: 2026-09-24 Topic: Sector guides Short answer: ChatGPT names financial planning firms it can match to a client's situation and check elsewhere: a clear niche such as employers or founders planning an exit, plain guides by named, qualified advisers, and a firm name that matches the FCA Register, VouchedFor, Unbiased and Google. Keep public guides general and fair, clear and not misleading, since targeted support now sits between information and advice. Say whether you are independent or restricted and how you charge. For example, the finance director of a 200-person engineering firm has been asked to review the company's workplace pension and employee benefits before the renewal. Her chief executive, who owns most of the shares, wants to sell the business within three years and needs to plan what happens to the proceeds. Before calling anyone, each asks ChatGPT who can help, and each gets a checklist and four firms. Your firm does exactly this work for local companies. It is not one of the four. This is now the normal first step for business buyers. [Forrester found in January 2026](https://www.forrester.com/blogs/b2b_buyers_make_zero_click_buying_number_one/) that 94% of business buyers use AI in their buying process, and that they are four times as likely as consumers to use Microsoft Copilot, which many have at work. Referrals from accountants, solicitors and existing clients still matter, but the people referred now check the name with an assistant, and people without a referral start there. Advisers are often among the least visible professionals in these answers, largely because compliance worries stop them publishing anything specific. The result is a website that promises "holistic, independent, bespoke financial planning" and little else. This guide is for the founder or managing director of a UK financial planning or wealth management firm who wants to be named when the right business and private clients ask. ## Key takeaways - Business owners, finance directors and private clients now check advisers with ChatGPT and ask it who specialises in their situation before they book a first meeting. - AI assistants name advice firms with a clear client niche, such as employers reviewing a workplace pension or founders planning an exit, not firms offering "holistic financial planning". - Your FCA Register entry, chartered or CFP status and reviews on VouchedFor, Unbiased or Google are what the assistants use to check you, so the names on each must match. - General education is still the safest content to publish, but whether a page is a financial promotion depends on what it says, and the FCA's targeted support regime has moved the line since 6 April 2026. - Saying plainly whether you are independent or restricted, and how you charge, answers two of the questions clients ask assistants most. - Start by choosing the client type you serve best, publishing three guides to their real decisions and making every listing agree. ## What do clients ask ChatGPT when they are looking for an adviser? They ask how to choose, what it costs and who understands their situation, usually at a moment of change. Business and private clients ask different questions: | Business owners and employers | Private clients | |---|---| | "Which financial advisers in Bristol help employers review a workplace pension for 200 staff?" | "How do I choose a financial adviser in the UK, and what should I ask them?" | | "How should I plan my personal finances before selling my company?" | "What is the difference between an independent and a restricted financial adviser?" | | "Which financial planners are best for company directors planning an exit?" | "How much does ongoing financial advice cost?" | | "What should an employee benefits adviser do for a company with 100 staff?" | "Which advisers understand NHS pensions for hospital consultants?" | | "How can directors take money out of a limited company in the most sensible way?" | "Which wealth managers near Leeds are chartered and have good reviews?" | Questions about how advice works and what it costs are answered from whoever has explained them most clearly. Questions that ask for a firm describe a type of client and a place, and only match firms whose websites describe that client in the same terms. We have not listed who ChatGPT names for these prompts, because the answer changes by city and from one run to the next; run them yourself in a clean session and note who appears. ## How does ChatGPT decide which advisers to name? It names firms it can match to the client's situation and confirm from independent sources. On money questions assistants are especially careful, and five kinds of evidence matter: - **A clear client niche**, described the way clients describe themselves. - **Guides to that niche's decisions**, written by named advisers and dated. - **Qualified, named people**, with qualifications such as Chartered Financial Planner or CFP and the clients they work with. - **Regulatory clarity**: a Register entry whose firm name, trading names, status and principal match your website. - **Client voice**: reviews on VouchedFor, Unbiased and Google that describe the client's situation and what changed. Many firms work under a network, act as appointed representatives, or have merged and kept old trading names. If your website, the Register and the review sites describe you differently, the assistant has several versions of you to reconcile, and where it cannot, it hedges or leaves you out. ## Why does a client niche matter more than holistic planning? Because clients ask about their situation, and an assistant matches situations to specialists. A firm clearly known for one or two client types will be named for those far more often than a generalist is named for anything. Most firms already have a centre of gravity: - **Employers**: workplace pension reviews, employee benefits and financial education for staff. - **Business owners before and after a sale**: extracting profit, planning the exit and deciding what to do with the proceeds. - **Families with wealth across generations**: estate planning, gifting and trusts, alongside solicitors. - **People approaching retirement**: income planning and pension decisions. - **Professionals with complex pensions**, such as NHS consultants or senior civil servants. Let the ones you are referred for most lead the website. A useful niche guide takes one decision and walks through it: the options, the trade-offs, the common mistakes, the questions to ask and when advice is worth paying for. "What to do in the first six months after selling your business" or "How to review a workplace pension scheme before renewal" are titles that match real questions. Each guide should name its author, say which tax year it reflects and be dated. Our article on [what makes ChatGPT treat content as expert rather than marketing](/insights/what-makes-expert-quality-content) explains why depth matters. ### Which financial planners are best for company directors planning an exit? In our view, the ones who start early and work alongside the corporate finance adviser, the solicitor and the accountant. An assistant can only give that answer about your firm if your site says it: that you plan for directors two to three years before a sale, what you do at each stage, the kinds of businesses you have worked with (anonymised) and which advisers in your region you work alongside. A page that sets that out, with a named adviser and a date, is what gets matched to the question. ## Where does education end and targeted support begin? Guides that explain how options work are information, but they must still meet the FCA's rules where they promote your services, and targeted support now sits between information and advice. Assistants reward the careful, general guidance the FCA expects. A page becomes a financial promotion when it is an invitation or inducement to engage in financial activity, communicated in the course of business, and every promotion must be [fair, clear and not misleading](https://www.fca.org.uk/firms/financial-promotions-adverts) whatever the medium, including websites and social media. A guide that ends "book a pension review with us" can be a promotion, so it goes through your financial promotions process. Since 6 April 2026, firms with permission can offer targeted support under the FCA's policy statement [PS25/22 (December 2025)](https://www.fca.org.uk/publications/policy-statements/ps25-22-consumer-pensions-investment-decisions-rules-targeted-support): suggestions designed for groups of consumers with common characteristics, for pensions and retail investments, which are not personal recommendations. | | General information | Targeted support | Advice | |---|---|---|---| | What it is | How options work, the trade-offs and what to ask | Suggestions for groups of consumers with common characteristics | A personal recommendation for one client's circumstances | | Permission | None beyond your usual rules; promotions must be fair, clear and not misleading | Firms must apply for the new permission | Advice permissions | | In a published guide | "How pension drawdown works and what to ask an adviser" | Needs a compliance decision before any page tells a defined group what to do | Not possible in a public guide | In practice: keep public guides general, say who they are written for, note that individual circumstances need individual advice, and if a page starts telling a defined group of readers what they should do, ask your compliance lead which side of the line it sits on. ### Say whether you are independent or restricted, and how you charge Clients ask assistants what the difference means. [MoneyHelper](https://www.moneyhelper.org.uk/en/getting-help-and-advice/financial-advisers/choosing-a-financial-adviser) describes independent advisers as able to "consider products from across the whole market", and restricted advisers as recommending "products from a certain number of providers and or within a specific product area". Explain yours in plain words. On charges, the Consumer Duty's [price and value outcome (PRIN 2A.4)](https://www.handbook.fca.org.uk/handbook/PRIN/2A/4.html) is about the relationship between what a client pays and the benefits they can reasonably expect, and the [consumer understanding outcome (PRIN 2A.5)](https://www.handbook.fca.org.uk/handbook/PRIN/2A/5.html) requires communications that help customers make decisions that are "effective, timely and properly informed". A clear page on your initial and ongoing charges, with worked examples, serves both. MoneyHelper notes that many advisers do not list their fees online, so a firm that does stands out. Agree a short set of content rules with your compliance lead or network at the start: what needs sign-off, which risk warnings apply and how tax figures are dated. Insurance brokers face the same tension between publishing and compliance, as our [guide for insurance brokers](/insights/ai-optimisation-insurance-brokers) shows from their side. ## Worked example: an appointed representative with four different names Example: a fictional planning firm that is an appointed representative of a network, and rebranded two years ago. | Where the firm appears | What it showed | The fix | |---|---|---| | FCA Register | The limited company's legal name, as an appointed representative of the network, at the old office | Ask the principal to update the address, and use the Register name on the website | | Website | The new brand only, with no company name or principal | Add one sentence giving the trading name, the company name, the FCA reference number and the principal firm | | VouchedFor | Adviser profiles still under the pre-rebrand trading name | Update the firm name on every adviser profile | | Unbiased | A second, older profile from the founder's previous practice | Close the old profile and complete the current one with the Register and trading names | | Google Business Profile | The new brand at the old office | Correct the address to match the Register | Before the fix, an assistant sees four names and two addresses and cannot be sure they are one firm. Treat the Register as the master version and make every other listing copy it. ## Which outside sources matter most for an advice firm? The ones that regulate, qualify and review you: - **The FCA Register.** Firm name, trading names, status and address, all matching your website. - **Qualifications.** Chartered Financial Planner and chartered firm status through the Chartered Insurance Institute's Personal Finance Society, CFP certification and Chartered Wealth Manager through the CISI, and SOLLA accreditation for later-life advice. Show them on individual profiles. - **Directories and reviews.** MoneyHelper, the government-backed guidance service, points people looking for an adviser to Unbiased, VouchedFor, SOLLA and the Personal Finance Society. Add Google reviews that describe the client's situation. - **Trade press.** FT Adviser, Money Marketing, Professional Adviser and Citywire's New Model Adviser and, for business work, Corporate Adviser and Employee Benefits. - **Your niche's own press and events.** A column in a regional business magazine, or a talk to a medical society about pensions, puts you where an assistant looks for a specialist. - **Your introducers' websites.** When an accountant or solicitor names you in a joint guide or a seminar write-up, it confirms that other professionals trust you. Co-author a guide with them, such as an accountant and a planner on what to do after a sale. Our [guide for accounting practices](/insights/ai-optimisation-accounting-practices) and [guide for law firms](/insights/ai-optimisation-law-firms) show their side. ## How do you tell whether it is working? Track a fixed list of 20 to 50 client questions, split between business and private clients, including a few about charges and status and a few that name your niche and region. Ask each several times a month in ChatGPT, Gemini, Claude, Copilot and Perplexity, in a clean session, because answers vary from run to run, and count a win only when it repeats. Record whether you are named, who is named instead and whether your status and charges are described correctly. Then ask every new client how they found you, and whether they checked you anywhere before calling. Assistants that search the web live can reflect a corrected listing or a new guide within weeks; answers drawn from a model's memory change only when it is retrained, which takes months. Our dated example is [ID-Pal](/work/id-pal), whose buyers include wealth managers: "UK buyer questions where AI names ID-Pal: none in May, two in August, four in September 2026." ## What to do this month 1. **Make your regulatory details agree.** FCA Register entry, trading names, principal or network, Google Business Profile and website. 2. **Choose your niche.** Pick the one or two client types you serve best, business or private, and agree they lead the website. 3. **Agree the compliance route for content** with your compliance lead or network before anyone writes. 4. **Publish three niche guides.** Three decisions those clients face, each with a named author, a tax year and a date. 5. **Publish a plain charges page** with your status, how fees work and worked examples. 6. **Ask for situation-specific reviews** from recent clients in that niche on VouchedFor, Unbiased or Google. ## Frequently asked questions ### Is a restricted financial adviser worse than an independent one? Not necessarily. MoneyHelper notes that both give tailored advice, and independent advisers may offer a wider choice of products. What matters is whether the adviser's range suits the client's needs and whether they explain it clearly. If you are restricted, say so plainly and explain what it means for the client. ### How do I check a financial adviser is FCA authorised and has no complaints? Search the FCA's Financial Services Register at register.fca.org.uk to check the adviser, or the firm they work for, is authorised to give advice, and that the contact details are genuine. The [Financial Ombudsman Service](https://www.financial-ombudsman.org.uk/decisions-case-studies/ombudsman-decisions) publishes its final decisions since 1 April 2013 in a database you can search by business name. Advisers should make both checks easy by printing their FCA reference number on every page. ### What does ongoing financial advice cost as a percentage in the UK? We do not quote market rates here. Firms charge ongoing fees in different ways, commonly as a percentage of the money they look after or as a fixed fee, and MoneyHelper suggests asking any adviser "How much will the advice cost me?" before going ahead. For an advice firm, the useful step is publishing your own charges with worked examples in pounds as well as percentages, so an assistant has your figures rather than someone else's. ### Can I trust VouchedFor ratings when choosing a financial adviser? They are a useful signal, not a verdict. [VouchedFor says](https://www.vouchedfor.co.uk/) it verifies the identity of all reviewers and never removes or edits negative reviews, and that it checks advisers are authorised. A rating still reflects the clients who chose to leave one, so read what reviewers say about situations like yours and check the Register as well. ### Will publishing guides count as giving advice? Not if they are general and educational. Explaining how an option works, what the trade-offs are and what to ask is information. A page becomes a personal recommendation when it tells someone what to do in their circumstances, and it may be targeted support if it makes suggestions to a defined group. Keep guides general and follow your financial promotions process. ### How often should we update our guides? At least once a year and after every Budget, with the tax year stated on each guide. An out-of-date figure on a money topic undermines trust with clients and assistants alike, so remove any guide you cannot maintain. Clients choose advisers on trust, and they now check that trust with an assistant before they call. A free [Growth Review](/growth-review) is a 30-minute call with a founder that checks what Google and ChatGPT say about your firm and two competitors, with three fixes in writing afterwards. ## Sources - [Forrester, B2B buyers make zero-click buying number one (January 2026)](https://www.forrester.com/blogs/b2b_buyers_make_zero_click_buying_number_one/) - [FCA, PS25/22: Supporting consumers' pensions and investment decisions: rules for targeted support (December 2025)](https://www.fca.org.uk/publications/policy-statements/ps25-22-consumer-pensions-investment-decisions-rules-targeted-support) - [FCA, Financial promotions and adverts (September 2026)](https://www.fca.org.uk/firms/financial-promotions-adverts) - [FCA Handbook, PRIN 2A.4 Consumer Duty: retail customer outcome on price and value (September 2026)](https://www.handbook.fca.org.uk/handbook/PRIN/2A/4.html) - [FCA Handbook, PRIN 2A.5 Consumer Duty: retail customer outcome on consumer understanding (September 2026)](https://www.handbook.fca.org.uk/handbook/PRIN/2A/5.html) - [MoneyHelper, Choosing a financial adviser (September 2026)](https://www.moneyhelper.org.uk/en/getting-help-and-advice/financial-advisers/choosing-a-financial-adviser) - [Financial Ombudsman Service, Ombudsman decisions (September 2026)](https://www.financial-ombudsman.org.uk/decisions-case-studies/ombudsman-decisions) - [VouchedFor, Find a financial adviser (September 2026)](https://www.vouchedfor.co.uk/) --- Title: How is AI changing the way B2B buyers choose suppliers? URL: https://fortitudemedia.ai/insights/ai-search-change-your-industry Author: Ross Williams Published: 2026-04-22 Last reviewed: 2026-09-24 Topic: How buying is changing Short answer: Buyers now start more of their research in AI assistants and arrive with a shortlist already formed. Forrester found in January 2026 that 94% of business buyers use AI in their buying process, and 6sense found buying groups had ranked preferred vendors before first contact in 94% of cases. Assistants now sit inside tools buyers use at work, such as Microsoft 365 Copilot and Google Search. Firms with specific, public, well-evidenced expertise get named; vague generalists are left out. ChatGPT ads do not change the answer itself. A few years ago, a finance director who needed a new audit firm would ask a peer, search Google and read three websites. In 2026 the same director is as likely to type the question into ChatGPT, or into the Copilot box next to their inbox, and read a short answer that names a handful of firms. The change has been fast and uneven, which is what makes it hard to plan for. The numbers are now large. [Forrester's Buyers' Journey Survey](https://www.forrester.com/blogs/b2b_buyers_make_zero_click_buying_number_one/), published in January 2026, found that 94% of business buyers use AI in their buying process, and 61% use private AI tools supplied by their own organisation. The [6sense 2025 Buyer Experience Report](https://6sense.com/newsroom/the-timeline-for-influencing-b2b-buyers-is-shrinking-insights-from-6senses-2025-buyer-experience-report/) (November 2025, more than 4,000 buyers) found that 94% of buyers used large language models while buying, mainly to summarise reviews or analyse data, and that buying groups had ranked their preferred vendors before first contact in 94% of cases. Neither survey is UK-only, but both cover business buyers broadly rather than software buyers alone. In our view, being named by AI assistants has become a business development channel in its own right. This guide sets out how buying has changed in 2026, which of the changes we expected have happened, what ChatGPT ads mean for who gets recommended, and which sectors feel it first. It ends with a dated log that we update each quarter. ## Key takeaways - Most business buyers now use AI while buying: 94% in Forrester's January 2026 survey, and 51% of software buyers in G2's April 2026 survey start their research in an AI chatbot. - Buyers form a shortlist before they speak to anyone, so the assistant's answer often decides who gets the first call. - AI now sits inside tools buyers already use at work, such as Microsoft 365 Copilot, Google Search and Chrome, so asking an assistant is becoming a habit rather than a decision. - Buyers act on what assistants say, but most also check it against reviews, websites and references. - Firms with specific, public, well-evidenced expertise get named; generalists with vague websites are left out. - ChatGPT has shown labelled ads to UK users since 11 August 2026, but OpenAI says ads do not change the answer, so being named in it is still earned. ## How is AI changing the B2B buying process in 2026? It is moving early research, and much of the shortlisting, into AI assistants. Buyers ask a question in their own words, read an answer that names a few firms, and contact fewer suppliers with more of their mind made up. Three changes stand out. First, research starts in an assistant more often: [G2's survey of 1,076 software buyers](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html), run in March 2026 and published in April, found that 51% now begin their software research with an AI chatbot more often than with Google, and 71% use chatbots at some point. Second, the shortlist forms early: in the 6sense research, buyers bought from their first choice 77% of the time. Third, the assistant changes minds: in G2's survey, 69% of buyers chose a different vendor from the one they first planned because of chatbot advice, and one in three bought from a vendor they had not known before. Buyers have not stopped checking. The [TrustRadius 2026 B2B Buying Disconnect Report](https://www.prnewswire.com/news-releases/trustradius-2026-b2b-buying-disconnect-report-reveals-ai-has-changed-how-buyers-research-but-not-what-they-trust-302825792.html) (July 2026, 1,862 technology buyers) found that 63% used AI during their purchase, 94% of those fact-checked its answers at least sometimes, and 74% used reviews. The assistant opens the door; your website, reviews and references still have to hold up when the buyer checks. Which assistants buyers use, and how they differ, is covered in [our comparison of ChatGPT, Gemini, Copilot, Perplexity and Claude](/insights/ai-search-landscape-chatgpt-perplexity-google-claude). ## What percentage of B2B buyers use ChatGPT to research suppliers? Most business buyers use some form of AI while buying, though no published survey isolates ChatGPT use among UK buyers of services. These are the best figures available. | Survey | Who was asked | Finding | Published | |---|---|---|---| | Forrester, Buyers' Journey Survey 2025 | Business buyers | 94% use AI in their buying process; 61% use private AI tools from their organisation | January 2026 | | 6sense, Buyer Experience Report | 4,000+ buyers in North America, EMEA and APAC | 94% used large language models while buying | November 2025 | | G2, The Answer Economy | 1,076 B2B software buyers | 51% start research in an AI chatbot; 71% use one at some point | April 2026 | | TrustRadius, B2B Buying Disconnect Report | 1,862 technology buyers | 63% used AI in their purchase | July 2026 | On ChatGPT itself, Forrester found business buyers twice as likely as consumers to use ChatGPT and four times as likely to use Microsoft Copilot. Copilot is what many buyers have at work: [Microsoft 365 Copilot passed 30 million paid seats](https://www.microsoft.com/en-us/investor/earnings/fy-2026-q4/press-release-webcast) in July 2026. OpenAI said in August 2026 that ChatGPT has [more than a billion weekly users](https://openai.com/index/expanding-access-to-ai-with-chatgpt-ads/). Read these figures with care. G2 and TrustRadius asked software and technology buyers, not buyers of professional services; none of the surveys is UK-only; and some come from vendors with an interest in the answer. ## Five changes we expected in 2026: where does each stand? When this guide was first published in April 2026, we expected five changes. By late September, one has happened, two are happening and two have not shown up in any evidence we can cite. | What we expected | Status in September 2026 | Evidence | |---|---|---| | AI answers built into the tools buyers already use | Happened | Microsoft 365 Copilot passed 30 million paid seats in July 2026; Gemini in Chrome reached UK desktop users on [14 July 2026](https://blog.google/products-and-platforms/products/chrome/were-expanding-gemini-in-chrome-to-users-in-the-uk/); AI Overviews sit at the top of many Google searches | | Sector-specific assistants recommending suppliers | Not yet | We have seen no evidence of sector assistants recommending suppliers; buyers still ask general assistants | | More confident recommendations that buyers act on | Happening | 69% of software buyers changed vendor because of chatbot advice (G2, April 2026), though most still fact-check (TrustRadius, July 2026) | | Procurement teams checking what AI says about suppliers | Not yet shown | No published tender clause or survey we can cite; 61% of business buyers use AI tools supplied by their employer (Forrester) | | A visible gap between named and unnamed firms | Happening | Helm Club: "1 in 5 new members this year came through an AI recommendation." | One change was not on our list and is now live: paid placements. ChatGPT has shown labelled ads to UK users on its Free and Go plans since 11 August 2026, which the next section covers. The Helm Club result shows what the gap looks like from the inside: "A channel that did not exist before now brings one in five new members: founders who ask ChatGPT, Gemini or Claude which peer group to join are told about Helm." The detail is in [the Helm Club case study](/work/helm-club). ## Will ChatGPT ads change which firms get recommended? Not according to OpenAI. ChatGPT has shown labelled ads to UK users on its Free and Go plans since [11 August 2026](https://openai.com/index/testing-ads-in-chatgpt/), and OpenAI says the ads are kept separate from the answer and do not influence it. The Plus, Pro, Business, Enterprise and Education plans show no ads, so a buyer using a company Business or Enterprise account sees none at all. So ads create a paid slot next to the answer, not a paid place in it. A firm can buy attention from someone reading a free ChatGPT answer; the recommendation itself still depends on what the assistant finds about you. Whether ads or being named is the better use of money is the subject of [our comparison of AI visibility and paid search](/insights/economics-ai-visibility-vs-advertising). ## Which UK industries are most affected by AI search? No UK study ranks industries by how much AI has changed buying, and the largest surveys asked software buyers, so technology and SaaS firms have the clearest evidence. In our view, the firms that feel it first are those whose buyers ask specific, researched questions before they call anyone, which covers every sector below. | Sector | What buyers ask, in their words | What assistants look for | Read more | |---|---|---|---| | Law | "Employment lawyer for a TUPE transfer in manufacturing" | Visible depth in named practice areas and sectors | [Law firms](/insights/ai-optimisation-law-firms) | | Accountancy | "Accountant who handles R&D tax relief for software firms" | Named specialisms, with sector clients and worked examples | [Accounting practices](/insights/ai-optimisation-accounting-practices) | | Consultancy | "Consultant for an operations turnaround after a merger" | Published thinking on that exact problem and evidence of solving it | [Management consultancies](/insights/ai-optimisation-management-consultancies) | | Financial advice and insurance | "Wealth manager for a founder planning an exit" | Regulatory standing, named advisers and consistent directory facts | [Financial advisory](/insights/ai-optimisation-financial-advisory), [insurance brokers](/insights/ai-optimisation-insurance-brokers) | | Recruitment | "Recruiter for senior finance roles in the North West" | A clear specialism, with placements and client sectors shown | [Recruitment agencies](/insights/ai-optimisation-recruitment-agencies) | | Technology and SaaS | "Managed IT provider for a 200-person firm with Cyber Essentials Plus" | Certifications, security posture and case studies from similar firms | [Technology companies](/insights/ai-optimisation-technology-companies) | | Fintech and regtech | "KYB checks for a UK fund administrator" | The regulator named, and a direct answer for the buyer's sector | [Fintech](/for/fintech) | | Agencies | "B2B agency that builds demand programmes for industrial firms" | A sector focus and results with numbers | [Marketing and creative agencies](/insights/ai-optimisation-marketing-creative-agencies) | The fintech row has a dated example. For ID-Pal, an identity verification firm: "Gemini now names ID-Pal for UK wealth managers. Claude names it for UK fund administrators, on a page rebuilt in August, and links NorthRow to ID-Pal in the same answer." The detail is in [the ID-Pal case study](/work/id-pal). ## Are procurement teams checking what AI says about suppliers? We have not seen published evidence that tender documents ask how a supplier is described by AI tools, so treat any claim that this is widespread with care. What the evidence does show is that the people who run buying processes use AI at work, often through tools their employer supplies, and that most check what it tells them. The practical conclusion is the same either way. Assume that someone on the buying side will ask an assistant about your firm before a shortlist is agreed, and make sure the answer is accurate. If it is wrong, [our guide to fixing how ChatGPT describes your firm](/insights/ai-conflicting-information-online) sets out the process. Where buyers go further and let an AI agent assemble the shortlist, the issues are covered in [what happens when an AI agent builds the shortlist](/insights/ai-agents-change-b2b-buying). ## Does ranking on Google still protect us? It helps, but it no longer does the whole job, so do not stop investing in it. Google's AI Overviews and AI Mode, ChatGPT search and Copilot all start from a search index, so ranking well makes you more likely to be found. But Google now runs several related searches for one question (query fan-out), so a firm that ranks for its main phrase can still be missing from the answer if it does not cover the follow-up questions or say plainly what it specialises in. How AI Mode differs from AI Overviews is explained in [our guide to Google AI Mode](/insights/end-of-ten-blue-links). Why a first-place ranking can still leave you out of ChatGPT's answer is covered in [we rank first on Google, so why does ChatGPT not mention us](/insights/google-rankings-wont-protect-you). The work that helps both is the same: specific, public, well-evidenced expertise. ## Why does timing matter? Because the firms named now gather the evidence that gets them named again: enquiries, case studies, reviews and coverage. What locks in and what does not, and the smallest sensible start, are set out in [whether you can afford to wait a year](/insights/first-mover-advantage-ai). ## How do we find out where we stand? Before you change anything, find out what the assistants say now. Write a fixed list of 20 to 50 questions your buyers ask, in their words, covering your most valuable services and main regions. Ask each one several times a month in a fresh session in ChatGPT, Gemini, Copilot and Google, because answers vary from run to run, and count a win only when it repeats. Do the same for your two closest competitors. You will usually find one of three things: you are named and described well; you are named but described vaguely or wrongly; or you are absent and a competitor is named instead. Each has a different fix. Our guide to [comparing your AI visibility with competitors](/insights/benchmarking-ai-visibility-competitors) turns the check into a monthly measure. ## What has changed in 2026 so far? A dated log of the changes that matter to B2B buyers. We update it each quarter. - **January 2026:** Forrester reports that 94% of business buyers use AI in their buying process. - **April 2026:** G2 reports that 51% of software buyers now start their research in an AI chatbot. - **July 2026:** Gemini in Chrome launches for UK desktop users (14 July). TrustRadius reports that 63% of technology buyers used AI in their purchase (15 July). Microsoft 365 Copilot passes 30 million paid seats (29 July). - **August 2026:** ChatGPT ads go live for UK users on the Free and Go plans (11 August). OpenAI says ChatGPT has more than a billion weekly users (31 August). ## What to do this month 1. Write down, in one sentence each, the three things your firm is best at and for whom. If you cannot, that is the first job. 2. Rewrite your homepage and service pages so those three sentences are the first thing a reader, or an assistant, sees. 3. Publish one substantial piece on each of those specialisms, with a named author and real experience in it. 4. Fix the facts assistants cross-check, so that Companies House, LinkedIn, Google Business Profile and directories all say the same thing about who you are and what you do. 5. Ask recent clients for reviews on the sites your buyers read, such as Google, Clutch, G2 or Trustpilot. 6. Write your list of buyer questions and record who is named today, for you and for two competitors. ## Frequently asked questions ### Which of our buyers are actually using AI assistants to find suppliers? More than most firms assume. Forrester found business buyers twice as likely as consumers to use ChatGPT and four times as likely to use Copilot. Senior buyers who say they "just Google" are often reading an AI Overview at the top of the results. Ask every new prospect how they found you, and within a quarter you will have your own figure. ### Is outbound still worth doing if buyers ask AI first? Yes. Researched outbound, where a person writes and approves every message, reaches buyers before any assistant is asked, and it works alongside being named. A buyer who has already heard from you, and then sees you named by an assistant, has two reasons to take the call. This is what [Pipeline That Fills](/pipeline-that-fills) does. ### We are a generalist firm. Are we in trouble? You are at a disadvantage, but it can be fixed. Assistants struggle to name firms they cannot describe in a sentence. Choose the two or three areas where you are strongest, make them prominent on your site, publish on them and gather evidence for them. You can keep doing everything else; just stop presenting everything equally. ### How quickly can a firm start being named? Assistants that search the web live, such as ChatGPT search, Perplexity, Copilot and Google's AI features, can reflect a new page or article within weeks. Answers drawn from a model's memory change only when the model is retrained, which takes months. ID-Pal is a dated example: "UK buyer questions where AI names ID-Pal: none in May, two in August, four in September 2026." ### What if AI assistants describe us wrongly? It is common, and usually comes from stale or inconsistent information: an old directory listing, a previous positioning on a partner's LinkedIn profile, or a press release about a service you dropped. Find the sources, correct them and make your own site the clearest, most current statement of who you are. Why assistants sound unsure when sources disagree is explained in [how AI cross-checks what the web says about you](/insights/ai-cross-references-business-sources). The next year will not change what makes a good firm; it changes how buyers find one. If you would like to know how your firm looks to buyers and assistants today, [book a free Growth Review](/growth-review): 30 minutes with David Adams, and three fixes in writing afterwards. The rest of our [guides on how buying is changing](/insights/pillar/ai-first-future) pick up from there. ## Sources - [Forrester, B2B buyers make zero-click buying number one (January 2026)](https://www.forrester.com/blogs/b2b_buyers_make_zero_click_buying_number_one/) - [6sense, 2025 Buyer Experience Report (November 2025)](https://6sense.com/newsroom/the-timeline-for-influencing-b2b-buyers-is-shrinking-insights-from-6senses-2025-buyer-experience-report/) - [G2, The Answer Economy: half of B2B software buyers now start with AI chatbots (April 2026)](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html) - [TrustRadius, 2026 B2B Buying Disconnect Report (July 2026)](https://www.prnewswire.com/news-releases/trustradius-2026-b2b-buying-disconnect-report-reveals-ai-has-changed-how-buyers-research-but-not-what-they-trust-302825792.html) - [Microsoft, FY26 fourth quarter earnings press release (July 2026)](https://www.microsoft.com/en-us/investor/earnings/fy-2026-q4/press-release-webcast) - [OpenAI, Testing ads in ChatGPT (updated August 2026)](https://openai.com/index/testing-ads-in-chatgpt/) - [OpenAI, A milestone in expanding access to AI (August 2026)](https://openai.com/index/expanding-access-to-ai-with-chatgpt-ads/) - [Google, Expanding Gemini in Chrome to users in the UK (July 2026)](https://blog.google/products-and-platforms/products/chrome/were-expanding-gemini-in-chrome-to-users-in-the-uk/) --- Title: How much does a B2B website cost in the UK? URL: https://fortitudemedia.ai/insights/real-cost-of-a-website Author: David Adams Published: 2026-04-22 Last reviewed: 2026-09-24 Topic: Websites that bring work in Short answer: Published UK prices run from about £2,000 for a freelance brochure site to £15,000 to £50,000 for a lead-generating B2B site from a specialist agency, and £60,000 or more for enterprise sites. The build is only part of the cost: hosting, maintenance, content, support and your team's time follow, and one agency suggests 10 to 20% of the build cost a year. Compare quotes on a three-year total covering the same items. Most founders meet the cost of a website as three quotes on a desk. One is suspiciously cheap, one sits in the middle, one makes you wince. They look comparable. They rarely are. One includes copywriting and another assumes you will write it. One includes hosting for a year, one never mentions it. None of them says what the site will cost to run once the agency has sent its final invoice. The published ranges are wide. UK agency guides put a freelance brochure site [from £2,000 and a custom agency build from £10,000](https://www.bigfork.co.uk/insights/how-much-does-a-website-cost) (Bigfork, March 2026), and an enterprise B2B site at [£60,000 to £100,000 or more](https://www.elcap.agency/insights/how-much-should-a-new-b2b-website-cost-in-the-uk) (Elcap, December 2025). Those are build fees only. The number your board actually needs is the cost of owning the site for three years: the build, hosting, security, content, support, your own team's time and the redesign that follows when none of that happens. This guide gives the UK price bands, the running costs, a worked three-year example, and the questions that get you to your own figure. ## Key takeaways - Published UK prices run from about £2,000 for a freelance brochure site to £60,000 or more for an enterprise B2B site, and one agency puts most lead-generating B2B sites at £15,000 to £50,000. - The build quote is only part of the cost; hosting, maintenance, content, support and your team's time make up the rest. - One UK agency suggests budgeting 10 to 20% of the build cost each year for maintenance and improvement. - Compare agencies on a three-year total across the same list of items, not on the build fee alone. - Pay upfront for work with a clear finish and monthly for work whose value fades when nobody touches it, as a website does. - For tax, HMRC treats the cost of creating a site as capital and regular updates as likely revenue; confirm the accounting treatment with your accountant. ## How much does a B2B website cost in the UK in 2026? A B2B website in the UK typically costs from about £2,000 with a freelancer to £15,000 to £50,000 with a specialist agency, and more for large or enterprise sites. The table brings together what UK agencies publish. All are sellers' own figures, so treat them as a guide to the market rather than a quote. | Route | Typical UK build price | Source | What you usually get | |---|---|---|---| | DIY builder such as Wix or Squarespace | From free, plus plan fees | Bigfork, March 2026 | A template you set up and write yourself | | Freelance designer | From £2,000 for a standard brochure site | Bigfork, March 2026 | An adapted theme, your own words, limited support | | Entry-level agency or studio | £5,000 to £15,000 | Elcap, December 2025 | A small site on a proven platform, some planning | | Custom design agency | From £10,000 | Bigfork, March 2026 | Custom design, more planning and templates | | Lead-generating B2B site | £15,000 to £50,000 | [Red Evolution](https://www.redevolution.com/blog/how-much-does-a-good-website-cost-an-in-depth-guide-for-b2b-cmos), June 2026 | Strategy, content, conversion and integrations | | Mid-market B2B | £25,000 to £50,000 | Elcap, December 2025 | Larger sites, more templates and integrations | | Enterprise B2B | £60,000 to £100,000 or more | Elcap, December 2025 | Complex, multi-region or heavily integrated sites | Copywriting is often priced separately: [Axon Garside](https://www.axongarside.com/blog/how-much-does-a-website-cost-in-2026) (May 2025) puts website copywriting at £1,000 to £10,000 or more. Whether you need a template or a custom build is a separate question, covered in [template or custom website: which does a B2B firm need?](/insights/template-vs-custom-website-b2b). For what we charge, see [what it costs](/what-it-costs). ## How much does it cost to run a website each month? For a small UK business, expect hosting at tens of pounds a month and maintenance in the low hundreds, before any content or improvement work. A site that is actively improved costs more, and most of that spend is people's time. | Running cost | Published UK range | Source | |---|---|---| | Hosting | £30 to £60 a month for decent hosting | Red Evolution, June 2026 | | Domain name | Around £10 to £30 a year | Axon Garside, May 2025 | | WordPress maintenance and updates | Roughly £125 to £250 a month, from Elcap's annual range | Elcap, December 2025 | | CMS licence, where the platform charges one | £50 to £351 a month | Red Evolution, June 2026 | | Ongoing management and improvement | About £1,000 a month as a sensible starting point | Red Evolution, June 2026 | | Rule of thumb | 10 to 20% of the build cost a year for maintenance and optimisation | Elcap, December 2025 | ## Why is the build quote not the real cost? Because a quote prices a project, and a website is a running asset. The agency quotes for what it controls: design, development and getting the site live. Everything that happens after launch, and much of what your own team must do before it, sits outside the number. That is rarely dishonesty. It is how the market quotes, and it leads firms to choose on the one line that matters least over the life of the site. The expensive parts are often the unglamorous ones. When we rebuilt [arum](/work/arum)'s website, the scope included "215 articles, research pieces and news items moved across, going back to 2014" and "Every old address redirected, so years of rankings carried over." Moving old content and redirecting old addresses are exactly the lines a cheap quote tends to leave out, and skipping them can cost years of search rankings. ## What goes into the total cost of a website? Eight things. Some are paid to an agency, some to other suppliers, and some in your own team's time. | Cost | What it covers | Often forgotten because | |---|---|---| | The build | Design, development, templates, rounds of changes | It is the only line most buyers compare | | Hosting, domain and security | Server, domain, certificate, backups, protection from attacks | Cheap hosting looks the same until the site goes down | | Maintenance | Software and plugin updates, checking nothing broke | It is invisible until a security flaw is found | | Content | Words, case studies, photography, articles | Agencies assume you will supply it | | Technical care | Speed, AI crawler access, structured data, broken links | Small, ongoing work that nobody owns | | Authority and PR | Mentions, articles and reviews on other sites | Often budgeted separately, but part of why a site is trusted | | Your team's time | Reviewing drafts, managing the project, writing case studies | It never appears on an invoice | | The next redesign | Replacing a site that has drifted out of date | It arrives years later, as a new quote | ## What do quotes commonly leave out? The items at the edges of the project. In our experience, these are the ones to ask about by name: - Writing or rewriting the content, as opposed to placing text you supply. - Photography of your team and offices, as opposed to stock images. - Moving existing pages and articles, and setting up redirects from old addresses. - Connecting the site to your CRM, email marketing or booking tools. - Hosting, security and backups after launch. - Licences for premium plugins, fonts or images, many of which renew every year. - Training for your team on how to edit the site. - Support after launch, and the hourly rate for changes once it ends. - Accessibility checks and speed testing before launch. If a quote does not mention an item, assume it is not included and ask. ## What does a three-year total look like? Example: three routes for the same firm over three years, using round, illustrative figures within the published ranges above. They are not quotes, and they exclude VAT. | | Cheap build, then left alone | Agency build plus support contract | Smaller build, then run and improved monthly | |---|---|---|---| | Build | £3,000 | £25,000 | £15,000 | | Hosting, maintenance and support over 36 months | £3,000 | £9,000 | Included below | | Monthly running and improvement | None | None | £1,250 a month, £45,000 | | Content writing | Your own team | £5,000 at launch | Included in the monthly fee | | Rebuild or refresh in year three | £3,000 | £10,000 | Not needed if improved steadily | | Three-year cash total | £9,000 | £49,000 | £60,000 | | State of the site after three years | Dated, due a rebuild | Sound, but three years old | Improved every month | The cheap route wins on cash and loses on almost everything else, and it hides your team's time. Between the other two, the difference in cash is smaller than the difference in what you are left with, when you pay and who carries the risk if the supplier is wrong. That is why the comparison has to be made over the same period, with the same items. ## Should we pay monthly or upfront for a website? Pay upfront for work with a clear finish and monthly for work whose value fades if everyone stops touching it. The test is simple: if the value holds, buy it as a project; if it fades, pay for it monthly. | Usually best as a project | Usually best monthly | Often best as both | |---|---|---| | A rebrand or visual identity | Content that answers buyers' questions | A website: a build fee, then a monthly fee to host, maintain and improve it | | A one-off piece of research | PR and independent mentions | | | Moving to a new platform, if your team will run it | Running and improving the website | | | An event with a fixed date | Tracking what AI assistants say about you | | Monthly fees keep the provider accountable and spread the cost, but they can drift into paying for activity nobody checks, so insist on a clear monthly scope and read the reports. We sell websites this way ourselves, so weigh our view accordingly. We compare the rebuild cycle with steady improvement in [should we redesign our website or keep improving the one we have?](/insights/website-redesign-vs-continuous-iteration). ## What is included in a website support retainer? A good support retainer covers hosting, security updates, backups, uptime monitoring and a set amount of changes each month, with a stated response time for problems. Better ones add regular improvements: new pages, speed and accessibility checks, and fixes when AI crawlers or search engines have trouble reading the site. Ask for the list in writing, the hours or tasks included each month, what happens to unused time, and the hourly rate for anything outside it. A retainer that only covers hosting is not really a different model from buying a site and hoping. ## Can we capitalise website development costs in our accounts? Possibly the build, but usually not the running costs, and your accountant should confirm. For tax, [HMRC's Business Income Manual](https://www.gov.uk/hmrc-internal-manuals/business-income-manual/bim35815) (updated August 2026) says the costs of creating an asset of enduring benefit are capital, while "the regular update costs of the site are likely to be revenue expenses". It compares a website to a shop window: building the window is capital, changing the display is revenue. How the build appears in your accounts depends on your accounting framework and policies. Monthly service fees are usually a running cost in the period you pay them. Either way, show the board the cost over the full three years. ## How long does a B2B website build take? Usually a few months. [Rudo](https://rudo.co.uk/insights/guides/b2b-website-redesign-timeline/), a UK agency, puts a mid-market B2B project at 10 to 16 weeks from kick-off to launch (April 2026), and says content production is almost always the bottleneck. Have case studies, service descriptions and team details ready before the build starts, because delays cost money as well as time. ## How do you compare website quotes fairly? Put every quote on the same list and compare three-year totals. It takes an afternoon and changes most decisions. ### Step 1: Send every agency the same list Use the eight cost areas and the list of commonly missing items above. Ask each agency to say, line by line, what is included, what is extra and what it costs. ### Step 2: Ask who does the content Find out whether the agency writes the key pages, edits what you supply, or expects finished text. If it is you, put a realistic figure on your team's time. ### Step 3: Ask what happens after launch Ask for the monthly cost of hosting, updates and support, what it covers, and the hourly rate for anything outside it. ### Step 4: Check the contract terms Look for a short notice period, a monthly scope that says what you get, and a written hand-over process if you leave. Make sure your firm owns the domain, content, design files and analytics accounts, registered to a company email address, with agencies given access rather than ownership. Our guide to [choosing a firm to help you get recommended by ChatGPT](/insights/how-to-compare-ai-optimisation-providers) has the full set of contract questions. ### Step 5: Total it over three years Add the build, the running costs, content and a reasonable estimate of your own time over three years. Compare those totals, not the headline build fees. ## Do we need an agency to build it at all? Not always. If you have capable people in-house, a clear plan can be worth more than an agency build. [TCC Group](/work/tcc-group) took that route: "1 week to a plan TCC's own team built the new site from", after "624 pages reviewed, each given a decision: migrate, merge or retire." In the end, "TCC chose to build it with its own team, and the new tcc.group is live, built to the plan, with nothing about it depending on Fortitude." A plan of that kind is what our [Paid Discovery](/discovery) produces. ## How do you know whether the website is worth what it costs? By comparing what it costs to run with the enquiries and work it brings in. Record, from the first month, how each enquiry found you and which became work. Within a year a simple spreadsheet will tell you whether the site pays its way. Our guide to [measuring the return from your website](/insights/measure-website-roi-b2b) sets out the method, and [what it costs to leave a website unchanged for years](/insights/hidden-cost-static-website) covers the other side of the ledger. For the board paper itself, see [how to get your board to approve the budget](/insights/business-case-ai-optimisation-board). ## What to do this month 1. List every cost connected with your current website over the last year, including your team's time, and add it up. 2. Write down who owns your domain, hosting, content and analytics accounts, and fix anything held by a former supplier or employee. 3. If you are getting quotes, send every agency the same line-by-line list and ask for a three-year total. 4. Decide who inside the firm will own the website after launch, and give them time in their week to do it. 5. Ask your accountant how a build would be treated in your accounts before the board sees the numbers. 6. Start recording the source of every enquiry, so that next year you can judge the site on results. ## Frequently asked questions ### Why do quotes for the same website vary so much? Because they rarely describe the same website. One agency adapts a theme and expects you to supply the words; another designs from scratch, writes the key pages and includes a year of hosting and support. Put every quote on the same line-by-line list and most of the difference becomes explainable. ### Should we buy a cheap website now and upgrade later? Only if you are confident you will upgrade, and soon. A cheap build is fine for a new firm testing its offer. For an established firm it often means a generic template, thin content and no support, which leads to a second project within a couple of years. ### How much should website maintenance cost per month for a small UK business? Published UK figures put hosting at £30 to £60 a month (Red Evolution, June 2026) and WordPress maintenance at roughly £125 to £250 a month, working from Elcap's annual range (December 2025). That keeps the site safe and working. Improving it, with new pages, content and fixes, costs more: Red Evolution calls about £1,000 a month a sensible starting point for ongoing management. ### Is a monthly fee just a way for agencies to lock in revenue? It can be, if the contract has a long minimum term and a vague scope. A fair monthly arrangement has short notice, a clear monthly scope and full ownership of what is produced, so the provider keeps the work only by being useful. ### How much of the website budget should go on content? More than most firms plan for. The words, case studies and articles are what persuade buyers and what AI assistants read when deciding whether to name you. When budgets are tight, protect the content and simplify the design. The cheapest website is the one that brings in enough work to pay for itself, and that depends more on how it is run than on what the build cost. To see one proposal that covers both building and running a site, look at [Website That Sells](/website-that-sells); our prices are on [what it costs](/what-it-costs). ## Sources - [Bigfork, How much does a website cost? (March 2026)](https://www.bigfork.co.uk/insights/how-much-does-a-website-cost) - [Elcap, How much should a new B2B website cost in the UK? (December 2025)](https://www.elcap.agency/insights/how-much-should-a-new-b2b-website-cost-in-the-uk) - [Red Evolution, How much does a good website cost? An in-depth guide for B2B CMOs (June 2026)](https://www.redevolution.com/blog/how-much-does-a-good-website-cost-an-in-depth-guide-for-b2b-cmos) - [Axon Garside, How much does a website cost in 2026? (May 2025)](https://www.axongarside.com/blog/how-much-does-a-website-cost-in-2026) - [HMRC, Business Income Manual BIM35815: costs of setting up a web site (August 2026)](https://www.gov.uk/hmrc-internal-manuals/business-income-manual/bim35815) - [Rudo, How long does a B2B website redesign take? (April 2026)](https://rudo.co.uk/insights/guides/b2b-website-redesign-timeline/) --- Title: How often do we need to publish for AI assistants to notice us? URL: https://fortitudemedia.ai/insights/how-often-publish-content-ai Author: David Adams Published: 2026-04-22 Last reviewed: 2026-09-24 Topic: Getting named by AI assistants Short answer: Publish at a rhythm you can keep for eighteen months, and make every piece substantive. For most founder-led B2B firms that is one thorough article a fortnight; monthly works where depth is the priority, and weekly only with several contributors. AI assistants judge the whole body of work and whether it is kept current, not the number of posts. Founder LinkedIn posts count too, but bursts of thin or AI-generated pages and date changes without real updates work against you. It comes up in almost every first conversation we have with a founder. "How often do we need to publish? Weekly? Twice a week? Our competitor seems to put something out every day." Underneath the question is a worry that you are falling behind, and a second worry, usually from whoever would have to write it all, that publishing constantly will produce rubbish. Both worries point at the wrong question. What matters to ChatGPT, Gemini, Claude and Perplexity is not how many articles you publish in a month. It is whether your site and your founders' profiles show a steady, growing body of substantive work on the subject you want to be known for, kept current. Recency does count: [Ahrefs found in July 2025](https://ahrefs.com/blog/do-ai-assistants-prefer-to-cite-fresh-content), across about 17 million citations, that pages cited by AI assistants were on average 25.7% fresher than the pages in Google's ordinary results. But fresh and frequent are different things. This guide sets out what a publishing rhythm does for you with AI assistants, what each frequency needs and achieves, where LinkedIn fits, and how to choose a rhythm your firm can keep for eighteen months. ## Key takeaways - AI assistants judge the whole body of work on a subject and whether it is kept current, not the number of posts in a month. - One substantive article a fortnight suits most founder-led B2B firms; monthly works when depth is the constraint, and weekly needs several contributors. - Thin pieces published to hit a quota lower the standing of the whole site, so never trade depth for frequency. - Publishing many AI-generated pages at once to catch searches can breach Google's scaled content abuse policy, however the pages were made. - Founder posts and LinkedIn articles count too: LinkedIn appeared in about 11% of AI answers in a Semrush study, and most cited authors posted regularly. - Changing the date on an unchanged article is not an update; revise the substance, then change the date. - Choose the rhythm you can keep for eighteen months, name one owner and keep a reserve of finished pieces. ## What does a publishing rhythm do for you with AI assistants? It builds the body of work assistants judge you on, keeps your site visibly current, and gives you enough connected pieces on one subject to be recognised as a specialist. Assistants do not count posts. They read what you have published and form a view of what your firm knows, so what matters is that the body keeps growing with pieces of substance. [Why ChatGPT ignores sites with hundreds of pages](/insights/ai-penalises-thin-content) explains what happens when volume replaces substance. Currency matters as well. [Seer Interactive found in July 2026](https://www.seerinteractive.com/insights/study-content-recencys-impact-on-ai-visibility-in-2026) that 75% of the pages cited by AI assistants in four consumer and finance industries had been updated in the previous twelve months. A regular rhythm, plus planned updates to your best pages, keeps that signal healthy. [How old is too old for ChatGPT](/insights/content-freshness-recency-ai) covers freshness in detail; this guide sticks to frequency, resourcing and who writes. And depth needs a set of pieces. To be the firm assistants name on, say, employee ownership trusts, you need a connected set covering the main question and its follow-ups, not one article. A rhythm is how that set gets built. [Organising content so AI assistants see you as the expert](/insights/topic-clusters-ai-understands) shows what the finished set looks like. ## Why does consistency beat volume? Because assistants judge the whole body of work, and a body of work that is mostly forgettable is not rescued by the occasional strong piece. Given a fixed amount of writing time, spending it on fewer, deeper pieces builds more standing than spreading it across more, thinner ones. Thin pieces are not neutral: they dilute the pages worth quoting and lower the standing of the whole site. So the choice is never "more but weaker". It is how often you can publish something good, and whether you can keep doing it. In our experience, a firm that publishes one good article a month and keeps going for two years ends up with a stronger body of work than one that published thirty pieces in a spring and then stopped. The steady firm has more substance, a site that is visibly maintained and a habit that keeps producing. ## Which publishing rhythm should we choose? The one you can sustain for eighteen months. For most founder-led B2B firms that is one substantive article a fortnight. | Rhythm | Pieces a year | What it needs | Suits | Main risk | |---|---|---|---|---| | Weekly | About 50 | Several contributors or a dedicated writer, plus a steady flow of client insight | Broad subjects, firms with several partners | Running out of things to say and publishing thin pieces | | Fortnightly | 26 | One named owner, a short founder interview per piece, a writer or a founder who writes | Most founder-led B2B firms building a specialism | Slipping in busy seasons | | Monthly | 12 | One well-researched piece from your best thinking | Narrow, deep specialisms; founders with little spare time | Long gaps if one slot is missed | | Quarterly | 4 | Research-grade work others will cite, with other visible activity in between | Firms producing original research | Looking inactive unless PR or LinkedIn fills the gaps | ### Weekly Weekly publishing sends a strong signal of active expertise, and it is expensive. It needs breadth to avoid running dry and either a dedicated writer or several experts sharing the load. For a smaller firm with a narrow specialism, weekly usually produces stretched pieces within a couple of months. ### Fortnightly Fortnightly is the balance we recommend to most founder-led B2B firms. It is frequent enough to keep the site visibly active and the body of work growing, and spaced enough for proper research, an interview with a colleague or client, and a second draft. Over a year it produces twenty-six substantive pieces, enough to build a clear specialism in one subject and start on a second. ### Monthly Monthly is underestimated. One thorough, well-argued piece a month forces depth, because you cannot fill the only slot with a slight idea. Over three years that is thirty-six substantial pieces. It suits firms where the people with the expertise also carry client work. ### Quarterly Four pieces a year can work only if each is exceptional: original research or a substantial guide that other sites cite. If your pieces are good rather than landmark, pair quarterly publishing with other visible activity. [How long PR takes to work, and what a year should look like](/insights/compounding-effect-consistent-pr-12-months) shows how coverage can carry part of the load. ## Did we follow our own advice? Not at first, and it is fair to ask. This site published 71 guides in March 2026 and 21 in April, then no new guides for five months. That is the burst pattern this guide warns against: a large body of work with a single date on it and nothing after. In September 2026 we began revising every guide, adding sources, dates, client results and the questions buyers ask. The plan from here is new guides and scheduled reviews of existing ones at a steady pace, not another batch. Judge us on whether that holds. ## Do LinkedIn posts count towards the body of work? Yes. Assistants cite LinkedIn often, so a founder's posts and articles count towards how the firm is seen, alongside the website. [Semrush's study](https://www.semrush.com/blog/linkedin-ai-visibility-study/) of 325,000 prompts in January and February 2026 found LinkedIn in about 11% of AI answers across ChatGPT search, Google AI Mode and Perplexity (14.3% in ChatGPT search, 13.5% in AI Mode). Frequency mattered there: about 75% of cited authors had posted five or more times in the previous four weeks. Engagement mattered less: most cited posts had 15 to 25 reactions, and 54% to 64% of cited posts shared knowledge or practical advice. ChatGPT search and AI Mode more often cited individual creators (59%), while Perplexity more often cited company pages (59%). So a sensible rhythm for most founders is short, useful LinkedIn posts every week or so, with a LinkedIn article or a website guide every fortnight or month, each saying the same thing about what the firm does. Posts keep the founder visible; the website guides hold the full answer. Helping founders do this in their own voice is part of [Reputation That Wins](/reputation-that-wins). ## Is it bad to publish lots of AI-written posts at once? Yes, if the pages add little. Google's [spam policies](https://developers.google.com/search/docs/essentials/spam-policies), last updated in August 2026, define scaled content abuse as "many pages generated for the primary purpose of manipulating search rankings and not helping users", "no matter how it's created", and give "using generative AI tools or other similar tools to generate many pages without adding value for users" as an example. Google's July 2026 [guide to its AI features](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide) adds that separate pages for every query variation, made mainly to manipulate AI answers, break the same policy. AI tools are fine for structure, tidying and speed. The expertise, the examples and the judgement have to come from your people. [Whether ChatGPT and Google quote content written with AI](/insights/thought-leadership-vs-keyword-stuffing) goes further. ## Does changing the date on an old article count as fresh content? No. Changing the date without changing the substance misleads readers, and Ahrefs, whose research shows assistants favour fresher pages, advises against it. It is tempting because models do lean towards recent dates: a [September 2025 study](https://arxiv.org/abs/2509.11353) found that adding newer publication dates to passages moved individual items by as many as 95 places when language models re-ranked them. That is a bias to respect, not exploit. A real update counts. Revise the figures, add new examples, answer a question buyers now ask, then change the modified date and say what changed. ## What about busy seasons and real life? Perfect consistency is an ideal. Accountancy practices have January, law firms have year end, everyone has a summer. The rhythm is your baseline, not a contract. Missing the occasional fortnight does no lasting harm. What harms you is committing to a pace, building an expectation and then falling silent without explanation. If a quiet period is coming, write two pieces ahead, or move to monthly for the quarter and keep to that. A small bank of two or three finished pieces, held for the weeks when nobody has time to write, removes most of the pressure. ## How do we choose the rhythm we can keep? Three questions settle it: - Do you have enough expertise, and enough angles on your subject, for one excellent piece a week? - Can someone research, draft and refine a piece every fortnight without other essential work suffering? - Can you commit to the chosen rhythm for the next twelve to twenty-four months, through your busy season? If the answer to the ambitious option is no, step down a level. A monthly rhythm you keep builds more standing than a weekly one you abandon after three months. Then name one owner. In our experience the firms that keep a rhythm have one person responsible for the calendar, even if several people write. If the writing is outsourced, [outsourcing content without losing your voice](/insights/outsourcing-content-brand-voice) explains how to keep the expertise yours while someone else drafts. ## What to do this month 1. Count the substantive pieces you published in the last twelve months and note the longest gap. That is your starting point. 2. Choose a rhythm using the three questions above, and err on the slower side. 3. Name one person as owner of the publishing calendar and fill the next six slots, each with a subject and an author. 4. Write two pieces ahead before you announce the rhythm, so the reserve exists from day one. 5. Agree a weekly LinkedIn habit for the founder that points back to the guides. 6. Book a quarterly review to check quality has held and to update your three most important pages. ## Frequently asked questions ### Is monthly publishing too infrequent for AI assistants to notice? No, provided each piece is substantial. Assistants weigh the body of work and whether it is current, not the count. Twelve thorough articles a year, plus updates to your best pages, give an assistant a clear picture of what you know and a site that is visibly maintained. ### How many articles do we need before ChatGPT sees us as an expert? There is no published threshold, and anyone quoting one is guessing. Aim to cover the main question your buyers ask and each follow-up: what it costs, how long it takes, the risks, the alternatives and how it differs by sector. Write those questions down and count them; that is your target set. [What an article that ChatGPT quotes looks like](/insights/anatomy-of-article-cited-by-ai) shows the standard each piece should meet. ### How much does it cost to outsource one B2B article a fortnight? It depends on who does the research, whether a founder is interviewed for each piece, and how much editing and sourcing the writer does. The hidden cost is founder time: a good piece still needs the founder's examples and opinions. Our own prices for this work are on [what it costs](/what-it-costs). ### Should we publish weekly if we can afford to? Only if you can sustain the quality. Weekly works for firms with a broad subject and several contributors. If in doubt, start fortnightly and move to weekly once you have three months of good pieces in hand. ### Do updates to existing articles count as publishing? Yes, and they are often better value than a new piece. Substantially updating an important article with new examples and revised figures keeps it current. Two new pieces and one major update a month is a sound pattern for a firm with a mature site. [Does ChatGPT prefer long articles or short ones?](/insights/long-form-vs-short-form-ai-prefers) covers how long each piece needs to be. ### What if we have published nothing for a year? Start again quietly rather than with a burst. Publish one good piece, then keep the rhythm you have chosen. A site that comes back to life and stays alive does more for you than ten pieces in one week followed by silence. The reliable voice becomes the authoritative voice. Pick the rhythm you can keep and let time do the rest. If you want a view of where you stand before you start, [book a free Growth Review](/growth-review): 30 minutes with David Adams and three fixes in writing. ## Sources - [Ahrefs, Do AI assistants prefer to cite fresh content? (July 2025)](https://ahrefs.com/blog/do-ai-assistants-prefer-to-cite-fresh-content) - [Seer Interactive, Content recency's impact on AI visibility in 2026 (July 2026)](https://www.seerinteractive.com/insights/study-content-recencys-impact-on-ai-visibility-in-2026) - [Semrush, LinkedIn AI visibility study (March 2026)](https://www.semrush.com/blog/linkedin-ai-visibility-study/) - [Google Search Central, Spam policies for Google web search (August 2026)](https://developers.google.com/search/docs/essentials/spam-policies) - [Google Search Central, Optimizing your website for generative AI features on Google Search (July 2026)](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide) - [arXiv, Fang and others, Do large language models favor recent content? (September 2025)](https://arxiv.org/abs/2509.11353) --- Title: What is domain authority, and do AI assistants actually use it? URL: https://fortitudemedia.ai/insights/what-is-domain-authority-ai Author: David Adams Published: 2026-04-22 Last reviewed: 2026-09-24 Topic: Getting named by AI assistants Short answer: Domain authority is a 1 to 100 score from SEO tools such as Moz, Ahrefs and Semrush that estimates how much other websites vouch for yours through links. It is not Google's number, and ChatGPT and other AI assistants never look it up. They respond to the evidence behind it: mentions and links on trusted, relevant sites, including mentions with no link. Earn coverage in publications your buyers read, and judge progress by whether assistants name you, not by the score. Somebody on your team has sent you a screenshot. A tool says your website has a domain authority of 23 and a competitor's has 41. Nobody is sure what the number means, but it feels like a school report and the mark is not good. Meanwhile a prospect mentioned last week that they asked ChatGPT for a shortlist of firms like yours and you were not on it. Those two things are connected, though not in the way most agencies describe. Domain authority is a third-party estimate of how much the rest of the web vouches for your site. ChatGPT, Gemini, Claude and Perplexity do not read that score. They respond to what it measures: how many trusted, relevant sites mention and link to you, and what they say when they do. The evidence behind the score does matter. In an [SE Ranking study of 216,524 pages, reported by Search Engine Journal](https://www.searchenginejournal.com/new-data-top-factors-influencing-chatgpt-citations/561954/) in November 2025, the number of referring domains (separate websites linking to a site) was the strongest predictor of whether ChatGPT cited it: sites with up to 2,500 referring domains averaged 1.6 to 1.8 citations, and sites with more than 350,000 averaged 8.4. This guide explains what the number is, why the tools disagree, what AI assistants actually respond to, and how a founder-led firm can build real authority without buying links. ## Key takeaways - Domain authority is a score from 1 to 100 made by SEO tool companies; Moz's Domain Authority, Ahrefs' Domain Rating and Semrush's Authority Score each measure it differently. - None of the scores is Google's, and no AI assistant looks them up, but the links and mentions behind them do shape who gets cited. - Google has long denied using a website authority score, yet documents leaked in 2024, which Google confirmed were authentic, list an attribute called siteAuthority. - A trade article that names you without a link still helps with AI assistants, because they read the text; links add a second benefit through search rankings. - There is no score you need to reach to appear in Google AI Overviews: Ahrefs found 31% of cited pages ranked outside the top 100. - Check the score quarterly as a rough barometer, and judge progress by whether assistants name you for the questions buyers ask. ## What is domain authority? It is an estimate, made by SEO software companies, of how likely a website is to rank on Google compared with others, built almost entirely from links: how many other sites point to yours, and how trusted those sites are in turn. The idea comes from academic citation. A paper cited by many others is probably worth reading; a paper cited by the leading researchers is worth reading closely. On the web, a link is a citation, and a link from a widely cited site counts for more. Two things matter when you read the number. It is not Google's: Moz, which invented Domain Authority, says on [its own explainer](https://moz.com/learn/seo/domain-authority) (March 2025) that "Domain Authority is not a Google ranking factor". And the scale gets harder as it rises: Moz says "it's easier to grow your score from 20 to 30 than it is to grow it from 70 to 80". So a competitor at 41 against your 23 is ahead, but not in the way the arithmetic suggests. ## Moz, Ahrefs or Semrush: why do the tools give different numbers? Because each company crawls a different slice of the web and scores it with its own formula. Compare scores from one tool only, never across tools. | Score | Made by | Scale | What it is built from | Free way to check | |---|---|---|---|---| | Domain Authority (DA) | Moz | 1 to 100 | Links, especially the number and quality of referring domains, modelled to predict ranking | Moz's free Domain Analysis tool | | Domain Rating (DR) | Ahrefs | 0 to 100 | The strength of a site's backlink profile compared with others in Ahrefs' database; ignores traffic, spam and domain age | Ahrefs' free Website Authority Checker | | Authority Score (AS) | Semrush | 1 to 100 | Link power, estimated organic traffic and six spam checks | Semrush's free account and free backlink tools, with limits | The differences are real. [Ahrefs says](https://help.ahrefs.com/en/articles/1409408-what-is-domain-rating-dr) DR "shows the strength of a website's backlink profile compared to the others in our database". [Semrush says](https://www.semrush.com/kb/747-authority-score) its Authority Score combines "Link Power", "Organic Traffic" and "Spam Factors". A site with plenty of links but little search traffic can score well on DR and poorly on Authority Score. Pick one tool, note the date, and compare yourself with the same five competitors each time. ## Does Google use a domain authority score? Not the third-party ones. Whether Google has its own site-level measure is less clear than its public statements suggested. For years Google's representatives denied using one; John Mueller said "we don't have website authority score", as [Mike King of iPullRank noted](https://ipullrank.com/google-algo-leak) in May 2024. That month, internal Google API documentation leaked, and [Google confirmed to The Verge](https://www.theverge.com/2024/5/29/24167407/google-search-algorithm-documents-leak-confirmation) on 29 May 2024 that the documents were authentic. King's analysis found a feature called "siteAuthority" among Google's stored quality signals, while noting "we do not know specifically how this measure is computed or used". For a founder, the practical reading is simple. Google appears to judge sites as well as pages, and whatever it measures is built from the same kind of evidence the tools approximate. The tools' numbers are still guesses at it. ## Do ChatGPT and the other assistants use the score? No. None of the assistants checks an SEO tool before naming you. They respond to the evidence the score approximates, which is why the score still tells you something. Your firm reaches an AI answer by two routes. The first is training: the model has read a large slice of the web, and firms that appear often on trusted sites, described consistently, are the ones it associates with a category. That changes only when the model is retrained, which takes months. The second is live search: ChatGPT search, Perplexity, Copilot, Gemini and Google's AI features search an index when a buyer asks, then read and cite pages. Search indexes weigh links, so well-linked pages are more likely to be retrieved, and that answer can change within weeks. So does Ahrefs DR matter for ChatGPT? Not the number itself. The SE Ranking study found link-based scores, including its own trust score, correlated with ChatGPT citations, but correlation is not cause: sites with many referring domains tend also to be well known, frequently written about and full of substantial content. [How AI cross-references what the web says about your firm](/insights/ai-cross-references-business-sources) covers how assistants weigh that outside evidence. ## Do brand mentions without a link help with AI search? Yes. An AI assistant reads the text of a page. When a trade article names your firm and describes what you do, that sentence is evidence whether or not it links to you. A mention helps in two ways, and a link adds a third: - Live search: if the trade article is retrieved when a buyer asks a question, the assistant reads your name in a trusted context and can cite that article. Muck Rack's [May 2026 analysis](https://www.globenewswire.com/news-release/2026/05/07/3290268/0/en/generative-pulse-earned-media-consistently-drives-ai-citations-holding-at-84.html) of more than 25 million links cited by ChatGPT, Claude and Gemini found 84% came from earned media. - Training: text naming your firm on trusted sites is part of what future models learn from. - A link, when there is one, also helps your own pages rank and be retrieved, because search indexes still weigh links. So do not turn down coverage because it comes without a link, and do not chase links that come without coverage. In our view, a named, accurate description of your firm on a site your buyers read is worth more than a link from a site nobody reads. ## What domain authority do you need to appear in Google AI Overviews? None. Google publishes no threshold, and the evidence argues against one. [Ahrefs found in March 2026](https://ahrefs.com/blog/ai-overview-citations-top-10) that 38% of the pages cited in AI Overviews ranked in the top ten for the search, 31.2% ranked 11th to 100th, and 31.0% ranked outside the top 100, because Google runs several related searches for each question. A page on a modest site that answers a follow-up question well can be cited. [What Google AI Overviews mean for your firm](/insights/google-ai-overviews-changing-search) covers the rest. ## Do nofollow links count for AI visibility? Partly. Since 2019, Google has treated nofollow, sponsored and ugc link attributes as ["hints about which links to consider or exclude"](https://developers.google.com/search/blog/2019/09/evolving-nofollow-new-ways-to-identify) rather than ignoring them outright. For AI assistants the attribute matters even less, because a nofollow link sits in text that names you, and the assistant reads the text. A nofollow mention in a respected trade title is worth having. ## Which mentions count most? Relevance and trust matter far more than volume: one mention from your sector's leading trade title outweighs a hundred directory listings, and links from sites that exist to sell links can count against you. In our experience the sources that do most for a B2B firm are trade publications with editors, national and regional business press, professional bodies, and partners and clients who describe working with you. [Whether backlinks still matter for ChatGPT](/insights/ai-evaluates-quality-referring-domains) goes into link quality, and [whether ChatGPT trusts press coverage more than your own site](/insights/earned-owned-paid-authority-signals) compares earned, owned and paid sources. ## What does this look like for a real firm? [ID-Pal](/work/id-pal), an identity verification firm, shows that being known is not the same as being recommended. It scored perfectly on recognition: "6 of 6 correct answers to 'what does ID-Pal do?', every engine, every run." Yet "AI could describe ID-Pal. It never recommended it." Its case study sums up the gap: "AI knew who it was, and had no reason to recommend it." Part of the problem was split authority. ID-Pal had bought NorthRow, a UK business verification platform, and "nothing on either site told Google or the AI engines that NorthRow and ID-Pal were the same company." The fix began with "About four hours of structured data work declaring NorthRow part of ID-Pal on both domains." Five industry pages were then rebuilt to answer the questions compliance leads ask. The results so far: "UK buyer questions where AI names ID-Pal: none in May, two in August, four in September 2026." And "AI answers still recommending NorthRow as a rival: six in May, one in September 2026." ## How do we build authority without buying it? Four honest routes, the same whether your goal is Google, ChatGPT or both. ### Earn coverage in publications your buyers trust Journalists and editors need expert comment, data and opinion. A founder who can explain what is changing in their sector, with a view, gets quoted, and each quote puts your name on a trusted site. Coverage on a respected title is read by live-search assistants as soon as it is indexed, often well before your own score moves. [Which kinds of PR win work for a B2B firm](/insights/pr-for-b2b-what-works) shows how to do this without a large budget. ### Publish something worth citing Original research, a benchmark, an annual survey of your clients, or a clear explainer of a regulation nobody has explained well will attract references without asking. [Whether publishing your own research helps ChatGPT recommend you](/insights/original-research-data-ai-trust) covers how. ### Turn relationships into references Your professional body, key suppliers, software partners and proud clients will usually mention you if asked and given something easy to publish: a member profile, a joint case study or a partner listing. ### Refuse anything that looks manipulative Bought links, private link networks, mass directory submissions and article-spinning services may nudge a tool's score for a while. They do nothing for whether an assistant trusts you, and if Google spots the pattern your organic visibility suffers too. When an agency promises a fixed number of links a month, ask where they come from. ## How should we use the score without being ruled by it? Check it quarterly, in one tool, against the four or five firms buyers shortlist alongside you, and watch the direction of travel rather than the number. Pair it with the measure that matters: whether ChatGPT, Gemini, Claude, Copilot, Perplexity and Google's AI answers name you for the questions your buyers ask. If the score is rising but the assistants still name competitors, the problem is usually how your site describes what you do or inconsistent facts across the web; [why ChatGPT describes your firm wrongly](/insights/ai-conflicting-information-online) covers that case. If the assistants have started naming you and the score has barely moved, ignore the score. ## What to do this month 1. List every trusted site that could plausibly mention you: trade titles, your professional body, partners, clients. Mark which already do, with or without a link. 2. Pick one piece of coverage to earn in the next sixty days, with a named contact and a specific angle. 3. Find and stop any link building you could not explain to a journalist without embarrassment. 4. Check your score and those of your five closest competitors in one tool, note the date, and set a reminder for three months. 5. Write down 20 to 50 questions your buyers ask and record which firms the main assistants name for them, so you judge progress by that. ## Frequently asked questions ### How do I check my domain authority for free? Use one tool's free checker and stay with it: Moz's Domain Analysis for Domain Authority, Ahrefs' Website Authority Checker for Domain Rating, or a free Semrush account for Authority Score. Check your five closest competitors at the same time. The scores differ between tools, so never compare a Moz number with an Ahrefs one. ### What is a good domain authority score for a small B2B firm? It depends on your sector and competitors. A specialist consultancy with a modest score may be the strongest firm in its niche, while a national firm with a higher one may lag its peers. Compare against the firms buyers shortlist alongside you, and treat the direction over a year as more meaningful than the number on any day. ### Can I improve my domain authority quickly? Not by legitimate means. The score is built from other sites' decisions to link to you, so it moves at the pace those decisions are made, usually over quarters. What can change faster is what live-search assistants say about you, because a mention on a trusted page can be read as soon as it is indexed. ### Do paid directory listings help with AI recommendations? A few relevant, well-kept directories help with consistency: your professional body's register, a respected sector directory, your Google Business Profile. Beyond that, paid directories add little, because a listing anyone can buy is not evidence that people who know your field chose to mention you. ### Should we hire a link-building agency? Be cautious. Ask where its links come from, whether you could name the publications to a client, and whether the sites have real readers. An agency that earns editorial coverage and places expert articles is doing PR, and that moves both the score and the assistants. ### Does domain authority matter more than the content on our own site? They work together. Outside evidence gets you considered; clear, specific pages let an assistant describe accurately what you do and for whom. A firm with high authority and a vague website gets named for the wrong things, and a firm with a clear site and no outside mentions rarely gets named at all. Authority is earned in the places your buyers read, which is the work of [Reputation That Wins](/reputation-that-wins). To see where you stand before worrying about any score, [book a free Growth Review](/growth-review): 30 minutes with David Adams and three fixes in writing. ## Sources - [Search Engine Journal, New data reveals the top 20 factors influencing ChatGPT citations (November 2025)](https://www.searchenginejournal.com/new-data-top-factors-influencing-chatgpt-citations/561954/) - [Moz, Domain Authority: what is it and how is it calculated (March 2025)](https://moz.com/learn/seo/domain-authority) - [Ahrefs Help Center, What is Domain Rating (DR)? (September 2026)](https://help.ahrefs.com/en/articles/1409408-what-is-domain-rating-dr) - [Semrush Knowledge Base, Authority Score (accessed September 2026)](https://www.semrush.com/kb/747-authority-score) - [iPullRank, Secrets from the algorithm: Google Search's internal engineering documentation has leaked (May 2024)](https://ipullrank.com/google-algo-leak) - [The Verge, Google confirms the leaked Search documents are real (May 2024)](https://www.theverge.com/2024/5/29/24167407/google-search-algorithm-documents-leak-confirmation) - [Ahrefs, AI Overview citations from top-ten pages (March 2026)](https://ahrefs.com/blog/ai-overview-citations-top-10) - [Muck Rack via GlobeNewswire, Earned media consistently drives AI citations (May 2026)](https://www.globenewswire.com/news-release/2026/05/07/3290268/0/en/generative-pulse-earned-media-consistently-drives-ai-citations-holding-at-84.html) --- Title: What is Google AI Mode, and how is it different from AI Overviews? URL: https://fortitudemedia.ai/insights/end-of-ten-blue-links Author: Ashley Kenerson Published: 2026-04-22 Last reviewed: 2026-09-24 Topic: How buying is changing Short answer: AI Mode is Google's conversational search, a separate tab launched in the UK in July 2025, where a buyer asks a long question and follows up. AI Overviews are the AI summaries shown automatically at the top of some normal results. Both break a question into many related searches, so pages outside the top ten are often cited. Get cited by being indexed, answering the follow-up questions buyers ask, writing what only you can write and earning mentions from credible sites. If you have run a business for more than a decade, you have a picture of how people find you online: someone types a few words into Google, gets ten links and clicks one of the top three. Every pound you have spent on your website, on search agencies and on adverts has been spent against that picture. The picture is out of date. For a growing share of searches, the first thing a buyer sees on Google is a written summary, an AI Overview, with a few links beside it. Next to the familiar results sits a second option, AI Mode, where the buyer asks a longer question and gets a written answer they can question further. Both are now very large: Alphabet told investors in [June 2026](https://blog.google/alphabet/investor-presentation-june-2026/) that AI Overviews have more than 2.5 billion monthly users and AI Mode more than 1 billion. And when a summary appears, people click less. [Pew Research Center](https://www.pewresearch.org/short-reads/2025/07/22/google-users-are-less-likely-to-click-on-links-when-an-ai-summary-appears-in-the-results/) found that US users clicked a traditional result on 8% of visits when an AI summary appeared, against 15% when it did not. This guide explains what AI Mode is, how it differs from AI Overviews, and how your firm gets cited in both. ## Key takeaways - AI Overviews are the AI summaries at the top of normal Google results; AI Mode is a separate tab where a buyer holds a conversation with Google Search. - AI Overviews reached the UK in August 2024 and AI Mode in July 2025, and since January 2026 a follow-up question in an AI Overview leads straight into AI Mode. - Both break a question into many related searches (query fan-out), which is why only 38% of AI Overview citations came from top-ten pages in Ahrefs' March 2026 study, down from 76% in July 2025. - Ranking still helps, because both sit on Google's index and core ranking systems, but covering the follow-up questions matters as much as ranking for the main one. - Clicks fall when an AI summary appears, but pages cited inside an AI Overview get far more clicks than pages left out. - Search Console counts AI Overviews and AI Mode in its main Performance report, and since June 2026 it also has dedicated generative AI reports. ## What replaced the ten blue links? A written answer, sometimes above the links and sometimes in place of them. The old model gave the buyer a ranked list and left the choosing to them; the new one reads many sources and writes a response that names a few options with reasons. | | The ten blue links | AI answers (AI Overviews, AI Mode, ChatGPT and others) | |---|---|---| | What the buyer types | A few keywords | A full question, often with context and follow-ups | | What comes back | A ranked list of pages | A written answer with a handful of cited links | | Who does the choosing | The buyer, by clicking | Largely the assistant, which summarises and names options | | What a lower position means | Fewer clicks, but still seen | Often no mention at all | | What the system rewards | Relevance and links, page by page | Relevance and links still, plus clear answers to the related questions and evidence that others agree | The old model rewarded pages that matched the words people typed, links from other sites and technical competence, which favoured big firms. Google's AI features still sit on its index, where links and ranking count, but each answer is assembled from many searches. That rewards what a sceptical buyer wants: a clear answer from someone who plainly knows the subject, backed up by other sources. It is good news for a specialist firm with deep, public expertise. How AI Overviews are changing search more broadly is covered in [what AI Overviews mean for your firm](/insights/google-ai-overviews-changing-search). ## What is Google AI Mode, and should my business care? AI Mode is Google's conversational search: a buyer asks a long or complex question, Google writes an answer with links, and the buyer can ask follow-up questions in the same thread. It launched in the UK on [28 July 2025](https://blog.google/company-news/inside-google/around-the-globe/google-europe/united-kingdom/ai-mode-search-uk/) as a tab on the search results page and in the Google app. Google says early users' questions were two to three times the length of traditional searches. Behind the scenes it uses query fan-out: it breaks the question into subtopics, runs many searches at once and writes one answer. Google says AI Mode is rooted in its core quality and ranking systems. You should care if your buyers ask the kind of question AI Mode is built for, such as "Which employment law firms handle TUPE transfers for manufacturers in the Midlands, and what should we ask them?" What matters is whether your pages are among those it reads and links. ## How is AI Mode different from AI Overviews? AI Overviews summarise; AI Mode converses. An AI Overview appears at the top of ordinary results when Google judges a summary helpful, while AI Mode is a separate place a buyer chooses to go for a longer back-and-forth. | | Classic results | AI Overviews | AI Mode | |---|---|---|---| | Where it appears | The main results page | At the top of the main results page, for some searches | A separate tab on the results page and in the Google app | | Is it the default? | Yes | Shown automatically when Google judges it helpful | No, the buyer chooses it or arrives through a follow-up question | | How the answer is built | Ranked pages for the words typed | A short summary, using query fan-out | A longer answer with follow-ups, using query fan-out | | How many sources it links | About ten results a page | A handful beside the summary; Google sets no fixed number | Links throughout the answer; Google sets no fixed number | | Model | None | Gemini 3 by default since January 2026 | A custom Gemini 2.5 model at UK launch, with newer Gemini models added since | | How it shows in Search Console | Performance report, Web search type | Counted in the Performance report (Web), and in the generative AI reports since June 2026 | The same as AI Overviews | Since January 2026 the two have been joined up. Google's [January 2026 update](https://blog.google/innovation-and-ai/products/google-ai-updates-january-2026/) made Gemini 3 the default model for AI Overviews globally and let people ask a follow-up question from an AI Overview, which takes them into AI Mode without clicking the tab. ## When did each arrive in the UK? AI Overviews came first, in August 2024, and AI Mode followed in July 2025. | Date | What happened | |---|---| | 15 August 2024 | AI Overviews launch in the UK | | 28 July 2025 | AI Mode launches in the UK as a tab, using query fan-out | | January 2026 | Gemini 3 becomes the default model for AI Overviews globally, and follow-up questions lead from an AI Overview into AI Mode | | May 2026 | At Google I/O, Google announces what it calls the next era for Search, including information agents that monitor topics in the background and a new Search box it calls its biggest upgrade in over 25 years | | 3 June 2026 | Search Console starts adding generative AI performance reports | | 31 August 2026 | The generative AI reports reach all sites | Google's [recap of its May 2026 announcements](https://blog.google/innovation-and-ai/technology/ai/google-ai-updates-may-2026/) gives no UK dates for the I/O features, so check availability before planning around them. ## Is Google AI Mode the default search in the UK? No. As of September 2026, AI Mode in the UK is a tab a buyer chooses; the default is still the familiar results page, with an AI Overview at the top of some searches. The line is thinner than it was, because a follow-up question in an AI Overview now leads into AI Mode. The searches most likely to show an AI summary are the ones B2B buyers make when they are researching. In Pew's March 2025 US data, 60% of searches that started with a question word produced an AI summary, against 18% of searches overall. ## Do AI Overviews only cite pages that rank on page one? No. Ahrefs found in [March 2026](https://ahrefs.com/blog/ai-overview-citations-top-10) that 38% of AI Overview citations came from pages in Google's top ten, down from 76% in July 2025; 31.2% came from positions 11 to 100 and 31.0% from beyond the top 100. The reason is query fan-out: Google runs several related searches for one question and cites a strong page for each part, so a page can be cited for a sub-question without ranking for the main phrase. For example, a buyer asks AI Mode how to move payroll to an outsourced provider before the year end. Google might also search for costs, timelines and HMRC requirements, and cite a different page for each. A firm with a clear page on switching timelines can be cited for that part while a larger rival ranks first for "outsourced payroll". Ranking still helps, because both features sit on Google's index and core ranking systems; covering the follow-up questions now matters as well. Why a first-place ranking does not guarantee a mention in ChatGPT is answered in [we rank first on Google, so why does ChatGPT not mention us](/insights/google-rankings-wont-protect-you). ## What do AI Overviews and AI Mode do to clicks? They reduce clicks overall and concentrate them on cited pages. In Pew's data, only 1% of visits with an AI summary led to a click on a link inside it. But [Seer Interactive's April 2026 study](https://www.seerinteractive.com/insights/aio-impact-on-google-ctr-2026-update) of 53 brands found that being cited inside an AI Overview brought about 120% more organic clicks than not being cited, though still fewer than when no AI Overview appeared. For a B2B firm the usual result is fewer visits with steady enquiries. We explain how to read that pattern in [why traffic falls when rankings have not](/insights/zero-click-search-your-business). ## Does AI Mode traffic show in Google Search Console? Yes, but mixed in with everything else. Google says traffic from AI Overviews and AI Mode is [included in the overall figures](https://developers.google.com/search/docs/appearance/ai-features) in the Performance report, under the Web search type, with no separate click count for each feature. Since June 2026, Search Console has also had [generative AI performance reports](https://developers.google.com/search/blog/2026/06/gen-ai-performance-reports), which show impressions for AI Overviews, AI Mode and Discover's AI features by page, country and device; they reached all sites by 31 August 2026. In Google Analytics 4, visits from AI Overviews and AI Mode arrive as Organic Search, not in the AI Assistant channel added in May 2026 for ChatGPT, Gemini, Copilot and others ([Nice Looking Data](https://www.nicelookingdata.com/blog/ga4-ai-traffic-chatgpt-referrals)). So judge AI Mode by its impressions and by enquiries, not clicks alone; [how to tell which of your leads came from AI](/insights/attribution-ai-era-track-leads) explains the method. ## How do I get my website cited in Google AI Mode? The same way you get cited in AI Overviews: be indexed and eligible for a snippet, answer the questions buyers ask and the ones they ask next, and give Google evidence that others rate you. Google says there are no additional requirements and no special markup needed to appear in either. ### Step 1: Check that you can be shown Your key pages must be indexed and eligible for a snippet. Check that no nosnippet or max-snippet setting is hiding them, and that nobody has switched on the [Search Console control](https://support.google.com/webmasters/answer/16908024?hl=en) that excludes a site from AI Overviews and AI Mode. ### Step 2: Answer the whole question List the follow-up questions a buyer asks about each service, such as costs, timelines, risks and alternatives, and answer each clearly on the relevant page, with the direct answer at the top. Helm Club's comparison pages show what this can do: "Every comparison page we have published ranks first for its question on Google and the AI engines." The detail is in [the Helm Club case study](/work/helm-club). ### Step 3: Do not build a page for every variation Google's [AI optimisation guide](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide), last updated in July 2026, warns that creating separate pages for every query variation or fan-out query, mainly to manipulate AI answers, breaks its scaled content abuse policy. One strong page that covers a topic and its follow-ups does more than twenty thin ones. Our guide to [organising content so AI sees you as the expert](/insights/topic-clusters-ai-understands) shows how. ### Step 4: Write what only you can write The same guide says content that offers first-hand expertise will influence presence in AI search more than anything else, and contrasts it with commodity content anyone could produce. Use examples from your own work, numbers you can stand behind and named authors. ### Step 5: Earn mentions and links from credible sources Both features sit on Google's index, where links from relevant, credible sites still count; trade press, professional bodies and client case studies are the durable route. More in [whether backlinks still matter for AI search](/insights/ai-evaluates-quality-referring-domains). ### Step 6: Keep your facts consistent Your name, services, sectors and locations should read the same on your website, Google Business Profile, LinkedIn, Companies House and directories. ## What carries over from our existing search work? Most of the work that mattered. What carries over badly is work that only ever made sense to an algorithm. | Carries over well | Carries over badly | |---|---| | A fast, well-structured site that is easy to read | Pages written to match keyword phrases rather than answer questions | | Clear pages about what you do, for whom and where | Links bought or swapped in bulk from irrelevant sites | | Useful content written by people who know the subject | Thin location pages and near-duplicate content | | Mentions and links from credible trade and industry sources | A page for every variation of a search phrase, or anything written to fool a ranking system | If a piece of work would impress a sceptical expert, it helps in both classic results and AI answers. How the two fit together is covered in [whether you still need SEO now buyers ask ChatGPT](/insights/seo-vs-llm-optimisation). ## What to do this month 1. Open Search Console's generative AI report and note which pages earn AI Overview and AI Mode impressions. 2. For your main service, list the follow-up questions a buyer asks and check each has a clear answer on your site. 3. Run your five most important buyer questions in AI Mode and note which sites it links for each part of the answer. 4. Merge or remove pages written for keyword variations rather than for readers. 5. Check your key pages are indexed and eligible for snippets, and that Search Console's AI features control is set the way you intend. 6. Pitch one expert article to a trade publication your buyers read. ## Frequently asked questions ### Does Google still matter for winning new business? Yes. Many buyers still search, and AI Overviews are read by people who think they are just Googling. Ranking in the list is no longer the whole objective: being cited in the answer is now the goal rankings were always a proxy for. ### Why does a smaller firm sometimes get cited ahead of a bigger one? Because query fan-out cites a strong page for each part of a question. A small firm that has published clearly on a narrow subject, with named authors, real cases and credible mentions, gives Google something specific to cite; a broad brochure site may not. ### Can we opt out of AI Overviews and AI Mode? Yes. Search Console has a control that excludes a site from AI Overviews, AI Mode and Discover's AI features without affecting classic ranking. For most B2B firms that would mean giving up the chance to be cited where buyers are reading, so we would not recommend it without a specific reason. ### Do we need special schema markup to appear in AI Mode? No. Google says its AI features need no special structured data. Markup still helps search engines read your facts correctly, so keep it accurate, but do not treat it as a ranking trick; Google stopped showing FAQ rich results on 7 May 2026. ### How long before AI Mode reflects changes to our site? Because both draw on Google's index, a new or improved page can be reflected within weeks of being recrawled. Answers drawn from a model's memory change only when the model is retrained, which takes months. The ten blue links were never the point: they were how buyers found someone to trust, and much of that job has moved into a written answer. If you want to see whether that answer names you today, [book a free Growth Review](/growth-review): 30 minutes with David Adams, and three fixes in writing afterwards. ## Sources - [Alphabet, Investor presentation (June 2026)](https://blog.google/alphabet/investor-presentation-june-2026/) - [Google, Introducing AI Mode in the UK (July 2025)](https://blog.google/company-news/inside-google/around-the-globe/google-europe/united-kingdom/ai-mode-search-uk/) - [Google, The latest AI news we announced in January (February 2026)](https://blog.google/innovation-and-ai/products/google-ai-updates-january-2026/) - [Google, The latest AI news we announced in May 2026 (June 2026)](https://blog.google/innovation-and-ai/technology/ai/google-ai-updates-may-2026/) - [Ahrefs, Only 38% of AI Overview citations come from the top ten (March 2026)](https://ahrefs.com/blog/ai-overview-citations-top-10) - [Pew Research Center, Google users are less likely to click on links when an AI summary appears (July 2025)](https://www.pewresearch.org/short-reads/2025/07/22/google-users-are-less-likely-to-click-on-links-when-an-ai-summary-appears-in-the-results/) - [Seer Interactive, AIO impact on Google CTR: 2026 update (April 2026)](https://www.seerinteractive.com/insights/aio-impact-on-google-ctr-2026-update) - [Google Search Central, AI features and your website (December 2025)](https://developers.google.com/search/docs/appearance/ai-features) --- Title: What makes ChatGPT treat our content as expert rather than marketing? URL: https://fortitudemedia.ai/insights/what-makes-expert-quality-content Author: Ross Williams Published: 2026-04-22 Last reviewed: 2026-09-24 Topic: Content that gets quoted Short answer: Five signals: an observation only experience could produce, specific figures and cases with named sources, an honest account of when the advice does not apply, visible reasoning, and a tone that informs a peer. In a 2024 Princeton and IIT Delhi study, adding quotations lifted a page's visibility in AI answers by about 41% and statistics by about 34%, against about 10% for an authoritative tone. A named author helps mainly through Google and through the same name appearing on LinkedIn and in the trade press. Your website reads well. The agency that wrote it was good. The case studies are polished, the service pages are persuasive, and the blog has a hundred posts with sensible titles. Yet when a buyer asks ChatGPT, Gemini or Claude who understands their problem, the assistant names a smaller competitor whose website looks like it was built in a weekend. Good writing and expert content are different things. A page can be beautifully written and still be marketing copy: text designed to move a reader towards a decision rather than help them understand something. In a [Princeton and IIT Delhi study presented at KDD 2024](https://arxiv.org/abs/2311.09735), adding quotations from relevant people raised a source's visibility in AI-generated answers by about 41%, adding statistics by about 34% and citing sources by about 30%. Rewriting the same text in a more authoritative tone raised it by only about 10%. It was tested on a research engine, so treat it as directional, but substance moved the result far more than style. The competitor with the weekend website has, somewhere on it, a partner writing frankly about what goes wrong, what it costs and when their approach does not apply. This guide sets out the signals, a phrasing table your writers can use, how to apply it to service pages, and a test you can run this afternoon. ## Key takeaways - AI assistants and buyers both favour content written to help a reader understand over content written to persuade them. - Expert content says something the writer could only know from doing the work: a specific observation, a number, a case or a failure. - In the best-known test, adding quotations, statistics and cited sources lifted visibility in AI answers far more than an authoritative tone did. - Admitting limits and trade-offs makes a page more credible, because that is how people who have done the work talk. - A named author helps mainly through Google's quality systems and through the same name appearing on LinkedIn and in the trade press; no published study shows ChatGPT weighs bylines directly. - Service pages need the same treatment as articles, because in our experience they are where "who should we hire" questions are usually answered. ## What is the difference between marketing copy and expert content? Marketing copy asks "how do I get this person to believe X". Expert content asks "what does this person need to know to judge X for themselves". Everything else follows from that. Most business content is marketing, even when it does not feel like it. It is written from the seller's side of the desk: benefits, nothing awkward, a call to action. "Our proven approach delivers measurable results" tells a reader, or an assistant, nothing except that a firm would like to be hired. Expert content is written from the reader's side: how something works, where it goes wrong and what it depends on. "For firms under about £5m, a full audit is usually unnecessary and a review will satisfy most lenders; above that the picture changes, and here is why" is expert content. It helps, and only someone who knows could have written it. ## What signals tell an assistant a page was written by an expert? Five things, in rough order of importance: something original, something specific, an honest account of limits, visible reasoning, and a tone that treats the reader as a peer. ### It says something new An expert notices what a summariser would not: that clients always ask the same question in month three, or that the standard advice suits large firms and fails small ones. Google's [guide to its generative AI features](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide) (updated July 2026) makes the same distinction, contrasting "commodity content" that "could originate from anyone" with first-hand experience. ### It is specific Expert content has numbers, cases and named sources: "a food manufacturer with three sites and about 200 staff", not "a client of ours". Where a figure has a source, name it. Where it is an observation, say so. Where you have none, do not invent one. ### It admits limits Compare "our approach delivers a 40 per cent improvement for every client" with "the improvement varies a great deal and depends mostly on how organised the firm was to start with". The second is more useful and more credible, because it reads like someone who has watched it happen. ### It shows its reasoning The page defines the problem, says why the answer works, names the alternatives and why they are less good, and says how it can fail. ### It treats the reader as a peer No urgency and no fear: "this works for X but not for Y" instead of "this is the answer". Our guide to [whether ChatGPT and Google quote AI-written or keyword-led content](/insights/thought-leadership-vs-keyword-stuffing) shows the opposite patterns side by side. ## What do AI assistants actually pick up on? Substance: quotations, figures and cited sources, far more than tone. The Princeton and IIT Delhi test measured how much each change to a page moved its visibility in AI-generated answers: | Change made to the page | Change in visibility (main measure) | |---|---| | Adding quotations from relevant people | About 41% higher | | Adding statistics | About 34% higher | | Citing credible sources | About 30% higher | | Using an authoritative tone | About 10% higher | | Keyword stuffing | About 8% lower | A 2025 study of ChatGPT citations points the same way, though it shows correlation, not cause. [SE Ranking found in November 2025](https://www.searchenginejournal.com/new-data-top-factors-influencing-chatgpt-citations/561954/) that pages with expert quotes averaged 4.1 citations against 2.4 without, and pages with 19 or more data points averaged 5.4 against 2.8 for pages with little data. Search-connected assistants such as ChatGPT search, Perplexity and Google's AI features retrieve a handful of pages, compare them and lift the passages that answer best. They do not verify each figure. Buyers do: in the [TrustRadius 2026 B2B Buying Disconnect Report](https://www.prnewswire.com/news-releases/trustradius-2026-b2b-buying-disconnect-report-reveals-ai-has-changed-how-buyers-research-but-not-what-they-trust-302825792.html) (July 2026), 94% of technology buyers who used AI said they fact-checked its answers at least sometimes. An unsupported figure may get past an assistant. It does not get past a buyer who clicks through. ## Does the author's name matter to ChatGPT? It matters to Google and to readers; for ChatGPT, no published study shows it checks bylines directly. The benefit is indirect, and it comes through two routes. The first is Google. Its guidance on [creating helpful content](https://developers.google.com/search/docs/fundamentals/creating-helpful-content) asks "Do pages carry a byline, where one might be expected?" and notes that "E-E-A-T itself isn't a specific ranking factor". Because Google's AI features, and assistants that draw on Bing, start from a search index, what helps a page rank helps it get retrieved, though ranking does not guarantee a mention. The second is corroboration: a partner whose name and subject appear consistently on your site, on LinkedIn and in the trade press is easier to connect to that subject. LinkedIn is itself a source: a [Semrush study](https://www.semrush.com/blog/linkedin-ai-visibility-study/) (March 2026) found that in ChatGPT search and Google AI Mode, 59% of LinkedIn citations came from individual members rather than company pages. [Why AI assistants care about E-E-A-T](/insights/eeat-why-ai-cares-more) explains the idea in plain terms. ## Why is our well-written content not getting us named? Usually because it was written to persuade, by someone who did not do the work, and nobody with expertise put their name to it. For example, a property consultancy's page on rent reviews explains what a rent review is and that the firm can help. It has no author, no example, and nothing on what goes wrong or when a tenant should not bother contesting. To an assistant it is one of a hundred similar pages. ID-Pal, an identity verification firm that had bought NorthRow and folded it into id-pal.com, had a version of this problem. Its case study puts it plainly: "AI could describe ID-Pal. It never recommended it." It records: "Across 36 recorded AI answers to the questions ID-Pal's buyers actually ask, ID-Pal was named three times, and only for its brand or its Salesforce integration. AI knew who it was, and had no reason to recommend it." Two things changed. One was technical: "About four hours of structured data work declaring NorthRow part of ID-Pal on both domains." The other was content: "Five industry pages rebuilt: fund administration, wealth management, legal, accountancy and estate agency. Each opens with the question a compliance lead types, names the regulator, and answers inside the first ninety words." The result the [ID-Pal case study](/work/id-pal) reports: "UK buyer questions where AI names ID-Pal: none in May, two in August, four in September 2026." The fix is rarely to write more. It is to put the expertise the firm already has onto the page. If outside writers produce your content, [how to outsource content without losing your voice](/insights/outsourcing-content-brand-voice) covers how to get the expertise in. ## Which phrases read as marketing, and what do experts write instead? The pairs are illustrative. For each phrase on the left, ask what you would say to a client across the table, and write that. | Marketing phrasing | Expert phrasing | |---|---| | "Our proven approach delivers results." | "For firms under £5m, a review usually satisfies lenders; above that, most ask for a full audit." | | "Industry-leading expertise." | "We work with manufacturers, retailers and law firms, and the hard part is different in each." | | "Tailored solutions for every client." | "We change two things for each client: the reporting schedule and who signs off." | | "Fast, hassle-free onboarding." | "Onboarding takes about four weeks. The second week is the slow one, because it waits on your data." | | "We guarantee results." | "We cannot promise the outcome. We can promise the work, the reporting and a named person to call." | | "Many clients see significant savings." | "The saving depends mostly on how many suppliers you start with; firms with one supplier often save little." | | "Research shows AI is changing how buyers shop." | "In G2's April 2026 survey, 51% of software buyers said they now start research in an AI chatbot." | | "You need to act now." | "If your renewal is more than a year away, there is time to compare three options properly." | | "A one-stop shop." | "We do the audit and the tax return. We refer payroll and legal work to firms we trust." | | "Best-in-class technology." | "We chose our software because it exports cleanly to the accounting systems most clients already use." | The G2 figure comes from [G2's "Answer Economy" research](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html) of 1,076 software buyers. Every figure in expert phrasing needs a source like that. ## How do we make service pages less salesy without hurting conversions? Answer the buyer's question in the first paragraph, show how the service works and where it does not fit, then put the proof and the call to action after the substance. Service pages matter here because, in our experience, "who should we hire" questions are answered from service pages, comparison pages, directories and reviews more often than from blog posts. Example of a rewritten opening: - Before: "We are a leading provider of outsourced payroll services. Our experienced team delivers accurate, compliant payroll for businesses of all sizes, so you can focus on what matters. Contact us today." - After: "We run payroll for UK firms with 20 to 250 staff that have outgrown doing it in-house. Most come to us after a missed pension deadline or a key person leaving. For the first month we run your payroll alongside your old system, so any gap in the data shows up before it costs anyone money. If you have fewer than about ten staff, payroll software is usually enough, and we will say so." The second names who it is for, why they buy, how it works and when not to buy, and it still ends in a call to action. Comparison pages follow the same rules. For Helm Club, a founder peer group: "We built the structured data AI engines read, then a comparison page for every question a founder asks when weighing up peer groups: Helm against YPO, Vistage and the rest." And: "Every comparison page we have published ranks first for its question on Google and the AI engines. The pages not yet built are the ones where a competitor still writes the answer." The [Helm Club case study](/work/helm-club) has the detail. ## How do we write content that reads as expert? Start from what you know that others do not, be specific, say where it does not apply, explain why, and leave out the sell. ### Step 1: Find the observation Ask the person who knows the subject what clients get wrong and what the standard advice misses. ### Step 2: Add the specifics The number, the case, the timeframe and the source. Anonymise clients but keep sector, size and outcome. ### Step 3: Write the limits When this does not apply, what it depends on, what you are not sure of. ### Step 4: Show the reasoning For each claim, say why, and say why you prefer this option to the alternatives. ### Step 5: Take out the sell Delete superlatives, urgency and the closing pitch; end with one calm sentence about where to go next. ### Step 6: Put a name on it Author, role and why they know this subject, consistent with their LinkedIn profile. ## How do we test whether a page is expert content? Read it and ask five questions. The more often the answer is the first option, the more expert the page. 1. Does it contain a thought someone else could not have had, or could anyone in the industry have written it? 2. Are the claims specific, with figures, cases and sources, or general assertions? 3. Does it say when the advice does not apply? 4. Can a reader follow the reasoning step by step? 5. Does it inform a peer, or persuade a prospect? In our experience most firms find their pages fail the first and third questions almost everywhere. The full method for scoring a whole site is in [how to audit existing content for AI assistants](/insights/audit-existing-content-ai-readiness), and [why ChatGPT never quotes most blogs](/insights/why-ai-ignores-most-blog-posts) covers the commonest failure. ## What to do this month 1. Run the five-question test on the five pages that matter most, starting with service pages. 2. Spend fifteen minutes per page with the partner who knows the subject: one observation, one example, one limit. 3. Rewrite each opening so the answer and the observation come first and the sell is gone. 4. Give your writers the phrasing table to apply line by line. 5. Add a named author with role and biography to every page that lacks one, matching their LinkedIn profile. ## Frequently asked questions ### Does ChatGPT check who wrote an article before quoting it? There is no published evidence that it does. Authorship helps indirectly, through Google and through a name that appears consistently on your site, on LinkedIn and in the trade press. Put a real name on every page anyway, because buyers look. ### Does content written with AI count as expert content for Google and ChatGPT? It can, if the expertise is on the page. Google's [guidance on AI-generated content](https://developers.google.com/search/blog/2023/02/google-search-and-ai-content) says "Using AI doesn't give content any special gains. It's just content." An AI draft becomes expert content when the person who did the work adds the observation, the specifics and the limits. ### Should our articles cite other firms' research or only our own? Both. In the Princeton and IIT Delhi test, citing credible sources raised visibility by about 30%, and your own data gives an assistant something it cannot find elsewhere. Always name the original source and its date. [Whether publishing your own research helps](/insights/original-research-data-ai-trust) covers producing it on a small budget. ### How many case studies does our website need to look credible to AI assistants? No study gives a number. Our view: one for each service and sector you want to be named for, each leading with a result. arum's rebuilt site is an example: "Each of the 12 case studies leads with a number, and each of the nine sector pages names the regulation in play." See the [arum case study](/work/arum). ### Does admitting limitations make us look weak to buyers? The opposite, in our experience. A buyer who reads "this approach does not suit firms with fewer than ten staff" trusts the rest of the page more, because the firm has shown it will tell the truth. Length is a separate question: [whether ChatGPT prefers long or short articles](/insights/long-form-vs-short-form-ai-prefers) covers it. The expertise is already in your firm; the job is getting it onto the page. If you want an outside view of how your pages read, [book a free Growth Review](/growth-review): 30 minutes with David Adams and three fixes in writing. If the site itself needs rebuilding around buyers' questions, that is [Website That Sells](/website-that-sells). ## Sources - [arXiv, Aggarwal and others, Princeton and IIT Delhi: source visibility in AI-generated answers, KDD 2024 (June 2024)](https://arxiv.org/abs/2311.09735) - [Google Search Central, Guide to optimizing for generative AI features on Google Search (July 2026)](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide) - [Search Engine Journal, New data: top factors influencing ChatGPT citations, SE Ranking study (November 2025)](https://www.searchenginejournal.com/new-data-top-factors-influencing-chatgpt-citations/561954/) - [TrustRadius, 2026 B2B Buying Disconnect Report (July 2026)](https://www.prnewswire.com/news-releases/trustradius-2026-b2b-buying-disconnect-report-reveals-ai-has-changed-how-buyers-research-but-not-what-they-trust-302825792.html) - [Google Search Central, Creating helpful, reliable, people-first content (December 2025)](https://developers.google.com/search/docs/fundamentals/creating-helpful-content) - [Semrush, LinkedIn AI visibility study (March 2026)](https://www.semrush.com/blog/linkedin-ai-visibility-study/) - [G2, The Answer Economy: half of B2B software buyers now start research with AI chatbots (April 2026)](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html) - [Google Search Central Blog, Google Search's guidance about AI-generated content (February 2023)](https://developers.google.com/search/blog/2023/02/google-search-and-ai-content) --- Title: What should a B2B website include? URL: https://fortitudemedia.ai/insights/what-should-a-b2b-website-include Author: David Adams Published: 2026-04-22 Last reviewed: 2026-09-24 Topic: Websites that bring work in Short answer: A B2B website should include a clear homepage, one page per service and per sector, dated case studies with results, a team page, an honest page on price and process, and an easy contact route. Together those pages must say who you serve, prove it, answer the questions buyers ask before a call, be readable by AI assistants, and make getting in touch easy from a phone. Give it a named owner and update it every month. Ten years ago, a B2B website had a modest job: exist, look professional and give people a phone number. Buyers met you through a referral, an event or a sales call, and the website was where they checked you were real. Many firms still run a site built on that brief. The brief has changed underneath them. Buyers now decide a great deal before they speak to anyone: [6sense's 2025 Buyer Experience Report](https://6sense.com/newsroom/the-timeline-for-influencing-b2b-buyers-is-shrinking-insights-from-6senses-2025-buyer-experience-report/) found buying groups had ranked their preferred vendors before first contact in 94% of cases, and bought from that first choice 77% of the time. More of that research now starts with an AI assistant. In [G2's April 2026 survey of software buyers](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html), 51% said they start in an AI chatbot, up from 29%. For [Helm Club](/work/helm-club), "1 in 5 new members this year came through an AI recommendation." So your website now has two readers making the same decision: an AI assistant deciding whether to name you, and a buyer deciding whether to believe it. A brochure satisfies neither. This guide sets out what a B2B website should include, page by page, and the five jobs those pages have to do. ## Key takeaways - A B2B website is now read by AI assistants deciding whether to name you and by buyers deciding whether to call. - It needs a clear homepage, one page per service, one page per sector, case studies with results, a team page, an honest page on price and process, and an easy contact route. - Those pages have five jobs: say who you serve, prove it, answer the questions buyers ask before the call, be readable by AI, and make contact easy. - Publishing how you price, how long work takes and who you are not right for wins more trust than saving it for the first call. - AI agents that browse and fill in forms for buyers need real buttons, labelled fields and a booking route that works without tricks. - A site that nobody updates drifts out of date, so it needs a named owner and a monthly routine. ## What should a B2B website include? A B2B website should include the pages a buyer needs to shortlist you without calling: who you serve, what you do, proof, how an engagement works and roughly costs, who will do the work, and how to start. Each page should answer one question clearly enough that an AI assistant could repeat it in a sentence. | Page | Its job | Must-haves | |---|---|---| | Homepage | Tell a stranger in seconds whether you are relevant | Who you serve and what you do on the first screen, one or two lines of proof, a clear next step, routes to services and sectors | | Service pages, one per service | Answer "can they do this for us?" | The problem it solves, who it is for and not for, how it works, timescale, how pricing works, a case study | | Sector pages, one per sector | Answer "do they understand our world?" | The sector's pressures and rules named plainly, relevant clients and results, the services that fit | | Case studies | Prove it | A named client where allowed, the problem, what you did, a specific result, a quote from a named person, a date | | Team and about | Show who will do the work | Named people with photos, roles and experience, the firm's history and ownership | | Price and process | Answer the questions buyers ask before a call | How engagements start, how pricing works, typical timescales, what happens after enquiry | | Guides or insights | Show how you think and answer buyers' questions | Articles organised by topic, linked to service pages, each answering one real question | | Contact | Make the next step easy | Phone, email, a short form with labelled fields, a booking link, what happens next and when | Add a current privacy notice and cookie information, which every UK business site needs. ## Job one: say plainly who you serve and what you do A buyer should know within a few seconds whether you are relevant, and an AI assistant should be able to state it in one sentence. That means naming your clients' sectors and size, the problems you solve and the services you offer, in plain words, on the homepage and on every service page. Vague positioning is the most common failure we see. "Integrated solutions for ambitious businesses" could describe anyone. "Outsourced finance teams for UK manufacturers with 50 to 500 staff" can be matched instantly. Structure does the same job at the scale of the whole site. [arum](/work/arum) advises banks, governments and major brands on collections technology, with "more than 500 projects in 23 countries, from offices in London, Atlanta and Sydney". In its rebuild, "Services organised the way a client buys them: five stages from advice to optimisation, plus three specialist services." The work covered "310 pages rebuilt across three regions, in weeks." As Jamie Waller, arum's CEO and Executive Chair, put it: "For most people, our website is the first time they meet arum. It now does us justice." ## Job two: prove it with specific, recent evidence Claims are cheap, so buyers and AI assistants both look for evidence. The strongest evidence on a B2B site is specific, attributed and recent: case studies with a clear problem and result, quotes from named clients, the people who will do the work, accreditations that matter in your sector, and writing that shows how you think. Lead with the number. On arum's new site, "Each of the 12 case studies leads with a number, and each of the nine sector pages names the regulation in play." Recency matters too: a case study from four years ago proves you once did the work; one from last quarter proves you are doing it now. We look at the signals that make buyers trust one firm over another in [what makes a buyer trust a website enough to get in touch](/insights/psychology-of-trust-website-conversions). ## Job three: answer the questions buyers ask before they call A good B2B site answers the questions a buyer would otherwise have to ask on a first call. How does an engagement start? How long does it take? What does it roughly cost? Who is it right for, and who is it not right for? What happens if something goes wrong? Firms often hold these answers back for the first call. The effect is the opposite of what they intend: a buyer who cannot find the answers picks a firm that provides them. An AI assistant asked how much an ISO 27001 audit costs a small business will draw on a firm that has published a clear answer, and it cannot quote a page that says "contact us". To find the questions, ask your sales team and read your inbox: the same dozen questions come up in almost every first conversation. We explain how to answer them in [should our content answer the questions customers actually ask](/insights/content-around-customer-questions). ## Should a B2B firm put its prices on its website? In our view, yes, at least as a starting price, a typical range or a clear explanation of how pricing works. Buyers who are shortlisting without calling use price to rule firms in or out, and a firm that publishes nothing is often ruled out by default. You do not need a rate card. Say what drives the cost, give typical ranges for common projects, and say what is and is not included. If every job really is bespoke, publish two or three worked examples with round figures. We publish our own prices on [one page](/what-it-costs) for the same reason. ## Job four: be readable by AI assistants The site has to let AI crawlers such as GPTBot, ClaudeBot and PerplexityBot in, and serve its words in the page itself rather than filling them in later with scripts. Clear headings, one page per service and sector, and structured data that matches the visible text help assistants describe you accurately. Our [technical checklist for an AI-ready website](/insights/technical-checklist-ai-ready-website) sets out what to ask your developer to check. ## Job five: make getting in touch easy Every page should offer a clear, low-effort next step: a visible phone number and email address, a short enquiry form, a plain statement of what happens next, and an option to book a call. Many buyers first look you up on a phone, often straight after an AI assistant named you. If the contact route is buried, slow or needs a long form, they move on. We cover the details in [do our B2B buyers really look at our website on their phones](/insights/mobile-first-design-b2b). A site that loads slowly, or cannot be used with a keyboard or screen reader, loses people at the moment they were ready to act, so aim for WCAG 2.2 AA. We explain the thresholds in [does our website's speed and accessibility really affect enquiries](/insights/website-accessibility-speed-matter). ## What do AI agents need from our contact form and booking link? AI agents that browse sites and complete tasks for people already exist, and they need the same plain forms that disabled visitors need. OpenAI launched [ChatGPT agent](https://openai.com/index/introducing-chatgpt-agent/) for paid plans in July 2025, and Google's [Chrome auto browse](https://blog.google/products-and-platforms/products/chrome/gemini-3-auto-browse/), which can fill in forms and get quotes, launched in the US in January 2026. [Gartner predicts](https://www.gartner.com/en/newsroom/press-releases/2025-10-21-gartner-unveils-top-predictions-for-it-organizations-and-users-in-2026-and-beyond) that by 2028, 90% of B2B buying will be AI agent intermediated. That is a prediction, not a measurement. Google's [guide to agent-friendly websites](https://web.dev/articles/ai-agent-site-ux) (April 2026) asks for real buttons and links rather than styled boxes, labels linked to their form fields, a stable layout, and no transparent overlays hiding the controls. In practice, for a B2B contact route: - Keep the enquiry form short, with a visible label on every field. - Use real buttons for "send" and "book a call", with plain wording. - Make the booking link open a calendar that works without an account or login. - Keep spam protection that does not demand a puzzle before the form can be sent, where your tools allow it. - Put your email address and phone number on the page as text, not only inside an image or a script. - Say what happens after an enquiry and how quickly, so an agent can report it back to the buyer. ## What is the difference between a brochure website and a lead generation website? A brochure website describes the firm; a lead generation website answers buyers' questions and asks for the next step on every page. The difference shows in what each page contains and whether anyone looks after it. | | Brochure website | Website that brings in work | |---|---|---| | Homepage | Slogan and stock image | Who you serve, what you do, proof and a next step | | Services | One page listing everything | One page per service, each answering a buyer question | | Proof | A logo wall | Dated case studies with results and named quotes | | Price and process | "Contact us" | How pricing works, timescales, what happens next | | AI assistants | Often unreadable or vague | Plain text, clear structure, consistent facts | | Updates | When someone remembers | A named owner and a monthly routine | | Measure | Visits | Enquiries from the right clients, and whether they become work | ## Should a B2B website have an AI chatbot in 2026? Only if it answers from your own published content, hands over to a person quickly, and does not slow the site. A chatbot cannot replace clear pages: AI crawlers read your pages, not your chat widget, so anything the bot knows but the pages do not say is invisible to the assistants buyers use. Chat widgets also slow a page's response to taps, so load one only when a visitor asks for it. For most founder-led firms, a clear page on price and process and a booking link do more. ## How long does it take to build a B2B website in the UK? Usually a few months. [Rudo, a UK agency, puts a mid-market B2B project at 10 to 16 weeks](https://rudo.co.uk/insights/guides/b2b-website-redesign-timeline/) from kick-off to launch (April 2026), and says content production is almost always the bottleneck. [Bigfork says a redesign can take anywhere between 4 to 14 weeks or more](https://www.bigfork.co.uk/insights/all-you-need-to-know-about-b2b-website-redesigns) (October 2024). What decides the timescale is usually content and sign-off rather than code. For arum, "No workshops and no weekly status meetings. Jamie reacted to real, working pages within weeks." Have your case studies, service descriptions and team details ready before the build starts. ## Why does the website have to keep changing? Because a site that stands still drifts out of date, and both buyers and AI assistants notice: people who have left, a newest case study that is years old, services you no longer sell. The fix is not constant redesign. It is a monthly routine owned by one named person: update one existing page, publish one new page or article, and add one new piece of proof. We explain the cost of letting it slide in [what it costs to leave our website unchanged for years](/insights/hidden-cost-static-website). ## How do you tell whether your website is doing these jobs? Score it out of ten on the five lenses we use in our free [Growth Review](/growth-review): 1. Found: what does Google return for the questions your buyers ask, and are you there? 2. Named by AI: when a buyer asks ChatGPT, Gemini, Claude, Perplexity or Google AI Mode who to use, are you named, and is the description accurate? 3. Website: does each page answer a buyer question, prove it and ask for the next step? 4. Authority: where does your firm exist outside its own site, in trade press, directories, review sites and LinkedIn, and do the facts agree everywhere? 5. Pipeline: what does a buyer see after the site, on the founder's LinkedIn and the company page? Then record where every enquiry came from and whether it became work. We set out a fuller method in [how to tell whether your website is working](/insights/evaluate-whether-website-working), and what the site should do for buyers who arrive already told about you in [what your website should do now that visitors arrive pre-sold by AI](/insights/death-cold-website-visit-ai-funnel). ## What to do this month 1. Rewrite the first screen of your homepage so it names who you serve, what you do and how to start. 2. Check your site against the page table above and list the service and sector pages that are missing. 3. Add or refresh one case study that leads with a specific, recent result and a named quote if the client agrees. 4. Write one page that explains how you price, how long work takes and what happens after an enquiry. 5. Fill in your own enquiry form and booking link on a phone, and fix every step that is slow or unclear. 6. Name one person as the website's owner and agree a monthly routine of one update, one new page and one piece of proof. ## Frequently asked questions ### How many pages should a B2B website have? Enough to give each service and each sector its own page, plus the core pages in the table above. One clear page per buyer question beats a few long pages that try to cover everything. ### Most of our work comes from referrals. Do we still need more than a brochure site? Yes, because referred buyers check you too. A recommendation gets you considered, but the buyer will still look at your website, and often ask an AI assistant about you, before calling. A site that confirms the recommendation turns more referrals into conversations. ### Do we need a blog? You need somewhere to publish useful writing, ideally a guides section organised by topic and linked to your service pages rather than a list of posts by date. Each piece should answer a real question your buyers ask. ### What should be on our homepage? Who you serve and what you do, one or two lines of proof and an obvious way to get in touch, all visible without scrolling. Below that, routes to your services and sectors, a recent case study and your people. ### Do we need a full rebuild to meet these standards? Not always. If your site is technically sound and easy to edit, rewriting key pages, adding proof and fixing the technical basics often gets you most of the way. Rebuild when the platform blocks those changes or the structure no longer matches how you sell. A website that does these five jobs becomes a working part of how you win business. To have one built to do that from day one, see how [Website That Sells](/website-that-sells) works, or book a free [Growth Review](/growth-review) to see your five scores and the three fixes we would make first. ## Sources - [6sense, 2025 Buyer Experience Report (November 2025)](https://6sense.com/newsroom/the-timeline-for-influencing-b2b-buyers-is-shrinking-insights-from-6senses-2025-buyer-experience-report/) - [G2, The Answer Economy: half of B2B software buyers now start their research with AI chatbots (April 2026)](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html) - [Gartner, top predictions for IT organisations and users in 2026 and beyond (October 2025)](https://www.gartner.com/en/newsroom/press-releases/2025-10-21-gartner-unveils-top-predictions-for-it-organizations-and-users-in-2026-and-beyond) - [web.dev (Google), Build agent-friendly websites (April 2026)](https://web.dev/articles/ai-agent-site-ux) - [OpenAI, Introducing ChatGPT agent (July 2025)](https://openai.com/index/introducing-chatgpt-agent/) - [Google, Chrome auto browse with Gemini 3 (January 2026)](https://blog.google/products-and-platforms/products/chrome/gemini-3-auto-browse/) - [Rudo, How long does a B2B website redesign take? (April 2026)](https://rudo.co.uk/insights/guides/b2b-website-redesign-timeline/) - [Bigfork, All you need to know about B2B website redesigns (October 2024)](https://www.bigfork.co.uk/insights/all-you-need-to-know-about-b2b-website-redesigns) --- Title: Which law firms does ChatGPT recommend, and how do we become one? URL: https://fortitudemedia.ai/insights/ai-optimisation-law-firms Author: Ross Williams Published: 2026-04-22 Last reviewed: 2026-09-24 Topic: Sector guides Short answer: ChatGPT recommends law firms it can match to a client's problem and check elsewhere: deep pages on a few specialisms, named solicitors with visible experience, Legal 500 and Chambers UK rankings, and an SRA record that matches the Law Society and your website. Most firm websites list every practice thinly. Go deep where you are strongest, answer clients' real questions in plain English, and publish your prices for fixed-fee and subscription work. For example, the founder of a software company with sixty staff has an offer from a trade buyer and six weeks of exclusivity. Before she calls the two firms her accountant suggested, she asks ChatGPT which law firms in Manchester are best for selling a tech company. She gets four names and a note on which are ranked in the Legal 500. Your corporate team completed three similar sales last year. You are not one of the four. Legal work still comes through referrals from accountants, bankers and other clients. But the people referred now check the name with an assistant, and new clients without a referral start there. [Forrester found in January 2026](https://www.forrester.com/blogs/b2b_buyers_make_zero_click_buying_number_one/) that 94% of business buyers use AI in their buying process, and that they are four times as likely as consumers to use Microsoft Copilot. A firm that is thinly described online loses at both stages. Most law firm websites make this harder than it needs to be: twenty practice areas with the same three paragraphs about a "partner-led, pragmatic service", and people pages with a photo and little else. This guide is for the managing partner or head of business development at a firm in England and Wales regulated by the Solicitors Regulation Authority (SRA). In Scotland the Law Society of Scotland regulates solicitors and runs its own Find a Solicitor directory, and in Northern Ireland the Law Society of Northern Ireland is the regulator; the principles are the same. ## Key takeaways - Clients now ask ChatGPT which firm to instruct, and check referrals with it, so a thin website costs you work at both stages. - AI assistants recommend firms they can match to a specific legal problem, so deep pages on a few specialisms beat twenty thin ones. - Legal 500 and Chambers UK rankings, the SRA's Solicitors Register and client reviews are the independent sources the assistants lean on most. - When the SRA record and the Law Society's Find a Solicitor disagree, fix the SRA record first, because the Law Society says it takes its information from the SRA. - Published prices for fixed-fee and subscription services answer one of the commonest questions clients put to assistants, and the SRA already requires them for several services. - Start by choosing two or three practices to lead with, rewriting those pages in full and checking your regulatory and directory details match. ## What do clients ask ChatGPT before they instruct a firm? They describe their problem and situation and ask who can help. They ask about a dismissal, a lease renewal, a shareholder dispute or a sale: - "Which law firms in Birmingham are best for employment law for employers?" - "Do I need a solicitor to sell my company, and how much will it cost?" - "Best commercial property solicitors in Leeds for a lease renewal on our warehouse" - "An employee has raised a grievance after being put on a performance plan. What should I do?" - "Can I get a fixed fee for reviewing a supplier contract?" - "What can I do if a customer disputes a £40,000 invoice and refuses to pay?" Most questions describe a problem in the client's words, not a practice area in a lawyer's. Questions like the last one are also put to collection agencies, and our [guide for debt collection and credit management firms](/insights/ai-optimisation-debt-collection-credit-management) shows how that side of the market competes for them. ## How does ChatGPT decide which law firms to name? It names firms it can match to the client's problem and confirm from several independent sources. Five kinds of evidence matter most: - **A clear match.** Pages that describe the problem, the type of client and the place in plain words. - **Named, credible lawyers.** Profiles showing experience, the matters they handle and what they have written. - **Directory recognition.** Rankings in the Legal 500 and Chambers UK for the practice and region in question. - **Regulatory clarity.** A firm whose name, SRA number and offices match across the SRA's register, Find a Solicitor and your website. - **Client voice.** Reviews on Google, ReviewSolicitors or Trustpilot that mention the kind of matter. Assistants are careful with legal questions, so they lean on sources that have checked you independently. The directories matter more here than in almost any other sector because they rank firms by practice and region, the shape of the questions clients ask. The Legal 500 says it reviews more than 60,000 submissions from law firms a year and [analyses confidential data from over 300,000 commercial law firm clients](https://www.legal500.com/about-us/); [Chambers says](https://chambers.com/about-us) it conducts over 366,000 research interviews and surveys with in-house counsel a year. ### What does a directory ranking give the assistant? A tier or band, named leading lawyers and a researched summary of the team, all tied to a practice and a region: | Prompt a client types | Directory pages an assistant would look for | What a ranked entry gives it | |---|---|---| | "Which law firms in Manchester are ranked by Legal 500 for employment law?" | The Legal 500 UK and Chambers UK employment tables for the region covering Manchester | A tier, the named lawyers and a summary of the employer or employee work | | "Best corporate lawyers in Leeds for selling a tech company" | The corporate and M&A tables for the region covering Leeds | Evidence of deal work at that size, and who led it | | "Which firms are ranked for private client work in Bristol?" | The private client tables for the region covering Bristol | A ranking and named partners an assistant can repeat with confidence | Cited sources change from run to run, so run these prompts for your own practices in Perplexity, which lists its sources under each answer, and note whether the directory pages appear and your firm is on them. ## What should we do when the SRA and the Law Society list us differently? Fix the SRA record first. The SRA describes its [Solicitors Register](https://www.sra.org.uk/consumers/register/) as the definitive source on the firms and people it regulates, and firms update it through mySRA. The Law Society says its [Find a Solicitor](https://www.lawsociety.org.uk/public/for-public-visitors/find-a-solicitor) service collects its information on solicitors from the SRA, so an old office or name there usually traces back to the SRA record or to a profile nobody has updated. Mergers, office closures and name changes leave old listings behind, and a firm that appears in two versions is easy for an assistant to leave out. Correct mySRA, then check Find a Solicitor, the Legal 500 and Chambers, Google Business Profile and the review sites, in that order. ## Why do twenty practice areas work against you? Because an assistant recommends firms for specific problems, and a page covering a whole practice in three paragraphs matches nothing well. Most full-service firms have two or three practices that carry the brand, with ranked partners and the most repeat work. Lead with those, each with pages for the specific problems clients bring, named lawyers and anonymised examples. Your other practices stay on the site; they are simply not where the effort goes first. ### How deep a practice page needs to go A practice page earns its place when it would help a client who read it before a first meeting. For employment work for employers, that means separate pages for grievances, disciplinary processes, redundancy, settlement agreements and tribunal claims. Each should explain what usually happens, what the employer should do first, where the risks are and how the firm charges, with a named author and a date. Open each page with the client's question and answer it in the first paragraph. Accountants often refer clients to law firms and field the same questions from their side, as our [guide for accounting practices](/insights/ai-optimisation-accounting-practices) shows. ## How do you write about the law without giving advice? Explain how the law generally works, what clients should consider and when they need advice, without advising on a specific case. A clear explanation of how a settlement agreement works shows the reader that the process has traps, and makes it more likely they call the firm that explained it. A short note that the page is general information, not advice on an individual matter, is enough for most firms. Keep it accurate and dated. [Paragraph 8.8 of the SRA Code of Conduct for Solicitors](https://www.sra.org.uk/solicitors/standards-regulations/code-conduct-solicitors/) requires that "any publicity in relation to your practice is accurate and not misleading, including that relating to your charges". Date every guide, name the author, say it covers England and Wales, and review it when the law changes. Update, date or redirect old articles, because an assistant may still quote them. Our article on [how old is too old for content](/insights/content-freshness-recency-ai) explains how assistants treat dated material. ## Should we publish fixed fees and subscription packages for business clients? Yes, wherever you offer them. "How much does a solicitor charge" is one of the commonest questions clients put to assistants, and a firm with a clear answer on its own site is the one quoted. For some services the SRA already requires it. The [SRA Transparency Rules](https://www.sra.org.uk/solicitors/standards-regulations/transparency-rules/) require price and service information for residential conveyancing, probate, immigration applications (excluding asylum) and immigration appeals, summary-only motoring offences and employment tribunal claims for unfair or wrongful dismissal brought by employees. For business clients they cover defending employment tribunal claims, debt recovery up to the value of £100,000, and licensing applications for business premises. Firms must also display their SRA number and the SRA's digital badge. For other commercial work, publishing is voluntary, which is why the firms that do it stand out. Paragraph 8.7 of the Code already requires the best possible information about how a matter will be priced. | Pricing model | Suits | What to publish | |---|---|---| | Hourly rates | Matters whose scope cannot be known at the start, such as disputes | Rates by seniority, what drives the total, and a typical range for common matters | | Fixed fee | Defined pieces of work, such as a contract review or a settlement agreement | What is included, what is not, what triggers extra charges, who does the work and turnaround | | Subscription | Businesses with a steady flow of routine questions | What the monthly fee covers, response times, any cap, the minimum term and what falls outside | State whether prices include VAT and disbursements. For the listed services the rules ask for the total cost or, where not practicable, an average or range; the same approach works for commercial fixed fees. ## Which outside sources matter most for a law firm? The independent sources that already regulate, rank and review you: - **The SRA.** Your firm's entry on the Solicitors Register, and the digital badge on your website. - **The Law Society.** Your Find a Solicitor listing, plus accreditations such as Lexcel, the Conveyancing Quality Scheme and the Wills and Inheritance Quality Scheme where relevant. - **Legal directories.** The Legal 500 and Chambers UK. Submissions take time, but assistants reach for them when asked for the best firms in a practice and region. - **Specialist bodies.** Resolution for family lawyers, STEP for trusts and estates, the Employment Lawyers Association and others that fit your practices. - **Legal and sector press.** Comment in the Law Society Gazette, Legal Futures or The Lawyer and, for a firm with sector specialisms, the trade press of the industries it serves. - **Reviews.** Google, ReviewSolicitors and Trustpilot, with clients describing the kind of matter in their own words. Financial advisers are often introduced alongside solicitors on private client matters and face similar questions, which our [guide for financial advisory firms](/insights/ai-optimisation-financial-advisory) covers. ## Why do individual solicitors matter as much as the firm? Clients instruct lawyers, not brands, and when a client asks about the person they are about to meet, the assistant builds the answer from that person's public record. A strong profile says what the solicitor does and for whom, the matters they have handled, their directory rankings, their publications and when they qualified, as text on the page rather than a PDF. Partners who publish short pieces under their own names build that record, and a partner too busy to write can be interviewed for twenty minutes and then check the draft. Named authors with visible experience are one of the strongest signals of expertise, as we explain in [why AI assistants care about E-E-A-T](/insights/eeat-why-ai-cares-more). ## How do you tell whether it is working? Track a fixed list of 20 to 50 client questions, in their words, including several that name a practice and a region. Ask each several times a month in ChatGPT, Gemini, Claude, Copilot and Perplexity, in a clean session, because answers vary from run to run, and count a win only when it repeats. Record whether the firm is named, which lawyers are mentioned and who is named instead. Ask every new client how they found you; referred clients who checked you on ChatGPT count too. Assistants that search the web live can reflect a corrected listing within weeks, while answers from a model's memory change only when it is retrained, which takes months. Our dated example is [ID-Pal](/work/id-pal), whose buyers include law firms: "UK buyer questions where AI names ID-Pal: none in May, two in August, four in September 2026." ## What to do this month 1. **Correct your regulatory record.** Update mySRA, then make Find a Solicitor, the Legal 500, Chambers, Google Business Profile and your website agree on names, offices and practices. 2. **Choose two or three practices to lead with**, the ones with ranked lawyers and the strongest client base. 3. **Rewrite one practice section in full**: five pages on the problems clients bring, each by a named solicitor, dated and in plain English. 4. **Publish your prices** for every service the Transparency Rules cover, and for your commercial fixed fees and subscriptions. 5. **Upgrade the profiles of the lawyers in that practice**: plain text, specific experience, rankings and publications. 6. **Set up your question list** and run it in clean sessions several times a month. ## Frequently asked questions ### Will publishing detailed legal guides reduce our billable work? In our experience, no. Clients who read a clear explanation usually see how much can go wrong and want someone to handle it, and the firm that explained the problem is the one they are most likely to call. ### How much do Legal 500 and Chambers rankings matter for this? A great deal. They are independent, researched and organised by practice and region, the same shape as the questions clients ask. If you are not ranked, a well-prepared submission with strong client referees is worth the effort alongside the website work. ### How do I check a solicitor is SRA regulated and has no disciplinary record? Search the SRA's Solicitors Register to confirm the person is a solicitor, where they work and that the SRA regulates the firm. The SRA also [publishes many of its regulatory and disciplinary decisions](https://www.sra.org.uk/consumers/solicitor-check/), searchable by name, and says they generally stay published for three years; not every decision is published. Firms make this easier by showing their SRA number and digital badge on every page. ### How much does a solicitor charge to handle the sale of a limited company? It depends on the size and structure of the deal and how much due diligence the buyer wants, so we do not quote market figures. Company sales are not on the SRA's transparency list, but the Code requires the best possible information about pricing. A firm that publishes what drives the cost and a typical range for its usual deal size gives the assistant, and the founder, a real answer. ### Is a fixed-fee commercial lawyer better than an hourly-rate firm for a small business? For defined work, a fixed fee gives a small business certainty. For disputes or deals whose scope changes, hourly billing with a clear estimate may be fairer to both sides. What matters is that the scope and the triggers for extra charges are written down before work starts. ### Should a startup use a law firm or a legal tech platform for its shareholders' agreement? A template platform can suit a simple arrangement between founders who agree on everything. A lawyer earns the fee when there are different share classes, investors, vesting or leaver terms that need thinking through. A firm that publishes an honest page on when each makes sense is the kind of source assistants quote. The firms that get named are the ones whose expertise is written down, dated and confirmed by others. If you want help turning your partners' knowledge into published guidance, directory recognition and press comment that clients and assistants both trust, [Reputation That Wins](/reputation-that-wins) is built for exactly that. ## Sources - [Forrester, B2B buyers make zero-click buying number one (January 2026)](https://www.forrester.com/blogs/b2b_buyers_make_zero_click_buying_number_one/) - [SRA, Transparency Rules (September 2026)](https://www.sra.org.uk/solicitors/standards-regulations/transparency-rules/) - [SRA, Code of Conduct for Solicitors, RELs and RFLs (September 2026)](https://www.sra.org.uk/solicitors/standards-regulations/code-conduct-solicitors/) - [SRA, Solicitors Register (September 2026)](https://www.sra.org.uk/consumers/register/) - [SRA, Check a solicitor's record (September 2026)](https://www.sra.org.uk/consumers/solicitor-check/) - [The Law Society, Find a Solicitor (September 2026)](https://www.lawsociety.org.uk/public/for-public-visitors/find-a-solicitor) - [Legal 500, About us (September 2026)](https://www.legal500.com/about-us/) - [Chambers and Partners, About us (September 2026)](https://chambers.com/about-us) --- Title: Why does ChatGPT name the big firms and not our consultancy? URL: https://fortitudemedia.ai/insights/ai-optimisation-management-consultancies Author: Ross Williams Published: 2026-04-22 Last reviewed: 2026-09-24 Topic: Sector guides Short answer: Boutique consultancies get named by ChatGPT when they describe the specific problems, sectors and outcomes they handle in plain words. The big firms win broad questions, but most buyers ask narrow ones. Publish anonymised case studies with a clear output and timescale, give partners visible profiles, get named in deal announcements, and back it up with MCA membership, your framework lot and coverage in your clients' sector press. For example, the chief operating officer of a food manufacturer with two sites wants help cutting waste and changeover times before the next round of price talks with the supermarkets. Before calling anyone, he asks ChatGPT which consultancies help UK food manufacturers improve OEE and reduce waste. He gets three large firms he has heard of and two specialists he has not. Your partners spent twenty years running food factories and have done this work for a dozen manufacturers. You are not mentioned. When he asks "should I use a big consultancy or a boutique for this", the boutiques it names are the same two. Consulting is still sold on reputation and relationships: partners' networks, private equity sponsors and the client who moves company and brings you with them. But buyers now use ChatGPT, Claude, Gemini, Copilot, Google's AI answers and Perplexity to build the long list and sense-check a recommendation, and the list forms early. [6sense's 2025 Buyer Experience Report](https://6sense.com/newsroom/the-timeline-for-influencing-b2b-buyers-is-shrinking-insights-from-6senses-2025-buyer-experience-report/), a survey of more than 4,000 B2B buyers, found buying groups had ranked their preferred vendors before first contact in 94% of cases, and bought from that first choice 77% of the time. When the assistant has little to say about you, you start behind, or not at all. Most boutique websites make this harder than it needs to be. "We partner with ambitious organisations to deliver sustainable transformation" could describe a thousand firms. This guide is for the founder or managing partner of a UK consultancy who wants to be named for the problems the firm solves best. ## Key takeaways - Buyers now ask ChatGPT which consultancy to hire and use it to check recommendations, so vague positioning quietly costs you work. - The big firms get named for broad questions, but most buyers ask narrow ones, which is where a boutique can win. - AI assistants name consultancies whose websites describe specific problems, sectors and outcomes in the words buyers use. - Anonymised case studies with a clear output and timescale, plain explanations of your method and strong partner profiles are the evidence assistants quote. - Private equity sponsors, deal announcements and the trade press that reports deals are a channel of their own, and they need their own evidence. - MCA membership, Chartered Management Consultant status, a place on a public framework and your clients' sector press confirm what your website claims. ## What do buyers ask ChatGPT before they hire a consultancy? They describe a problem, a sector and sometimes a constraint, and ask who can help or how to go about it. The person asking is usually a chief executive, operations director, finance director or private equity operating partner: - "Which consultancies help mid-sized UK manufacturers improve operational efficiency?" - "Should we use a big four firm or a boutique for a finance transformation?" - "Which consultancies do private equity firms use for value creation in the UK lower mid-market?" - "Should I hire a consultancy or an interim manager for an operations turnaround?" - "Which consultancies are on the government framework for NHS work?" - "Which consultancies specialise in procurement for housing associations?" The broad questions produce the famous names, because they are the most written-about firms. The narrow ones, which are most of them, produce whichever firms the assistant can match to the sector and problem. "How should we" questions are answered from whoever has explained the approach most clearly, so a consultancy with a plain guide to integrating two businesses of similar size is the natural firm to name alongside that answer. ## How does ChatGPT decide which consultancies to name? It names firms it can match to the buyer's problem and describe with confidence. For a consultancy, that confidence comes from four kinds of evidence: - **A specific match.** Pages that name the problem, the sector, the size of client and the outcome, in the buyer's words. - **Proof of work.** Case studies, even anonymised ones, that show the situation, the approach and what changed. - **Credible people.** Partners with visible industry experience, qualifications and published thinking. - **Independent confirmation.** Trade body membership, framework places, press coverage and sponsor mentions that agree with your own description. A boutique cannot match the volume written about the big firms and does not need to. It can be the best-documented firm for a narrow combination: procurement for housing associations, operational improvement for food manufacturers, commercial due diligence on software deals. For those questions the assistant is looking for the closest match, not the biggest name. ## Why does abstract positioning make you invisible? Because an assistant cannot match a buyer's problem to a sentence that could describe anyone. "Transformation partner for ambitious organisations" gives it nothing. "We help UK food and drink manufacturers with one to five sites cut waste and changeover time, led by partners who ran factories" gives it a clear match for a real question. Narrowing the description does not have to narrow the business. Lead the homepage with your strongest two or three offers and give each a proper section: the problem, the sector, the typical client, the approach, the outcomes, the people and the case studies. You can still take other work. Explain your method in plain words, not trademarks. A named model such as "our 5D framework" can help in a pitch, but an assistant will not match a trademark to a buyer's question. Say how you diagnose, what you measure, how you involve the client's team and how you make improvements stick. Our article on [how ChatGPT tells real expertise from keyword-stuffed content](/insights/thought-leadership-vs-keyword-stuffing) explains why specificity wins. ## How do you publish case studies when every client is confidential? Anonymise the client, keep the specifics of the work and describe outcomes in ranges or relative terms where exact figures are sensitive. A useful case study covers: - The client type, such as "a family-owned food manufacturer with two sites and around 400 staff". - The trigger: a margin squeeze, an acquisition, a failed system implementation or a new chief executive. - What you did, in a few plain paragraphs, including what was difficult. - The output and the timescale: what the client had at the end, and how long it took. - What changed, as precisely as the client allows. "Changeover time roughly halved on the main line" beats "significant improvement". - What the client's own team could do afterwards without you. Ask for permission to publish at the start of an engagement, not at the end. Many clients who will not be named will agree to a detailed anonymous write-up, and where they will not, publish aggregated lessons, such as what a dozen post-merger integrations taught you. ## Worked example: how we write up an engagement Here is that structure applied to our own work for [TCC Group](/work/tcc-group), which provides regulatory compliance for UK financial services through three brands: TCC for advisory and managed services, Recordsure for compliance AI and Momenta for specialist resourcing. The quoted lines are from our case study. | Element | How the case study puts it | |---|---| | The trigger | "ChatGPT already placed TCC top for outsourced suitability reviews. But on Consumer Duty and Section 166, the two areas the group backs hardest, the assistants named other firms and not TCC. One had stopped recognising the UK firm altogether, and none of them connected the three brands." | | The evidence | "36 recorded answers from ChatGPT, Gemini and Claude, every conversation kept and reproducible." and "624 pages reviewed, each given a decision: migrate, merge or retire." | | What we found | "What failed was underneath: a site organised around internal teams rather than the questions buyers ask, pages that half-answered the same question so neither was cited, and machine-readable data that declared a name, a logo and nothing else." | | Output and timescale | "1 week to a plan TCC's own team built the new site from." | | The new structure | "Ten solution pages, one for each problem a buyer types, from Section 166 and FCA intervention to vulnerable customers, plus sector hubs and one group page telling the story once." | | What the client did without us | "TCC chose to build it with its own team, and the new tcc.group is live, built to the plan, with nothing about it depending on Fortitude." | Two things make this quotable: a named output with a timescale, and a plain statement of what the client could do afterwards. In our experience, most consultancy case studies have neither. ## Why do your partners' profiles matter as much as the firm's? Buyers hire people, and when they ask about the partner coming to pitch, the answer depends on that person's public record. A strong profile says what the partner does and for whom, the roles they held before consulting, the sectors and problems they know, their qualifications and what they have written, as plain text rather than a PDF CV. Partners who write a short monthly piece on the problems they see, on the firm's site and on LinkedIn, build a record an assistant can connect to the firm. Written thinking also reaches the people who shape decisions quietly. In the [2025 Edelman and LinkedIn B2B Thought Leadership Impact Report](https://www.edelman.com/expertise/Business-Marketing/2025-b2b-thought-leadership-report), a survey of US executives, 64% of "hidden" decision-makers said they trust thought leadership more than marketing materials to assess a firm's capabilities. Show [Chartered Management Consultant status](https://www.managers.org.uk/membership/chartered-management-consultant/) where partners hold it. It is created and managed by the Management Consultancies Association and the Chartered Management Institute. We could not find a public register of holders, so state the year awarded on each profile. ## How do private equity firms find consultancies, and how do you get named? Through the same deal network as always, now checked with an assistant. An operating partner asking which consultancies do value creation for lower mid-market portfolio companies gets the firms whose names appear where deals are recorded, so that is where your evidence needs to be: - **Deal announcements that name advisers.** Commercial due diligence and value creation advisers are often listed in deal press releases and in the trade press that reports deals, such as Real Deals and Insider Media. Ask sponsors to name you, and keep your own dated list of the deals you advised on. - **Sponsor and portfolio pages.** A private equity firm that lists you as an adviser, or a portfolio company case study that names you, is independent confirmation. - **Case studies written for sponsors.** The investment thesis, the hundred-day plan and what changed by exit, anonymised where needed. - **Partner profiles that say "private equity"**, with the fund sizes and sectors worked with. Our [guide for private equity and venture capital firms](/insights/ai-optimisation-private-equity-vc) shows the questions those sponsors are asking about their own portfolios. ## Which outside sources matter most for a consultancy? The ones buyers already trust to vouch for consultancies: - **The Management Consultancies Association.** The [MCA says](https://www.mca.org.uk/membership) all member firms "have to meet stringent entry criteria" and "are audited periodically", and it publishes a member directory. The MCA Awards are a further signal if you have a strong entry. - **Public sector frameworks.** The Government Commercial Agency, formerly the Crown Commercial Service, runs [Management Consultancy Framework Four (MCF4, RM6309)](https://www.gca.gov.uk/agreements/RM6309) from 29 July 2025 to 28 July 2027, with ten lots: Business; Strategy and policy; Complex and transformation; Finance; HR; Procurement and supply chain; Health, social care and community; Infrastructure; Environment and sustainability; and Restructuring and insolvency. State your lot on the matching service page. Example: "We are a supplier on MCF4 (RM6309), Lot 6: Procurement and supply chain, until 28 July 2027." - **Rankings.** The Financial Times' annual list of the UK's leading management consultants is compiled with Statista from surveys of consultants and clients, so there is no form to fill in; you are recommended onto it. - **Consulting press and research.** A firm profile and news coverage on Consultancy.uk, and mentions in the work of Source Global Research, the analyst firm that studies the consulting market. - **Clutch**, for digital and technology consultancies, where reviews carry the client's sector and project size. - **Your clients' sector press.** For a specialist, a column in the food manufacturing, housing or healthcare trade press is worth more than any consulting publication. Earned coverage matters most: [Muck Rack's May 2026 study](https://www.globenewswire.com/news-release/2026/05/07/3290268/0/en/generative-pulse-earned-media-consistently-drives-ai-citations-holding-at-84.html) of more than 25 million links found earned media made up 84% of the links AI tools cited, and paid content 0.3%. Keep the firm's name, description and partner names consistent across all of them. ## How do you tell whether it is working? Track a fixed list of 20 to 50 buyer questions, including the "big four or boutique" question for your sector and several that name your sector and problem. Ask each several times a month in ChatGPT, Gemini, Claude, Copilot and Perplexity, in a clean session, because answers vary from run to run, and count a win only when it repeats. Record who is named, and in Perplexity, which lists its sources, what evidence each named boutique had. Then ask every new prospect how they came to you, and count invitations to tender from sectors you have written about. Corrections to listings can show within weeks when assistants search the web live; becoming the firm named for a sector and problem takes months. ## What to do this month 1. **Name your three problems.** Agree the three combinations of problem and sector where you are strongest, and rewrite the homepage to lead with them. 2. **Rewrite one service section in full**: the problem, the typical client, your method in plain words, the outcomes, the people and two case studies. 3. **Upgrade partner profiles**: prior roles, sectors, qualifications and published work, in plain text. 4. **List your deals and sponsors.** Collect every deal announcement that names you and ask sponsors to name you on their sites. 5. **Agree a case study routine.** Ask for publication permission at the start of every engagement, and write up one completed project a month. 6. **Set up your question list** and run it in clean sessions several times a month. ## Frequently asked questions ### Can a small consultancy be named ahead of the big firms? For narrow questions, yes. The big firms dominate broad questions because so much is written about them. Most buyers ask about a specific problem in a specific sector, and there a boutique that documents that combination clearly can be the better match. ### What is a typical day rate for a boutique management consultancy in the UK? We do not quote market rates here, because they vary widely by seniority, sector and how the work is scoped. Buyers do ask assistants what consultants cost, so explain how you price: fixed fee or time-based, how you scope the work and what pushes the cost up or down. Some firms give indicative ranges for typical projects. ### Should I hire a consultancy or an interim manager for an operations turnaround? In our view, an interim manager suits a gap at the top: someone to run the operation for a period. A consultancy suits a team that is in place but needs a diagnosis, a method and extra hands to deliver it. If you offer both, publish a page that says when each fits, because it is exactly the question buyers put to assistants. ### Is MCA membership worth it for a small consultancy? It depends on who you sell to. Membership means passing the MCA's entry criteria and periodic audit and committing to its Consulting Excellence standards, and the MCA runs an SME Group for smaller firms. For a boutique selling to large companies or the public sector, that independent check is evidence an assistant can find. For a firm selling on sector expertise, the sector's own trade body may matter more. ### Will publishing our method let competitors copy it? Rarely in any way that matters. Competitors cannot copy your partners' experience, your client relationships or your judgement in the room. A clear explanation reassures buyers that there is a method, and gives assistants something specific to quote. ### Do awards and rankings help? They help as independent confirmation, because assistants treat them as someone else vouching for you. On their own they will not get you named for a specific problem. Combined with clear service pages, case studies and credible partners, they make an assistant more confident describing you. The consultancies that get named are the ones whose work is described as plainly as it is delivered. That record takes months to build. Researched outbound, where a person writes and approves every message, reaches buyers before any assistant is asked, and [Pipeline That Fills](/pipeline-that-fills) puts you in front of decision-makers in the sectors you know best while it builds. ## Sources - [6sense, 2025 Buyer Experience Report (November 2025)](https://6sense.com/newsroom/the-timeline-for-influencing-b2b-buyers-is-shrinking-insights-from-6senses-2025-buyer-experience-report/) - [Edelman and LinkedIn, 2025 B2B Thought Leadership Impact Report (June 2025)](https://www.edelman.com/expertise/Business-Marketing/2025-b2b-thought-leadership-report) - [Muck Rack, What is AI reading? (May 2026)](https://www.globenewswire.com/news-release/2026/05/07/3290268/0/en/generative-pulse-earned-media-consistently-drives-ai-citations-holding-at-84.html) - [Government Commercial Agency, Management Consultancy Framework Four (MCF4) (September 2026)](https://www.gca.gov.uk/agreements/RM6309) - [Management Consultancies Association, Membership (September 2026)](https://www.mca.org.uk/membership) - [Chartered Management Institute, Chartered Management Consultant (September 2026)](https://www.managers.org.uk/membership/chartered-management-consultant/) --- Title: Why does ChatGPT recommend other recruitment agencies to employers? URL: https://fortitudemedia.ai/insights/ai-optimisation-recruitment-agencies Author: Ross Williams Published: 2026-04-22 Last reviewed: 2026-09-24 Topic: Sector guides Short answer: ChatGPT and the other assistants recommend recruitment agencies with a clearly stated niche and published proof of it: dated salary guides, market notes and placement stories. They check that against REC or APSCo membership, trade press and reviews from clients and candidates. Executive search and RPO firms need their own evidence, such as AESC or APSCo OutSource membership. Choose one niche to lead with, publish a dated salary guide and make your listings match. The founder of a growing medical devices company needs a head of quality who knows device regulation inside out. Last time, a generalist agency sent her CVs from people who had never worked in a regulated environment. This time she asks ChatGPT: "which recruitment agencies specialise in quality and regulatory roles in medical devices in the UK". She gets four names. Your desk has placed more quality leaders in medtech than any of them. You are not on the list. Hiring managers still call the consultant who filled the last role. What has changed is what happens when the hiring manager is new, the role is new to the business or the finance director wants three agencies to compare. The first names now often come from an AI assistant. Forrester found in January 2026 that 94% of business buyers use AI somewhere in their buying process, and that business buyers are four times as likely as consumers to use Microsoft Copilot, which is built into Microsoft 365 ([Forrester, January 2026](https://www.forrester.com/blogs/b2b_buyers_make_zero_click_buying_number_one/)). Candidates do the same from the other side, asking which recruiters to trust. Most agency websites are job boards with a paragraph of "about us" and a list of sectors ending in "and more", which gives an assistant nothing to separate one agency from the next. This guide is for the founder of a UK recruitment agency, executive search firm or RPO provider who wants to be named when employers in their niche ask for help. ## Key takeaways - Employers and candidates now ask AI assistants which recruitment agencies to use, and the answers favour agencies with a clearly stated niche. - A dated salary guide with a stated method and sample size is among the most quotable things a recruitment agency can publish. - Executive search and RPO are separate niches with their own sources, such as AESC membership for retained search and APSCo OutSource for recruitment process outsourcing. - REC or APSCo membership, trade press and reviews confirm what your website claims, but only when the details match everywhere. - Dated explainers on the Employment Rights Act 2025 and the Fair Work Agency give hiring clients a reason to read you and assistants a reason to cite you. - Start by choosing one niche to lead with, publishing a short salary guide for it and making your listings match. ## What do employers and candidates ask AI assistants? They ask about the role, the market, the law and who can help, in very specific terms. Typical employer questions: - "Which recruitment agencies specialise in finance roles in Leeds?" - "Which recruitment agencies are best for fintech roles in London?" - "Retained or contingent search for a head of sales: which is better?" - "What is the going salary for a senior quantity surveyor in Bristol?" - "What does the Employment Rights Act mean for our probation periods?" - "Is it worth using a recruiter now that AI tools can screen CVs?" Candidates ask "best recruiters for data engineering jobs in Manchester", "what salary should I ask for as a newly qualified solicitor outside London?" and "is this recruitment agency good to work with?" Two patterns stand out. The employer questions nearly always name a function, a sector or a level, and often a place, and an assistant can only match those words to agencies whose websites use them about real work. And many questions are about pay, law and process rather than agencies. The assistant answers those from whoever has published the clearest information, and often names that source. ## How does ChatGPT decide which agencies to name? It names agencies it can describe with confidence: a clear niche, evidence of success in it, named consultants with visible experience and independent sources that agree. The assistants that matter most are ChatGPT and Google's AI Overviews and AI Mode, then Microsoft Copilot and Gemini, then Perplexity and Claude, and all of them reward the same things. - **A niche in plain words.** "Permanent and interim finance recruitment for businesses in Yorkshire" can be matched to a question. "Delivering talent solutions across multiple sectors" cannot. - **Evidence you know the market.** Salary data, market commentary and placement stories for that niche. - **Named consultants.** Profiles that say which market each person covers and for how long. - **Independent confirmation.** Trade body listings, trade press, reviews and LinkedIn all pointing the same way. Agencies that have grown by adding desks often list twelve sectors with the same two paragraphs under each. To an assistant, that reads as a generalist with no depth anywhere. Most good agencies combine two lines of specialism, such as finance roles in Yorkshire or engineering roles in defence, and that combination is exactly the phrase buyers type. The pattern holds in any regulated market. When we rebuilt ID-Pal's industry pages, "Each opens with the question a compliance lead types, names the regulator, and answers inside the first ninety words." An agency's niche page can do the same for a hiring manager. ### Can an agency pay to appear in the answer? It can pay for a labelled advert, which is not the same as being recommended. ChatGPT has shown labelled ads to UK users on its Free and Go plans since 11 August 2026, and OpenAI says the ads do not change the answer itself; its Plus, Pro, Business, Enterprise and Education plans show no ads ([OpenAI, August 2026](https://openai.com/index/testing-ads-in-chatgpt/)). Being named in the answer is still earned. ## How do executive search and RPO firms get named? Treat each as its own niche, because the buyers, the questions and the sources differ. A retained search firm is chosen by a chair, chief executive or investor for one senior hire. An RPO provider is chosen by an HR director or procurement team to run hiring for a contract period. | Niche | Who asks | What they type | What the assistant can check | |---|---|---|---| | Contingent agency | Hiring managers, HR, founders | "finance recruitment agency in Leeds" | Niche pages, salary guide, REC or APSCo listing, reviews | | Executive search | Chairs, chief executives, investors | "executive search firms that place CFOs for private equity backed businesses" | Named partners, AESC membership, Executive Grapevine listing, board-level press | | RPO | HR directors, talent acquisition heads, procurement | "RPO providers in the UK for high-volume hiring" | Programme case studies, APSCo OutSource membership | For retained search, publish partner profiles that say which boards and functions each partner covers, anonymised searches described by role and ownership type, and your process from brief to appointment. Show your membership of the Association of Executive Search and Leadership Consultants, whose [member firm directory](https://www.aesc.org/membership/aesc-member-firms/) is public, and keep your entry in Executive Grapevine's database of executive recruiters current. For RPO, the buyer wants proof you can run volume without losing quality: case studies stating the roles, programme size, contract length and the measures agreed, with figures the client has approved. APSCo OutSource describes itself as the trade association for recruitment process outsourcing, managed service providers and statement of work ([APSCo OutSource](https://outsource.apsco.org/)), and membership there does for an RPO provider what REC or APSCo membership does for an agency. ### Retained or contingent: which should a client choose? For a senior, confidential or scarce role, such as a head of sales, retained search usually suits better, because one firm is paid to map the market and approach people who are not looking. Contingent recruitment suits active markets where several agencies can present strong candidates quickly. Explain this on your own site and say which model you offer. | | Contingent | Retained | |---|---|---| | Fee | Paid only when a hire is made | Paid in instalments through the search | | Exclusivity | Often several agencies at once | One firm, exclusively | | Method | People the agency can reach quickly | Full market map and direct approach | | Best for | Active markets, repeat hires | Senior, confidential or scarce roles | ## Why do salary guides get an agency named? Because pay is one of the questions employers and candidates ask most, and the assistants need a source to answer it. An agency that publishes clear, current salary data for its niche becomes that source. The guide needs to be specific to the roles you place, clear about its method, dated and written in prose as well as tables, because numbers locked in a PDF or an image are much harder for an assistant to read. Example: the excerpt below shows the format with round, made-up figures. It is not salary data. **Financial controller, permanent, base salary. Based on our placements from July 2025 to June 2026. Published September 2026.** | Region | Typical range (illustrative) | Placements behind the range | |---|---|---| | Leeds | £60,000 to £70,000 | 20 | | Manchester | £60,000 to £75,000 | 30 | | Birmingham | £55,000 to £65,000 | 15 | Commentary: "Manchester ranges widened as more private equity backed firms hired their first financial controller. Candidates with group consolidation experience were offered the top of the range in all three cities." That is the whole pattern: one role, one contract type, a stated period, the number of placements behind each range, a publication date and two sentences of plain commentary. Our article on [publishing your own research](/insights/original-research-data-ai-trust) covers how to present data so it is trusted. ## What should agencies explain about the Employment Rights Act 2025? Explain what changes for your clients' hiring and when, in dated plain English. The Act received Royal Assent on 18 December 2025 ([legislation.gov.uk](https://www.legislation.gov.uk/ukpga/2025/36/introduction)), and its measures arrive over 2026 and 2027. This table follows the government's timeline, checked on 24 September 2026. | Date | What changes | |---|---| | 6 April 2026 | Statutory Sick Pay without waiting days or the lower earnings limit; day-one paternity and unpaid parental leave | | 7 April 2026 | The Fair Work Agency set up | | 1 October 2026 | Time limit for tribunal claims rises from three to six months | | 30 October 2026 | Employers must take all reasonable steps to prevent sexual harassment, and must not permit harassment by third parties | | 1 January 2027 | Unfair dismissal qualifying period falls to six months, and compensatory awards are uncapped | | 2027, after consultation | Right to guaranteed hours and reasonable notice of shifts, with provisions for agency workers; regulation of umbrella companies | The government says future dates "remain subject to parliamentary processes and may change" ([GOV.UK, August 2026](https://www.gov.uk/government/publications/implementing-the-plan-to-make-work-pay-and-employment-rights-act/plan-to-make-work-pay-and-employment-rights-act-timeline-update)), so put a "last checked" date on any page you write and review it when the timeline changes. In our experience three short pages work best: one for permanent hiring clients on probation and the new qualifying period, one for temporary and contract clients on guaranteed hours and umbrella companies, and one on who enforces what. Enforcement has moved. The Conduct of Employment Agencies and Employment Businesses Regulations 2003 still set the rules on terms with hirers ([legislation.gov.uk](https://www.legislation.gov.uk/uksi/2003/3319/contents)), but from 7 April 2026 the Fair Work Agency took over from the Employment Agency Standards Inspectorate and the Gangmasters and Labour Abuse Authority ([REC, March 2026](https://www.rec.uk.com/our-view/news/fair-work-agency-launches)). Update any page that still names the old bodies. ## How do you show a track record without breaching confidentiality? Write short, anonymised placement stories: the kind of client ("a private equity backed engineering group with four UK sites"), the role and why it was hard, how you ran the search in two or three sentences, the outcome and a line from the client given with permission. For example, an agency placing around a hundred engineers a year could publish twenty such stories grouped by role and client type, which gives an assistant far more than a page of logos. Candidates' experience feeds the same answers. Reviews on Google, Trustpilot and Glassdoor shape the summary a hiring manager sees, so a pattern of ghosted candidates costs you clients as well as candidates. Our guide on [what ChatGPT tells candidates about working at your firm](/insights/ai-search-recruitment-talent) covers the employer side of this. ## Which outside sources matter most for an agency? Independent sources that confirm what your website says, with the same name and description everywhere. | Source | What it confirms | What to check | |---|---|---| | Recruitment and Employment Confederation (REC) | Corporate members pass a compliance assessment every two years ([REC](https://www.rec.uk.com/recruiters/compliance/code-professional-practice)) | Your REC directory entry matches your site | | Association of Professional Staffing Companies (APSCo) | Members follow a Code of Conduct and pass a risk assessment every three years ([APSCo](https://uk.apsco.org/discover-apsco/why-use-an-apsco-member/quality-and-compliance)) | Your sector, specialism and region tags in "Find an APSCo Member" | | AESC or APSCo OutSource | Standing in retained search or RPO | Membership shown on partner and service pages | | Companies House | The company behind the brand | The link to your trading name is stated | | Trade press | Comment in Recruiter, Recruitment International and your clients' sector titles | Consistent name and bylines | | Reviews | Client and candidate views | Reviewers mention the niche in their own words | Set out your approach to the Conduct Regulations, the Agency Workers Regulations and IR35 plainly on the site. Agencies that have rebranded or merged desks often have old names live on directories and job boards, which leaves the assistant with two versions of the same firm. ## How long does it take, and how do you tell it is working? Fixes to pages and listings can show within weeks in assistants that search the web live, such as ChatGPT search, Perplexity, Copilot and Google's AI features. Answers drawn from a model's own memory change only when it is retrained, which takes months. For ID-Pal, an identity verification firm we work with: "UK buyer questions where AI names ID-Pal: none in May, two in August, four in September 2026." The [ID-Pal case study](/work/id-pal) explains how. Track a fixed list of 20 to 50 questions your clients and candidates ask, each run several times a month in a clean session, and count a win only when it repeats. Our guide on [finding out whether ChatGPT recommends your firm](/insights/track-ai-recommending-your-business) sets out the method. Then ask every new client how they found you and record it in your CRM. ## What to do this month 1. **Choose the niche you lead with.** Pick the one or two combinations of function, sector, level and region where you place most and best, and rewrite your homepage opening to say so. 2. **Publish a short salary guide.** Use last year's placements for that niche, with a method, a sample size and a date, as a web page. 3. **Write one dated Employment Rights Act explainer.** Cover the next twelve months for your clients, with a "last checked" date and a link to the government timeline. 4. **Write five placement stories.** Anonymised, specific and tied to the niche. 5. **Fix your listings.** Match your name and description across Companies House, REC or APSCo, AESC if relevant, Google Business Profile, job boards and LinkedIn, and replace any mention of the EAS or GLAA with the Fair Work Agency. ## Frequently asked questions ### What is the average recruitment agency fee for a permanent hire in the UK? We do not quote a market figure here, because fees vary with the level of the role, the market, exclusivity and whether the search is retained or contingent. Permanent fees are most often a percentage of the candidate's first-year salary, with a rebate or replacement if the hire leaves early. A page that explains how you charge, what changes the fee and what your guarantee covers gives the assistant something accurate to quote. ### Which recruitment agencies are best for fintech roles in London? An assistant answers this by looking for agencies whose pages, listings and consultants all say "fintech" and "London" about real placements. To be in that answer, publish a fintech page naming the roles you fill, placement stories from fintech clients, a London salary guide for those roles and consultant profiles that say they cover the market, and make your APSCo tags match. ### How do I check if a recruitment agency is REC or APSCo accredited? Search the trade body's own directory rather than trusting a logo. REC runs a member directory, and its corporate members pass a compliance assessment every two years. APSCo's "Find an APSCo Member" can be filtered by sector, specialism and region. For executive search, AESC publishes its member firms. If an agency shows a logo but is not in the directory, ask why. ### Is it worth using a recruiter now that AI tools can screen CVs? In our view, yes, for roles where the best people are not applying. Screening tools sort applications that have already arrived; a good recruiter finds people who have not applied, judges fit and persuades them to move. Agencies should answer this on their own site: what AI tools do in their process and what they do not trust them with. ### Will publishing salary data upset our clients? Rarely, if you publish ranges based on enough placements that nobody can be identified. Most clients welcome a credible benchmark, because it helps them set pay before they brief you. Say where the data comes from, how many placements sit behind it and when it was last updated. ### Should our consultants post on LinkedIn, or should the agency? Both, but consultants matter more. LinkedIn appears in about 11% of AI answers, articles make up 50 to 66% of LinkedIn citations, and many cited posts had only 15 to 25 reactions ([Semrush, March 2026](https://www.semrush.com/blog/linkedin-ai-visibility-study/)). A short monthly article from each consultant on their niche builds a record the agency's pages can link to. The agencies that get named are the ones whose knowledge of a market is written down and confirmed by others. If you want help turning your consultants' market knowledge into salary guides, press comment and a presence employers can find, [Reputation That Wins](/reputation-that-wins) is built for that, and our [page for recruitment firms](/for/recruitment) shows how we work with agencies. ## Sources - [Forrester, B2B buyers make zero-click buying number one (January 2026)](https://www.forrester.com/blogs/b2b_buyers_make_zero_click_buying_number_one/) - [OpenAI, Testing ads in ChatGPT (August 2026)](https://openai.com/index/testing-ads-in-chatgpt/) - [GOV.UK, Plan to Make Work Pay and Employment Rights Act: timeline update (August 2026)](https://www.gov.uk/government/publications/implementing-the-plan-to-make-work-pay-and-employment-rights-act/plan-to-make-work-pay-and-employment-rights-act-timeline-update) - [legislation.gov.uk, Employment Rights Act 2025 (December 2025)](https://www.legislation.gov.uk/ukpga/2025/36/introduction) - [REC, The Fair Work Agency launches (March 2026)](https://www.rec.uk.com/our-view/news/fair-work-agency-launches) - [REC, Code of Professional Practice (September 2026)](https://www.rec.uk.com/recruiters/compliance/code-professional-practice) - [APSCo, Quality and compliance (September 2026)](https://uk.apsco.org/discover-apsco/why-use-an-apsco-member/quality-and-compliance) - [Semrush, LinkedIn AI visibility study (March 2026)](https://www.semrush.com/blog/linkedin-ai-visibility-study/) --- Title: Are guest articles in trade magazines still worth our time? URL: https://fortitudemedia.ai/insights/guest-articles-authority-building-ai Author: Ross Williams Published: 2026-03-31 Last reviewed: 2026-09-24 Topic: Press, reviews and being trusted Short answer: Yes, when an editor at a trade title your buyers read chooses the article on merit. Paid guest posts on link sites are worthless, and Google treats keyword-rich links in them as link spam. A real one gives you a byline, independent evidence of expertise that ChatGPT and other assistants read, and often a new relationship with an editor, even if the author link is nofollow. Pitch short, specific ideas about the editor's readers and mention your firm once at most. Someone has suggested you write for the trade magazine your buyers read. Your first reaction may be doubt. You have heard that "guest posting" is a spam tactic, and perhaps an agency once sold you guest posts on websites nobody had heard of that did nothing. Both are true, because two different activities share the name. Paying to place a thin, link-stuffed article on a site that sells placements is worthless, and Google treats it as link spam. Writing a useful article that an editor at a respected trade title chooses to publish under your name remains one of the best ways for a founder-led firm to build trust, with buyers and with the AI assistants they consult. In the [2025 Edelman and LinkedIn B2B Thought Leadership Impact Report](https://www.edelman.com/expertise/Business-Marketing/2025-b2b-thought-leadership-report), a US survey, 64% of the "hidden" decision makers who shape purchases said they trust thought leadership more than marketing materials when judging a firm's capabilities. And AI assistants lean on other people's words: in [Muck Rack's May 2026 analysis](https://www.globenewswire.com/news-release/2026/05/07/3290268/0/en/generative-pulse-earned-media-consistently-drives-ai-citations-holding-at-84.html) of more than 25 million links cited by ChatGPT, Claude and Gemini, earned media made up 84%. This guide is about the second kind: why it works, where to write, how to get an editor to say yes, and what to do once it is published. ## Key takeaways - A guest article counts when an editor at a respected title chose it on merit; a paid placement on a link site counts for nothing and can break Google's link spam policy. - Write for the trade titles, professional body journals and regional business press your buyers read, whichever sector they sit in. - The article should make one clear argument with specific examples and your own evidence, and mention your firm once at most. - Even when the author link is nofollow, the byline and the text are still there for buyers and AI assistants to read. - Founder LinkedIn articles also get cited by AI assistants, so treat them as a companion to guest articles. - Link every published article from an author page on your site, and ask each title to link your bio to that page. ## Why does a real guest article still carry weight? Because an editor who answers to their readers decided your thinking was worth publishing. Everything else follows from that. It puts your ideas in front of your buyers in a place they already trust, ties a named person at your firm to the subject, and often starts a relationship with the editor that leads to quote requests. It matters for AI assistants too. ChatGPT, Claude, Gemini, Copilot and Perplexity learn from, and search, the same trade titles your buyers read. An article there, explaining a problem in your sector under your founder's name, is independent evidence that your firm knows its subject: your words, published to someone else's standards. For example, the managing director of a facilities management firm might write on how new building safety rules change what clients should ask contractors. When a facilities manager later asks an assistant which contractors understand the rules, there is an independent source connecting that firm to the subject. ## Guest article or paid guest post: how do you tell the difference? Ask who decided to publish it, and why. An editor choosing it for readers is a guest article; a fee or a link swap is a paid placement. | | Editorial guest article | Paid guest post | |---|---|---| | Where it appears | A title with its own journalists, editors and regular readers | A site whose articles are mostly written by outsiders promoting something | | Why it was accepted | It is useful to readers, and it was probably edited | A fee was paid or a link was traded | | Mentions of your firm | Light, usually in the biography | Packed with links, often with keyword-heavy link text | | What it cost you | Time | A placement fee | | How Google sees it | Normal editorial content | Link spam, unless the links are marked as paid | Google's spam policies name "links with optimized anchor text in articles, guest posts, or press releases distributed on other sites" as link spam, along with paying for articles that include links passing ranking credit ([Google Search Central](https://developers.google.com/search/docs/essentials/spam-policies)). Clearly labelled sponsored content in a real title can be a reasonable way to reach an audience, but it is advertising, and buyers and assistants read it as such. Why paid, owned and earned coverage carry different weight is set out in [whether ChatGPT trusts press coverage more than your website or your ads](/insights/earned-owned-paid-authority-signals). ## Do guest post links still help SEO if the magazine makes them nofollow? A little, and less than the article itself. Some trade titles mark contributor links as nofollow or sponsored, which tells Google not to associate the two sites; Google says such links will generally not be followed ([Google Search Central](https://developers.google.com/search/docs/crawling-indexing/qualify-outbound-links)). So the link passes little ranking credit. That is not the main prize. The byline, the biography and the argument stay on a trusted site whatever the link attribute, and buyers and AI assistants read the text. Assistants that search the web start from indexes such as Google's and Bing's, so an article that ranks for its subject can surface in answers either way. For how links matter now, see [whether backlinks still matter for ChatGPT and AI search](/insights/ai-evaluates-quality-referring-domains). ## Which publications should we write for? The titles your buyers read for work, which are often in their sector rather than yours. For most founder-led B2B firms that means a short list of eight to twelve. - **The trade press in your buyers' sector.** If you sell to construction firms, the titles construction directors read. - **Professional body journals.** Institutes and associations publish journals for members, and their readers hold budgets. - **Regional business press.** Useful when your buyers are regional, and usually more open to new contributors. - **Partner publications.** A respected partner's newsletter, where the audiences overlap. ### How do I find which trade magazines in my industry accept guest articles? Ask your five best clients what they read for work, and look at where the competitors ChatGPT names have been published. Then read recent issues of each candidate for opinion or comment sections with outside bylines and a contributor guidelines page. Be wary of any site whose main offer is "write for us" with a price list. As a starting point, these are examples of UK titles to check in the five sectors we work with. Whether each takes contributed pieces, and at what length and notice, changes, so confirm with the title before you pitch. | Your sector | Kinds of title your buyers read | Examples to check | |---|---|---| | Professional services | Your clients' trade press, professional body journals, legal and finance titles | Law Society Gazette, Legal Futures, Accountancy Age, AccountingWEB | | Technology and SaaS | Your buyers' sector titles, IT and technology press | Computer Weekly, UKTN | | Fintech | Financial services and fintech trade press | Finextra, FinTech Futures, AltFi | | Recruitment | HR and people titles your client employers read, recruitment trade press | People Management, HR Magazine, Personnel Today, Recruiter | | Agencies and consultancies | Marketing and management titles, your clients' sector press | Campaign, Marketing Week, Consultancy.uk, Management Today | For length, count the words in three recent contributed pieces and match them. For lead time, ask the editor: monthly print titles plan further ahead than websites. ## How do you get an editor to say yes? Make it easy for them: a short, specific pitch about their readers, from someone who has clearly read the title. ### Step 1: Read the last few issues Note the kinds of contributed pieces, the length, the tone and who writes them, then pitch something that fits. An editor can tell within a line whether you have read their publication. ### Step 2: Find the right person Most titles list their editors on a contact or masthead page. Look for the commissioning editor for opinion or features, not the general inbox. If a journalist on the title has quoted you before, they are the best route in. ### Step 3: Write a short pitch Five short paragraphs at most: a clear subject line with the angle; why this matters to their readers now; what the article will argue and what readers will take away; why you are the right person to write it; and a simple offer of an outline. For example, a pitch from the founder of a payroll bureau might read: "Many care providers opening a second or third site are still running payroll on systems built for one. I'd like to write around 900 words for your operations section on the payroll checks worth doing before expanding, drawn from what we see across the care groups we work with. I have run a payroll bureau for care providers for over a decade. Happy to send an outline if it is of interest." ### Step 4: Follow up once, then deliver If you hear nothing after a week or two, send one polite follow-up, then move on. When a title says yes, hit the deadline and the word count and accept edits gracefully. Editors come back to contributors who make their lives easier, as we cover in [how to get trade journalists to call you for comment](/insights/building-relationships-journalists-editors). ## What should the article say? One clear argument, backed with specifics, with the selling left out. - **Lead with the point.** Open with the insight rather than background. Readers decide in the first few lines whether to carry on, and assistants lean on the top of a page too: a 2026 analysis of 1.2 million ChatGPT answers by Kevin Indig found 44% of citations came from the first 30% of a page ([Search Engine Land](https://searchengineland.com/chatgpt-citations-content-study-469483)). - **Be specific.** Name the regulation, the process, the kind of client and the typical mistake. - **Use your own evidence.** A figure from your own work, anonymised, or a finding from a small study, makes an article more credible and more quotable. In the [research paper that first tested this](https://arxiv.org/abs/2311.09735) (Princeton and IIT Delhi, 2024), adding quotations, statistics and cited sources each raised a source's visibility in AI-generated answers by roughly 30% to 40% (a research engine test, so directional). [How to turn your own data into trade press coverage](/insights/leverage-data-research-press-coverage) shows how to produce that evidence. - **Structure it clearly.** Where the format allows, use descriptive subheadings, short paragraphs and the main idea at the start of each section, as we set out in [what an article that ChatGPT actually quotes looks like](/insights/anatomy-of-article-cited-by-ai). - **Mention your firm once, if at all.** Most editors allow a link in the biography and perhaps one passing mention. The quality of the article is the advertisement. ## How much time and money does a guest article take? Mostly the founder's time: agreeing the idea, sharing the substance and editing the draft until every sentence is theirs. Who does the rest depends on who you have. | Who writes it | The founder's part | Works well when | Watch for | |---|---|---|---| | The founder alone | Everything | The founder writes easily and has time | Articles that slip when the diary fills | | An in-house marketer | Idea, interview, final edit | The marketer knows the subject and the buyers | Drafts that sound like marketing | | A freelance ghostwriter | Idea, interview, final edit | The writer knows your sector | Generic copy written without that interview | | A PR or content agency | Idea, interview, final edit | You also want pitching and editor relationships | Paid placements sold as "guest articles" | ## What should we do once it is published? Publication is the start of the article's useful life. Make it easy for buyers and assistants to connect the byline to your firm. - **Keep an author page for each person who writes.** List their role, qualifications and every article they have published elsewhere, each with the title's name, the date and a link. - **Mark the author page up.** Add Person structured data, with sameAs links to the author's LinkedIn profile and any contributor profile pages the titles keep; Google's author markup guidance recommends url or sameAs links to a page that identifies the author ([Google Search Central](https://developers.google.com/search/docs/appearance/structured-data/article)). Markup helps search engines read facts correctly but is no ranking trick, as [whether you need schema markup for AI assistants](/insights/structured-data-schema-markup-ai) explains. - **Ask the title to link the author bio to that page.** A bio link to a page about the person is more useful to readers than one to your home page, and easier for an editor to justify. - **Share it and use it.** Post it on LinkedIn with a sentence on why it matters, give it to your sales team as a follow-up, and thank the editor with another idea. ## Are LinkedIn articles worth as much as guest articles? Not as independent endorsement, because no editor chose them, but they matter more for AI answers than many founders assume. In [Semrush's March 2026 study](https://www.semrush.com/blog/linkedin-ai-visibility-study/), LinkedIn appeared in about 11% of AI answers (14.3% in ChatGPT search, 13.5% in Google AI Mode), articles made up 50% to 66% of LinkedIn citations, and individual members accounted for 59% of LinkedIn citations in ChatGPT search and AI Mode. So a founder's LinkedIn articles are worth writing in their own right, and a strong one can become a pitch. We have no comparable data for Medium, so treat it as your own channel. ## How do you tell whether it is working? Look at conversations and AI answers rather than keyword rankings. Signs include editors asking for more, journalists asking for comment, prospects mentioning your piece, and AI answers that describe your firm in terms matching what you wrote or cite the article. Assistants that search the web can reflect a new article within weeks; answers from a model's own memory change only when it is retrained, which takes months. Several articles over a year, in the two or three titles that matter, are what change how you are seen. ## What to do this month 1. Ask five of your best clients which publications they read for work, and list the ones that run contributed articles. 2. Read the last three issues of your top two titles and note the length, tone and topics of their contributed pieces. 3. Write down three article ideas drawn from questions clients ask you every week. 4. Send one short, specific pitch to the right editor at your first-choice title. 5. Create an author page for whoever will write, with Person markup, ready to list their published articles. 6. Turn one of the three ideas into a LinkedIn article under the founder's name while you wait for a reply. ## Frequently asked questions ### How often should we pitch the same publication? Only when you have a new idea that suits it. For most trade titles, a piece every few months is plenty. The better route to regular appearances is to become a reliable source, so the editor asks you. ### Should the article be about our firm or about the industry? About a problem your readers have. Your firm's experience should be the evidence rather than the subject. An article on the mistakes firms make when outsourcing payroll, drawing on what you have seen, will be accepted and read; one on why your payroll service is excellent will be rejected. ### Can someone else write it for the founder? Yes, if the ideas and expertise are the founder's. Many business leaders work with a writer who interviews them and drafts the piece for them to edit. A generic article written without that input does not work, because editors and readers can tell. The byline is a promise that the named person stands behind every sentence. ### How long should a guest article for a trade magazine be? As long as that title's contributed pieces usually are. There is no single right length, so match the recent examples you counted or the figure the editor gives you, and put the main argument in the first few lines. ### How much does a ghostwriter charge for a trade magazine article in the UK? Rates vary widely with the writer's experience, sector knowledge and whether they also pitch and place the piece, and most charge per article or by the month. Ask what is included and to see pieces published in titles like yours. Our own prices are on [what it costs](/what-it-costs). ### Can I republish my guest article on my own website? Ask the title first, because some allow it after a period and some do not. Duplicate text is not a spam violation in itself: Google calls some duplication normal and picks one version to show ([Google Search Central](https://developers.google.com/search/docs/crawling-indexing/canonicalization)), which may not be yours. We usually suggest a shorter adapted version that adds something new, says where the article first appeared and links to it. If you want help finding the right titles, shaping the ideas and getting pieces placed under your founder's name, that is a core part of [Reputation That Wins](/reputation-that-wins). A free [Growth Review](/growth-review) will tell you where to start. ## Sources - [Edelman and LinkedIn, 2025 B2B Thought Leadership Impact Report (June 2025)](https://www.edelman.com/expertise/Business-Marketing/2025-b2b-thought-leadership-report) - [Muck Rack, Generative Pulse: earned media consistently drives AI citations (May 2026)](https://www.globenewswire.com/news-release/2026/05/07/3290268/0/en/generative-pulse-earned-media-consistently-drives-ai-citations-holding-at-84.html) - [Google Search Central, Spam policies for Google web search (August 2026)](https://developers.google.com/search/docs/essentials/spam-policies) - [Google Search Central, Qualify your outbound links to Google (December 2025)](https://developers.google.com/search/docs/crawling-indexing/qualify-outbound-links) - [Search Engine Land, ChatGPT citations content study (February 2026)](https://searchengineland.com/chatgpt-citations-content-study-469483) - [Aggarwal and others, Princeton and IIT Delhi, research paper presented at KDD 2024 (June 2024)](https://arxiv.org/abs/2311.09735) - [Semrush, LinkedIn AI visibility study (March 2026)](https://www.semrush.com/blog/linkedin-ai-visibility-study/) - [Google Search Central, Article structured data (September 2026)](https://developers.google.com/search/docs/appearance/structured-data/article) --- Title: Can we just do content, or do we need PR and a better website too? URL: https://fortitudemedia.ai/insights/compound-returns-content-pr-ai-together Author: David Adams Published: 2026-03-31 Last reviewed: 2026-09-24 Topic: Costs, returns and the business case Short answer: Usually you need all three, because each makes the others work harder. Content gives PR something to say, coverage tells AI assistants that others vouch for you, and a clear website lets assistants read you and turns visitors into enquiries. Returns build over time: in our worked example, cumulative gross profit passes spend only in year three. If budget is tight, fix the website and core pages first, then add small, regular coverage within six months. You have probably tried each of these before, one at a time. A content agency wrote two blog posts a month for a year. The posts were fine, and nothing happened. A PR firm got you into a couple of trade titles, which looked good on LinkedIn, and nothing happened. Then the website was rebuilt, it looked much better, and the phone still did not ring any more than before. Now buyers are asking ChatGPT, Gemini, Copilot, Claude and Perplexity who to hire, and you need to decide where to put the money. The instinct is to pick one thing and test it properly. That is usually a mistake, and the reason explains why the earlier attempts disappointed. AI assistants name a firm when they can find a clear, consistent and well-supported story about it in several places, and most of what they cite is written by someone other than the firm. In [Muck Rack's May 2026 study](https://www.globenewswire.com/news-release/2026/05/07/3290268/0/en/generative-pulse-earned-media-consistently-drives-ai-citations-holding-at-84.html) of more than 25 million links cited by ChatGPT, Claude and Gemini, 84% came from earned media and just 0.3% from paid content. Your own content, other people's coverage and your website each supply a different part of that story. This guide explains how the three work together, what the returns look like over three years, and what to fund first. ## Key takeaways - Content, PR and your website each do a different job, and AI assistants need all three before they name a firm with confidence. - Most of what AI assistants cite is earned coverage, so useful content needs other people vouching for it before it carries full weight. - A clear website lets assistants read and quote you accurately, and turns the buyers they send into enquiries. - In a worked example, cumulative gross profit passes cumulative spend only in year three, which is why judging each strand alone at month six looks like failure. - A founder's LinkedIn adds a steady voice that assistants do cite, but it is the firm speaking about itself and does not replace independent coverage. - If budget is tight, fix the website and core pages first, then add small but regular coverage within six months. ## What does each of the three actually do? Each answers a different question an assistant needs answered before it names you. Content answers "what do they know?" Coverage answers "who else says so?" The website answers "what exactly do they do, and for whom?" ### Content: something worth quoting Pages and articles that answer the questions your buyers ask, with your own knowledge behind them. When a buyer asks ChatGPT how to handle a problem, the assistant looks for a clear, specific answer it can use. We describe what those pages look like in [what an article that ChatGPT quotes looks like](/insights/anatomy-of-article-cited-by-ai). ### PR and coverage: someone else vouching for you Your name and your experts' views in places you do not control: trade press, directories, association listings, awards, reviews and speaking programmes. Reviews count here too: in [G2's April 2026 survey of software buyers](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html), 45% said citations of review sites were the most confidence-inspiring signal in an AI answer. Coverage turns "this firm claims expertise" into "this firm is known for it". ### The website: the place everything points to Where assistants check the facts and buyers land when they want to know more. It needs to say plainly what you do, who for and where, in a way both people and AI crawlers (the programs that read websites for AI assistants) can follow. A confusing or thin website undermines every piece of content and coverage that points to it. ## Which sources do AI assistants actually cite? A mix of all three, with independent sources carrying most weight. We know of no published UK study of B2B citations, so the figures below come from large US and global studies, and the patterns shift as the assistants change. | Type of source | Which strand it belongs to | What the studies found | |---|---|---| | Press, trade titles and other people's sites | PR and coverage | Muck Rack (May 2026): 84% of links cited by ChatGPT, Claude and Gemini were earned media, and 27% were journalism | | Review, analyst and community sites | PR and reputation | [Profound (2025, 680 million citations)](https://www.tryprofound.com/blog/ai-platform-citation-patterns): G2 and Gartner appeared among B2B sources; Perplexity leaned heavily on Reddit | | LinkedIn articles and posts | Founder content | [Semrush (March 2026)](https://www.semrush.com/blog/linkedin-ai-visibility-study/): LinkedIn appeared in about 11% of AI answers | | Your own pages | Website and content | [SE Ranking (November 2025)](https://www.searchenginejournal.com/new-data-top-factors-influencing-chatgpt-citations/561954/): the number of other websites linking to a site (referring domains) was the strongest predictor of ChatGPT citations | The last row makes the point: even your own pages get cited more when other sites link to you. The three strands feed each other. ## Why does one on its own usually disappoint? Each, done alone, hits a ceiling that only the other two can lift. - **Content without coverage.** Good articles on a site nobody references are easy to overlook. The assistant may read them but has no independent sign that they come from a firm worth naming. - **Coverage without content.** A quote about industry trends or a new-hire announcement gives assistants little to connect you with, and a buyer who follows the link lands on a general homepage rather than a page that proves the point. - **A website without either.** A handsome site with nothing new to quote and nobody pointing to it is a clear shop window on a quiet street. This is why so many founders conclude that "content does not work" or "PR does not work". Each was tested alone, against a standard only the combination could meet. We explain the same pattern from the PR side in [what a year of consistent PR looks like](/insights/compounding-effect-consistent-pr-12-months). ## How do the three feed each other? Each piece of work gives the next one a head start. For example, imagine a managed IT firm whose prospects keep asking what IT checks to run before acquiring a business. It writes a clear guide based on what its engineers find in acquisitions, with a checklist. - **The content gives PR something to say.** The firm pitches a short piece on IT risks in acquisitions to a title read by finance directors. Because the guide exists, the pitch has substance and the editor has something to link to. - **The coverage vouches for it.** The article names the firm and its technical director as the source, so the expertise now appears in two places, one of them independent. - **The website makes it easy to act on.** The guide sits on a clear page, linked from the service pages, with the director's profile and a way to book a call. - **The result feeds the next round.** A buyer asks ChatGPT about IT due diligence and the firm is named. The enquiry becomes a client, the client becomes a case study, and the case study becomes the next article and the next pitch. Each loop makes the next easier. ## What does compounding look like in numbers? The returns build because the pages and coverage keep working after you have paid for them, and the clients they bring stay for years. The table is an example with illustrative figures, to show the mechanism, not a forecast. Example: you spend £40,000 a year on the three together. An average client pays £15,000 a year in fees for three years, at a 50% gross margin. As the body of work grows, it brings in two new clients in year one, four in year two and six in year three. | Year (example) | New clients | Clients paying | Fees | Gross profit | Cumulative gross profit | Cumulative spend | |---|---|---|---|---|---|---| | Year 1 | 2 | 2 | £30,000 | £15,000 | £15,000 | £40,000 | | Year 2 | 4 | 6 | £90,000 | £45,000 | £60,000 | £80,000 | | Year 3 | 6 | 12 | £180,000 | £90,000 | £150,000 | £120,000 | Three things stand out. - **Years one and two look like a loss.** Judged on gross profit, the work is still behind at month 24. Judged at month twelve, many boards would stop. - **Cumulative profit passes cumulative spend only in year three.** Spend stays flat while profit climbs, because earlier pages, coverage and clients are still working. - **The growth in new clients is the assumption to test hardest.** If the work brought in half as many (one, two and three), cumulative gross profit would reach £75,000 by year three, short of the £120,000 spent, so payback would depend on those clients staying and the work continuing to bring new ones. Paid advertising works the other way: results stop when spend stops, as we compare in [whether being named by ChatGPT is cheaper than Google Ads](/insights/economics-ai-visibility-vs-advertising). To run the sum with your own figures for a board, see [how to get board approval for AI search](/insights/business-case-ai-optimisation-board). ## Can the founder's LinkedIn replace PR? No, but it is one of the cheapest ways to add a steady voice, and assistants do read it. [Semrush's March 2026 study](https://www.semrush.com/blog/linkedin-ai-visibility-study/) found LinkedIn in about 11% of AI answers, rising to 14.3% in ChatGPT search and 13.5% in Google's AI Mode. Articles made up 50 to 66% of the LinkedIn citations, and the typical cited post had only 15 to 25 reactions, so a useful post does not need a large following. The difference is who is speaking. A founder's post is the firm talking about itself, like your website. Coverage is someone else choosing to quote you, which is the kind of source that dominated the Muck Rack figures. Use LinkedIn to publish the thinking, then turn the best posts into pitches, guest articles and pages on your site. We run founder LinkedIn as part of [Reputation That Wins](/reputation-that-wins). ## How do we plan the three together? As one piece of work with a shared list of buyer questions, not three suppliers with three briefs. ### Step 1: Agree the buyer questions you want to own Pick 20 to 50 questions your best prospects ask, in their words. Every page, pitch and profile over the next year should help answer one of them. Track the same list every month, each question asked several times in a fresh chat, and count a win only when it repeats. ### Step 2: Fix the website foundations first In the first quarter, make sure the site says clearly what you do, who for and where, that AI crawlers can read it, and that your core pages answer the most important buyer questions. It need not take long. One example is arum, which advises banks, governments and major brands on collections technology: - "310 pages rebuilt across three regions, in weeks." - "Proper structure and markup on every page, a plain summary AI assistants can read, facts stored once so two pages never disagree, and every old address redirected so years of rankings carry over." The detail is in the [arum case study](/work/arum), and [what a B2B website should include](/insights/what-should-a-b2b-website-include) covers the basics. ### Step 3: Add coverage built around those pages From the second quarter, pitch commentary and articles on the topics your new pages cover, and correct directory and association listings so they describe you the same way your website does. ### Step 4: Settle into a monthly rhythm For most founder-led firms, a sustainable rhythm is one useful page, one piece of coverage and one piece of proof, such as a case study or review, each month. Consistency matters more than volume. ### Step 5: Feed results back in Every new client, case study and good question from a prospect is raw material for the next page and pitch. Put a monthly half hour in the diary to capture them. ## How do we judge it without stopping too early? Judge the combination, not the parts. Measure whether you are named more often for your buyer questions, whether the assistants describe you accurately, and whether enquiries mention an AI assistant. Set review points as targets, not promises. By month three, the foundations should be done and a baseline recorded. By month six, the target is to be named for at least some of your questions: assistants that search the web live can reflect a new page or article within weeks, while answers drawn from a model's memory change only when it is retrained, which takes months. By month twelve, aim to count enquiries that came through AI answers. Our guide to [building an AI visibility dashboard](/insights/build-ai-visibility-dashboard) sets out a simple format. "Did the PR produce leads?" is the wrong question when the PR's job was to make the content credible. Ask whether the whole effort is moving you into the answers your buyers see. ## What if we can only afford part of it? Keep all three in play at a smaller scale rather than dropping one. A small monthly budget should buy: - **Website:** the core pages fixed first, then one page a month answering a buyer question, with facts kept consistent. - **Coverage:** one good piece of trade commentary a month, built around a page you have already published, plus a few new reviews. - **Listings:** correcting directories and profiles yourself, which your team can do in a few afternoons. - **Tracking:** a monthly check of what the assistants say, by hand or with a low-cost tool. We package the website and reputation work as [Website That Sells](/website-that-sells) and [Reputation That Wins](/reputation-that-wins), and our prices are on [what it costs](/what-it-costs). What does not work is a cheap package that promises all three and delivers a few generic posts, as we explain in [whether a cheap AI search package will work](/insights/cheap-ai-optimisation-doesnt-work). ## What to do this month 1. List the 20 to 50 buyer questions you want to be named for, with your sales team's help. 2. Check your website against the top five: is there a clear page answering each? 3. Ask those five in ChatGPT, Gemini, Copilot, Claude and Perplexity, and note who is named and which sources are cited. 4. Choose one question where you have strong expertise and plan a page, a pitch and a piece of proof around it. 5. Agree your month three, six and twelve review points with your leadership team before you spend anything. ## Frequently asked questions ### Which of the three should we start with? The website and your core pages, because everything else points to them. If your site does not say clearly what you do, coverage and content elsewhere send people and assistants to a page that does not convince them. Once the foundations are sound, usually within the first quarter, add coverage. ### Is PR or SEO better for getting mentioned by ChatGPT? You need both, because they do different jobs. SEO makes your pages easy to find and read, and ranking helps, though it does not guarantee a mention. PR earns the independent references that assistants lean on, and those references also make your own pages more likely to be cited. ### Do press releases help with AI search? Rarely on their own. A release on your site or a wire service is still your own words. It helps when it gives a journalist something to write about, because the resulting article is independent coverage. Put the effort into a story or data worth covering rather than into distribution. ### How much does B2B PR cost per month? It depends on scope: a few hours of founder commentary a month is a different job from a full programme of pitching, articles and awards. Compare what each proposal buys: named titles, pitches a month, founder time needed and how results are reported. Our own prices are on [what it costs](/what-it-costs). ### Can we use different suppliers for each part? You can, but someone must own the joined-up plan. The common failure is three suppliers working to three briefs, each reporting its own activity. Give them the same list of buyer questions, share the calendar and hold one monthly review where all three are discussed together. ### Does this apply to firms that win most work through referrals? Yes. In our experience, referred buyers often check the name they were given with an AI assistant or a quick search before they call. If the assistant describes you vaguely, or suggests alternatives, some referrals leak away. The three strands make sure that check confirms the recommendation. ### What if we already have strong PR but a weak website? Then the website is probably where your coverage leaks, because people who follow your press mentions land on pages that do not answer their questions. Building pages around the topics your coverage already addresses is often the quickest win for a firm in that position. If you want to know which of the three is holding you back, [book a free Growth Review](/growth-review): 30 minutes with David Adams, and three fixes in writing afterwards. ## Sources - [Muck Rack, What is AI reading? Earned media drives 84% of AI citations (May 2026)](https://www.globenewswire.com/news-release/2026/05/07/3290268/0/en/generative-pulse-earned-media-consistently-drives-ai-citations-holding-at-84.html) - [G2, New G2 research: half of B2B software buyers now start their research with AI chatbots (April 2026)](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html) - [Profound, AI platform citation patterns (2025)](https://www.tryprofound.com/blog/ai-platform-citation-patterns) - [Semrush, LinkedIn AI visibility study (March 2026)](https://www.semrush.com/blog/linkedin-ai-visibility-study/) - [Search Engine Journal, New data: top factors influencing ChatGPT citations (November 2025)](https://www.searchenginejournal.com/new-data-top-factors-influencing-chatgpt-citations/561954/) --- Title: Do awards and accreditations make ChatGPT more likely to recommend us? URL: https://fortitudemedia.ai/insights/awards-accreditations-ai-trust Author: Ross Williams Published: 2026-03-31 Last reviewed: 2026-09-24 Topic: Press, reviews and being trusted Short answer: Yes, if someone independent checked them: registration with your regulator or professional body, audited certifications such as ISO 27001 or Cyber Essentials Plus, respected rankings and trade awards with named judges. Paid or vanity awards add almost nothing. State each credential in plain text on your site, with the awarding body and date, mark it up for search engines, and make sure the awarding body and trade press say the same thing. The email arrives most weeks. Your firm has been "selected" for a regional business award, and the winners' package, with a trophy, a logo and a table at the dinner, is available for a fee. Meanwhile your operations director is months into ISO 27001, and your marketing manager wants to enter the trade magazine's awards. Which of these do buyers, and the AI assistants they consult, take notice of? ChatGPT, Gemini, Claude, Copilot and Perplexity treat recognition the way a sceptical buyer would. They weigh who gave it, how hard it was to get and who else talks about it. Buyers look for the same independent evidence inside AI answers: in [G2's April 2026 survey of 1,076 software buyers](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html), 45% said citations of review sites were the most confidence-inspiring signal in an AI answer. An audited certification or an entry on your professional body's register is a fact an assistant can check and repeat. An award judged by a respected trade title is evidence that people who know the sector rate you. A badge you paid for is a claim you made about yourself. This guide sorts the kinds of recognition by how much they help, shows how to tell a real award from a paid one, and explains how to make the recognition you already have visible to buyers and assistants alike. ## Key takeaways - AI assistants weigh awards and accreditations by who granted them, how selective they were and whether independent sources repeat them. - Registration with your regulator or professional body, and audited certifications such as ISO 27001 or Cyber Essentials, are the most useful recognition for a B2B firm. - An award judged by a respected trade title or industry body is worth entering; an award you can buy is worth ignoring. - A badge image tells an assistant almost nothing, so state every award and certification in plain text, with the awarding body and the date. - Recognition becomes visible to assistants when the awarding body, the trade press and your own site all say the same thing. - Let buyer demand decide which certifications to pursue: the ones your tenders and questionnaires keep asking for come first. ## How do AI assistants treat awards and accreditations? As claims, each weighed by its source. No assistant publishes rules for this, but the pattern follows from how they work. When a buyer asks for "ISO 27001 certified IT support firms in Bristol" or "the best employment law firms for retailers", the assistant looks for evidence that a firm fits. Some of it is on the firm's own website. More convincing evidence sits elsewhere: the professional body's register, the trade magazine's list of winners, a directory's rankings, a client's case study. When the firm's claim and an independent source agree, the assistant can state the fact with confidence; when the only source is the firm, it may give the claim less weight or leave it out. A certification on a public register, in a trade article and on your own site is well evidenced. A trophy that appears only as a logo on your homepage is barely evidence at all. This is the cross-checking assistants apply to everything they say about a firm, explained in [why ChatGPT sounds unsure about your firm](/insights/ai-cross-references-business-sources). It also maps to what Google calls trustworthiness, covered in [what E-E-A-T is and why AI assistants care about it](/insights/eeat-why-ai-cares-more). ## Which kinds of recognition count most? For a UK B2B firm, roughly in this order, from most useful to least. | Kind of recognition | Who checks it | Examples | Weight with buyers and assistants | |---|---|---|---| | Regulation and professional body registration | Regulator or chartered body | SRA, ICAEW, ACCA, RICS, FCA register | Highest: the reference record of what you are | | Audited certifications | Accredited certification body | ISO 9001, ISO 27001, ISO 14001, Cyber Essentials Plus | High: specific, verified and renewed | | Independent rankings and directories | Researchers or verified reviewers | Legal 500, Chambers, G2, Clutch | High where your sector has one | | Verified standards on people and impact | Independent assessors | B Corp, Investors in People | Medium: strong when the buyer asks about them | | Judged trade and industry awards | Named judges | Your buyers' trade magazine awards | Medium: evidence that the sector rates you | | Pay-to-win and vanity awards | Nobody independent | "You have been selected" schemes | Little or none, and can raise doubts | ### Regulation and professional body registration If your field is regulated or has a chartered body, your entry on its register is the foundation, and the first thing a careful buyer checks. Make sure the details match your website exactly. ### Audited certifications ISO 9001 for quality, ISO 27001 for information security, ISO 14001 for environmental management, and Cyber Essentials or Cyber Essentials Plus all involve an outside assessment and have to be renewed. Certificates come from certification bodies rather than from [ISO](https://www.iso.org/standard/27001) itself, and ISO notes that a certificate from an accredited body adds confidence because someone has independently checked that body's competence. Procurement teams ask for these by name, and buyers put them into their questions to AI assistants too. ### Independent rankings and directories Some sectors have rankings compiled by researchers who interview clients and peers; others have review platforms that verify reviewers. Where your sector has one, a listing there is among the most quoted forms of recognition, because journalists, buyers and assistants treat it as a reference. | Your sector | Rankings and directories to consider | What we know about AI citations | |---|---|---| | Professional services (law) | Legal 500, Chambers and Partners | Treated as references by buyers and journalists; we have seen no UK citation study | | Technology and SaaS | G2, Capterra, Gartner Peer Insights | G2 made up 1.1% of all ChatGPT citations, and Gartner 7.0% of Perplexity's top ten sources, in [Profound's 2025 study](https://www.tryprofound.com/blog/ai-platform-citation-patterns) | | Agencies and consultancies | Clutch, sector award lists | Review sites such as Clutch appear in B2B answers, in our experience | | Recruitment | REC or APSCo membership, Google reviews | We have seen no citation study specific to recruitment | | Fintech | FCA register, G2 for software products | As for SaaS where you sell software; the register for regulated activity | Reviews on Google, Trustpilot, G2 or Clutch are recognition of a different kind, from clients rather than judges; which sources ChatGPT cites for B2B questions is covered in [which sources ChatGPT cites when a B2B buyer asks who to hire](/insights/ai-optimisation-b2b-vs-b2c). ### Awards judged by trade titles and industry bodies Awards run by the trade magazine your buyers read, or your industry association, with published judges and criteria. Winning, or being shortlisted, shows that people who know the sector rate your work, and the coverage is often worth more than the trophy. ## Do B Corp, Investors in People and similar badges count? Yes, when they are independently assessed, and both of these are. They belong on your credentials page, but they answer a different question from ISO or a regulator's register: how you treat people, communities and the environment, not whether you can be trusted with the work. B Lab says [B Corp certification](https://www.bcorporation.net/en-us/certification/) is "independently verified and audited by a third party, based on ISO 17021-1 requirements", starting from a self-assessment against its standards. [Investors in People](https://www.investorsinpeople.com/) accreditation is assessed by a practitioner, drawing on a survey and direct conversations with a cross-section of staff, and is awarded at Standard, Silver, Gold or Platinum. Does B Corp make ChatGPT more likely to recommend you? Only when the buyer's question involves it, for example "B Corp certified accountants in Bristol", or when a tender scores social value. Then it works as a filter, like ISO 27001. Investors in People matters most when candidates ask what you are like to work for, which is covered in [what ChatGPT tells candidates about working at your firm](/insights/ai-search-recruitment-talent). Charters and pledges you sign without assessment are statements of intent, and should be described that way. ## How do you tell a real award from one you are buying? A real award has independent judges, published criteria, a real chance of losing and an organiser your buyers respect. A bought award has none of those. Run any invitation through these questions. - Who runs it: a trade title or industry body your buyers know, or a company whose main business seems to be running awards? - Who judges it: named, credible people from the sector, or nobody you can identify? - Did you enter, or were you told you had been selected? - Do some entrants lose, or are the shortlists huge and the categories endless? - What does it cost? A modest entry fee and a dinner ticket are normal. A fee to receive the award, use the logo or appear in the winners' list is not. - Would your best client recognise the name? If the answers are uncomfortable, decline politely and spend the money on an award your buyers' trade press runs, or the certification your prospects keep asking about. ## Why do accreditations often matter more than awards? Accreditations are verified, specific and renewed, so they answer a buyer's real question: can this firm be trusted with our work? Awards answer a softer one: did some judges like this firm's entry? The difference shows in how buyers ask AI assistants for help. A procurement lead asks "Which payroll bureaus in the Midlands are ISO 27001 certified?" or "Which managed IT providers near Manchester have Cyber Essentials Plus?". Those are filters. If your certification is not stated clearly in text on your site, and recorded independently wherever the scheme allows, you drop out of the answer before quality is considered. ## How do you make your recognition visible to buyers and assistants? Write it out in text, give it a proper home on your site, label it for machines, and make sure independent sources say the same thing. ### Step 1: Write it in text, not only as a logo A badge image tells an AI crawler, the software assistants use to read websites, almost nothing. State each certification and award in words: its full name, who granted it, the year and, for certifications, the scope and the certifying body. Give logos descriptive alt text as well. ### Step 2: Give it a proper home on your site An about page or a dedicated credentials page listing every current certification, registration and award, with dates, is the page buyers and assistants will use. Mention the relevant ones on service pages too: ISO 27001 belongs on the pages about work where data security matters. ### Step 3: Label it for machines Structured data is a small block of code that labels facts on a page so search engines can read them reliably. Many small-firm sites carry none at all; if yours has Organization markup, it can list awards, memberships and credentials using the schema.org properties award, memberOf and [hasCredential](https://schema.org/hasCredential), which schema.org defines as "a credential awarded to the Person or Organization". A simplified example with made-up values: ```text Organization name: Example Payroll Ltd award: Payroll Bureau of the Year 2025, Example Trade Awards memberOf: Example Institute of Payroll Professionals hasCredential: ISO/IEC 27001:2022 certificate, Example Certification Ltd, issued March 2026 hasCredential: Cyber Essentials Plus, certified June 2026 ``` Keep the markup matching the visible text. Google says its AI features need no special markup, so treat this as a way to state facts clearly rather than a ranking trick, as [whether you need schema markup for AI assistants](/insights/structured-data-schema-markup-ai) explains. ### Step 4: Make sure independent sources say the same Check that the awarding body lists you correctly, that register entries match your trading name and address, and that coverage of a win names you accurately. ### Step 5: Get coverage of the result A shortlisting or a win is a reason to contact the trade journalist who covers your area and to tell your clients. Coverage turns a line on your own site into a mention on someone else's, as [whether ChatGPT trusts press coverage more than your website](/insights/earned-owned-paid-authority-signals) explains. ### Step 6: Keep it current Remove lapsed certifications and old awards that suggest you have stood still. Cyber Essentials and Cyber Essentials Plus certificates expire after 12 months ([IASME](https://iasme.co.uk/cyber-essentials/frequently-asked-questions/)), so an out-of-date year on your site tells a careful buyer the certificate may have lapsed. ## Does Cyber Essentials help us win contracts? Often, yes. UK central government made Cyber Essentials mandatory in 2014 for contracts advertised after 1 October that year involving personal information and certain ICT products and services ([GOV.UK, Procurement Policy Note 09/14](https://www.gov.uk/government/publications/procurement-policy-note-0914-cyber-essentials-scheme-certification)), and the [NCSC](https://www.ncsc.gov.uk/cyberessentials/overview) says a growing number of organisations require suppliers to be certified to bid for work. The basic level combines a self-assessment with independent checking; Cyber Essentials Plus adds hands-on technical testing, so it carries more weight with buyers who know the difference. For AI answers it works like any filter: state the level, the date and the certifying body in text, and keep it renewed. ## How do you choose which awards to enter? Enter fewer, and choose them the way a buyer would, rather than whoever sent the most persuasive email. 1. **List what your buyers' procurement asks for.** Any certification that appears repeatedly in your last ten tenders comes ahead of any award. 2. **List the awards your buyers' trade press and industry bodies run.** In most sectors only two or three matter. 3. **Check each one against the questions above,** and be realistic about where you could be shortlisted this year. 4. **Write entries with evidence.** Judges reward specific results and client testimony, and a strong entry gives you material for case studies. 5. **Plan what happens after the result.** A shortlisting announced well is worth more than a win nobody hears about. ## What to do this month 1. Pull the last ten tenders or supplier questionnaires you answered and list the certifications they asked for. 2. Check your entries on your regulator's or professional body's register and correct anything that differs from your website. 3. Create or update a credentials page that states every current certification, registration and award in plain text, with dates and awarding bodies. 4. Ask your web developer to add award, memberOf and hasCredential to your Organization markup, matching that page. 5. Stop responding to vanity award schemes and remove any paid-for badges from your site. 6. Put the entry deadline for the one award your buyers' trade press runs, and that you could realistically win, in the diary. ## Frequently asked questions ### Are paid business awards a scam, and how do I spot one? Not always a scam, but often a purchase dressed as recognition. The signs are an invitation saying you have been "selected" before you entered, no named judges, endless categories, and a fee to receive the award or use the logo. Such awards add little for buyers or assistants, and a row of them can make a serious buyer wonder why you chose to display them. ### Should we show award logos on our website or write them out in text? Both, but the text does the work. A logo is recognised at a glance by people who know the scheme; the written line (the full name, the awarding body, the category and the year) is what search engines and AI assistants can read and repeat. Give every logo descriptive alt text as well. ### How much does ISO 27001 certification cost a small UK firm? The largest cost is usually your own team's time to put the controls, policies and records in place, which in our experience takes months rather than weeks. On top of that come the certification body's audit fees, which depend on your size and scope, and any consultant you use. Get quotes from two or three accredited certification bodies. If buyers mainly want reassurance on basic cyber security, Cyber Essentials may answer the question with far less effort. ### Is it worth mentioning that we were shortlisted but did not win? Yes, if the award is credible. Being shortlisted for a respected trade award is a real achievement, and it is fair to say so. State it precisely, naming the award, the category, the organiser and the year; what damages trust is describing a shortlisting as a win. ### How quickly will a new award or certification show up in ChatGPT's answers? It depends on the route. Assistants that search the web live, such as ChatGPT search, Perplexity, Copilot and Google's AI features, can reflect a new page or article within weeks; answers from a model's own memory change only when it is retrained, which takes months. Our dated example is [ID-Pal](/work/id-pal): "UK buyer questions where AI names ID-Pal: none in May, two in August, four in September 2026." Coverage on other sites helps both routes, which is why the announcement matters as much as the award. If you want a clear view of which recognition would move your buyers, and help getting it seen once you have it, that is part of [Reputation That Wins](/reputation-that-wins). A free [Growth Review](/growth-review) is a quick way to see where you stand. ## Sources - [G2, New G2 research: half of B2B software buyers now start their research with AI chatbots (April 2026)](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html) - [Profound, AI platform citation patterns (June 2025)](https://www.tryprofound.com/blog/ai-platform-citation-patterns) - [ISO, ISO/IEC 27001:2022 Information security management systems (October 2022)](https://www.iso.org/standard/27001) - [NCSC, Cyber Essentials overview (September 2026)](https://www.ncsc.gov.uk/cyberessentials/overview) - [IASME, Cyber Essentials frequently asked questions (September 2026)](https://iasme.co.uk/cyber-essentials/frequently-asked-questions/) - [GOV.UK, Procurement Policy Note 09/14: Cyber Essentials scheme certification (May 2016)](https://www.gov.uk/government/publications/procurement-policy-note-0914-cyber-essentials-scheme-certification) - [B Lab, About B Corp Certification (September 2026)](https://www.bcorporation.net/en-us/certification/) - [Investors in People, Accreditations (September 2026)](https://www.investorsinpeople.com/) --- Title: Do backlinks still matter for ChatGPT and AI search? URL: https://fortitudemedia.ai/insights/ai-evaluates-quality-referring-domains Author: Ashley Kenerson Published: 2026-03-31 Last reviewed: 2026-09-24 Topic: Press, reviews and being trusted Short answer: Yes, but not the way your SEO report counts them. Links help AI assistants find and trust your pages, and an SE Ranking study found referring domains were the strongest predictor of ChatGPT citations. What other trusted sites say about you matters more: in an Ahrefs study of 75,000 brands, web mentions correlated with AI Overview visibility about three times as strongly as backlinks. Earn mentions in trade titles, professional bodies, review sites and LinkedIn, and stop buying links. Your agency's monthly report says the number of sites linking to you is up again. The graph goes the right way. Then a prospect mentions that they asked ChatGPT for three firms like yours and you were not one of them. Your sales director asks the obvious question: if all those sites link to us, why has nobody noticed? Backlinks still matter for AI search, but the count in your report is the wrong measure. In a [study of 75,000 brands published by Ahrefs in May 2025](https://ahrefs.com/blog/ai-overview-brand-correlation/), how often a brand was mentioned across the web correlated far more strongly with appearing in Google's AI Overviews (0.664) than the number of backlinks did (0.218). Links help assistants find and trust your pages. What other trusted sites say about you, in words, does more. This guide explains where links still fit, which sites that link to you or mention you count, and how to audit what you have in an afternoon, with free tools if you do not pay for an SEO platform. ## Key takeaways - Backlinks still matter: an SE Ranking study found referring domains were the strongest predictor of whether ChatGPT cited a page. - Brand mentions matter more: in Ahrefs' study of 75,000 brands, web mentions correlated with AI Overview visibility about three times as strongly as backlinks. - A link does three jobs: it records an editor's judgement, helps your pages get found when assistants search, and ties a mention to your firm. - The sites that count most are trade titles, professional bodies and respected press, plus the community and review sites assistants cite heavily, such as LinkedIn, Reddit, YouTube, G2 and Clutch. - Directories, press release wires and paid links are weighed lightly, and buying links that pass ranking credit breaks Google's spam policies. - You can audit your links for free with Google Search Console and Bing Webmaster Tools, then compare them with the competitors ChatGPT names instead of you. ## Do backlinks still matter for ChatGPT visibility in 2026? Yes. Links remain one of the ways search engines judge which pages to trust, and most AI assistants start from a search index. An [SE Ranking study reported in November 2025](https://www.searchenginejournal.com/new-data-top-factors-influencing-chatgpt-citations/561954/) found the number of referring domains was the strongest predictor of whether ChatGPT cited a page. Google AI Overviews and AI Mode use Google's index; Copilot uses Bing's, and ChatGPT search has drawn on Bing. So ranking helps, but it does not guarantee a mention. [Ahrefs found in March 2026](https://ahrefs.com/blog/ai-overview-citations-top-10) that only 38% of AI Overview citations came from pages ranking in the top ten, down from 76% in July 2025, because Google now runs several related searches for one question. Links get your pages into the running; covering the follow-up questions and being described well by others gets you chosen. ## Are brand mentions more important than backlinks for AI search? On the evidence so far, yes. The [Ahrefs study of 75,000 brands](https://ahrefs.com/blog/ai-overview-brand-correlation/) measured how strongly different signals correlated with a brand appearing in AI Overviews: | Signal | Correlation with AI Overview visibility | |---|---| | Branded web mentions | 0.664 | | Branded anchor text | 0.527 | | Branded search volume | 0.392 | | Domain Rating | 0.326 | | Referring domains | 0.295 | | Number of backlinks | 0.218 | Two cautions. Correlation is not cause, and the study covered Google's AI Overviews in 2025, not every assistant. But the direction fits how these systems work: they read text. A trade article that describes your firm in a full sentence teaches an assistant what you do, with or without a link. The best coverage gives you both. ## What jobs does a link do for AI search? Three separate ones, and it helps to keep them apart because they lead to different decisions. - **It records someone else's judgement.** An editorial link, one a journalist or editor chose to include, shows that someone with a reputation to protect thought your firm worth pointing readers to. That is the corroboration assistants look for. - **It helps your pages get found in live search.** When an assistant searches the web to answer a buyer's question, well-linked pages are more likely to be among those it reads. A service page with nothing pointing to it may never be fetched. - **It ties the mention to you.** A link makes clear which firm is being discussed, which matters if your name is close to another business's or your trading name differs from your registered one. That last job has limits. A link or a redirect does not explain how two names relate. With ID-Pal, which had bought NorthRow: "The redirect from northrow.com was technically correct, but nothing on either site told Google or the AI engines that NorthRow and ID-Pal were the same company." The fix: "About four hours of structured data work declaring NorthRow part of ID-Pal on both domains." The result we track: "AI answers still recommending NorthRow as a rival: six in May, one in September 2026." ([ID-Pal case study](/work/id-pal)). Links point; words and markup explain. ## What is a referring domain, and why does the count mislead? A referring domain is a separate website that links to yours at least once. If a trade magazine links to you ten times, that is still one referring domain. SEO tools count them because the number is easy to measure and tends to rise. The count treats every site the same. A free directory listing nobody visits adds as much to the total as a feature in the magazine your buyers read. An agency paid to grow the number will go after the cheap ones. The score most agencies quote alongside the count is explained in [what domain authority is and whether AI assistants use it](/insights/what-is-domain-authority-ai). For example, two management consultancies might each have about 200 referring domains. The first has mostly directory listings and a press release copied onto a hundred news-wire sites. The second has fewer links, but they include two trade press features, an entry on its professional body's register and a quote in a regional business paper. On the report they look level. The second is the one an assistant has read about in a context that explains what it does. ## Which sites that link to you or mention you count most? Those with real editors, real readers and a record in your field, plus the community and review sites assistants cite heavily. No assistant publishes a list, but in our experience the order below holds. | Tier | Examples | How assistants treat it | |---|---|---| | 1. National and business press | Financial Times, The Times, BBC | Strong, widely cited, rare for most B2B firms | | 2. Trade titles and professional bodies | Your sector's trade press, regulator and member registers | Usually the most valuable: right audience, right category | | 3. Smaller editorial sites | Regional business press, specialist newsletters, client case studies | Less each, but together they show you are discussed widely | | 4. Community and review platforms | LinkedIn, Reddit, YouTube, G2, Clutch, Trustpilot, Google reviews | Cited heavily in AI answers; what people say there matters | | 5. Directories, wires and link networks | Generic directories, press release wires, paid guest posts | Weighed lightly, with a few curated exceptions | | 6. Harmful sites | Link farms, spam, sites flagged as unsafe | No help, and can suggest bought links | Tier four is the one most link reports miss. In Google's AI Overviews, [Ahrefs found](https://ahrefs.com/blog/most-cited-domains-ai-overviews/) YouTube (22.9%) and Reddit (18.5%) had the largest shares of mentions among the most-cited sites in the US. A [Semrush study in March 2026](https://www.semrush.com/blog/linkedin-ai-visibility-study/) found LinkedIn in about 11% of AI answers. And in [G2's April 2026 survey](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html) of 1,076 software buyers, 45% said citations of review sites were the most confidence-inspiring signal in an AI answer. For a B2B firm, that means your founder's LinkedIn articles, honest client reviews on the platforms your buyers use, and being discussed accurately where your buyers talk. Tier five is not worthless. For "best firm" questions, assistants do cite curated directories, review platforms and ranked lists: a [Peec AI study reported by Search Engine Land](https://searchengineland.com/ai-search-engines-cite-reddit-youtube-and-linkedin-most-study-473138) found review sites such as G2 common in recommendation answers. Keep accurate entries on the handful of curated directories in your sector. The long tail of generic directories adds little. ## Why does an editor's choice count for more than a listing you submitted? Because someone who answers to their readers made a judgement. An editorial link sits inside written content, in a sentence about what you do. A submitted link sits in a list, in the same template as thousands of other firms. It proves you exist, not that anyone thinks well of you. For example, a recruitment agency sends a press release about a new office through a wire service, and it appears word for word on a hundred small news sites. The same week, a trade editor quotes its founder on hiring shortages and links to its salary guide. The hundred copies add a hundred to the count. The trade piece is what tells an assistant what the agency is known for. Treat press releases as a way to reach journalists; the article a journalist writes because of one is what counts. ## Why do the words around a mention matter? Because assistants learn from the sentence, not just the link. - **The topic of the page.** A piece on cyber security for law firms that mentions a managed IT provider tells an assistant which category that provider belongs in. The same provider in a lifestyle blog's list of apps teaches almost nothing. - **How you are described.** "Example Ltd, which handles employment disputes for hospitality groups" teaches something specific. "Other firms include Example Ltd" does not. If every link to you uses the same keyword phrase as its anchor text, it looks placed rather than earned. - **How much is said.** A case study that spends several paragraphs on your work gives an assistant far more than a roundup of twenty names. - **Whether sources agree.** Use one plain description of what you do, who for and where, everywhere you control, so earned coverage repeats it. When sources conflict, assistants hedge, as covered in [why ChatGPT sounds unsure about our firm](/insights/ai-cross-references-business-sources). ## Does it matter when and how quickly links arrive? Yes. Assistants that search the web live find this year's coverage more readily than old pieces, and buyers read dates. A steady run of good mentions over a year looks like a firm people keep writing about; a burst of dozens of links in one week looks like a purchase. Repeat coverage from the same trade title is a good sign: a sector editor who quotes you three times a year is telling readers you are a regular voice. Why that steady rhythm matters is set out in [how long PR takes to work](/insights/compounding-effect-consistent-pr-12-months). ## How do we audit the sites that link to us? A useful first pass takes an afternoon. You do not need a paid tool. ### Step 1: Export the sites that link to you With Ahrefs, Semrush or Moz, export every referring domain. Without one, use the free [Links report in Google Search Console](https://support.google.com/webmasters/answer/9049606?hl=en), which shows top linking sites, top linked pages and top linking text; Google notes it is a sample, with tables limited to 1,000 rows. Bing Webmaster Tools, also free, lists sites linking to you too, and Bing's index feeds Copilot. ### Step 2: Sort each site into the six tiers Mark each as national press, trade or professional, smaller editorial, community or review, directory or wire, or harmful. Most firms find the bulk in tier five. That is normal, and it tells you what the headline number is made of. ### Step 3: Find the mentions that carry no link Search for your firm's name in quotation marks, excluding your own site, and set up a free alert such as Google Alerts for your firm and founder names. Add anything in tiers one to four to your list, with or without a link. ### Step 4: Read the best mentions properly Open the pages in tiers one to four and read what they say. Is the description accurate, current and in the right category? Note any old article about a service you no longer offer, because assistants may repeat it. ### Step 5: Compare with the competitors ChatGPT names Ask ChatGPT, Gemini, Copilot and Claude the question your buyers ask and note who is named. Perplexity and ChatGPT search show their sources: list the trade titles, bodies and review platforms that mention those competitors and not you. That gap is your target list. ## Should we stop paying for link building? Stop paying for links; keep paying for work that earns coverage. [Google's spam policies](https://developers.google.com/search/docs/essentials/spam-policies) treat buying or selling links that pass ranking credit as link spam, including paid articles and press releases with optimised anchor text. Link packages, paid guest-post networks and mass directory submissions add to the count without adding to the evidence. Work that earns mentions is different: expert comment, original research, bylined articles, speaking and client case studies. Ask any provider what it will earn, where, and how it will report it. The routes are compared in [digital PR or traditional PR for a B2B firm](/insights/digital-pr-vs-traditional-pr) and [which kinds of PR win work](/insights/pr-for-b2b-what-works). ## What to do this month 1. Export your linking sites from Search Console or your SEO tool and sort them into the six tiers. 2. Set up alerts for your firm and founder names, and list unlinked mentions worth keeping accurate. 3. Put the question your buyers ask to the main assistants, note who is named, and list the sources that mention them and not you. 4. Pause spending on directory submissions, wire distribution or guest-post packages until you know what they have delivered. 5. Pick the two trade titles your buyers read most and find out who edits the sections your firm would fit. ## Frequently asked questions ### How many referring domains does a B2B website need? There is no fixed number, and anyone who quotes one is guessing. It depends on how crowded your category is and how well your competitors are covered. Compare your tiers one to four with those of the firms being named, not with a target figure. ### How do I find unlinked mentions of my company online? Search for your firm's name in quotation marks while excluding your own domain, set up free alerts for your firm and founder names, and check the sources assistants cite when they answer your buyers' questions. Paid mention trackers do the same at scale. If a journalist mentioned you without a link, a polite request is reasonable, but the mention is already doing much of the work. ### Are any directories worth being listed in? Yes, a few. Registers kept by your regulator or professional body, the recognised directories in your sector, such as Legal 500 or Chambers for law firms, and well-used review platforms carry weight because they are curated or verified. Keep those entries accurate and consistent. ### Should we remove low-quality links pointing at our site? Usually not. Google says [most sites will not need to use its disavow tool](https://support.google.com/webmasters/answer/2648487?hl=en), because in most cases it can assess which links to trust. The exception is links you or a previous agency paid for or arranged at scale; take advice before acting. ### Do links from our own other websites count? Very little. Links from sites you own, such as a sister company or an old microsite, do not show that anyone else vouches for you. A case study on a client's website, in their own words, is worth more than a dozen links from your own properties. ### Can we just tell ChatGPT to correct what it says about us? Not in any lasting way. Correcting it in your own chat does not change what the model has learned or what other people see. Change the sources it reads; answers based on live search usually update within weeks, and the model's own knowledge follows when it is retrained, which takes months. If you would rather have someone run this audit and then earn the coverage that closes the gap, that is what [Reputation That Wins](/reputation-that-wins) is for. We start with the firms ChatGPT names instead of you and work back from there. The free [Growth Review](/growth-review) is a quicker first look. ## Sources - [Ahrefs, AI Overview brand visibility study of 75,000 brands (May 2025)](https://ahrefs.com/blog/ai-overview-brand-correlation/) - [Search Engine Journal, SE Ranking data on the top factors influencing ChatGPT citations (November 2025)](https://www.searchenginejournal.com/new-data-top-factors-influencing-chatgpt-citations/561954/) - [Ahrefs, AI Overview citations from top-ten pages (March 2026)](https://ahrefs.com/blog/ai-overview-citations-top-10) - [Ahrefs, The most-cited websites in Google AI Overviews (June 2026, updated September 2026)](https://ahrefs.com/blog/most-cited-domains-ai-overviews/) - [Semrush, LinkedIn AI visibility study (March 2026)](https://www.semrush.com/blog/linkedin-ai-visibility-study/) - [G2, The Answer Economy: half of B2B software buyers now start research with AI chatbots (April 2026)](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html) - [Google Search Central, Spam policies for Google web search (August 2026)](https://developers.google.com/search/docs/essentials/spam-policies) - [Google Search Console Help, Links report (September 2026)](https://support.google.com/webmasters/answer/9049606?hl=en) --- Title: Do our B2B buyers really look at our website on their phones? URL: https://fortitudemedia.ai/insights/mobile-first-design-b2b Author: David Adams Published: 2026-03-31 Last reviewed: 2026-09-24 Topic: Websites that bring work in Short answer: Yes. Phones made up 46.11% of UK web page views in August 2026 (StatCounter), and many B2B buyers first check a supplier on a phone, often from a LinkedIn post or an AI answer. Google indexes and ranks the mobile version of your pages. Design key pages for a phone first: a clear headline, one proof line, a tap-to-call number and a short form, then test them from inside LinkedIn every quarter. Most B2B websites are signed off in a meeting room. The agency plugs a laptop into the big screen, the partners look at the homepage, and everyone agrees it looks the part. Almost nobody in that room picks up their phone and tries to find the contact number with one thumb. Your buyers do. A finance director hears your name from a colleague at lunch and looks you up on the walk back to the office. An operations manager asks ChatGPT for three firms who could take on a contract and taps through to each site from the train. A managing director taps a link in a LinkedIn post on Sunday evening, before a Monday meeting. The first time many buyers see your website, they see it on a screen the size of a playing card. Phones are no longer the minority channel. [StatCounter's figures for August 2026](https://gs.statcounter.com/platform-market-share/desktop-mobile-tablet/united-kingdom) show phones accounted for 46.11% of web page views in the UK, against 49.17% for desktops and 4.72% for tablets. That covers all websites rather than B2B alone, but it means the phone version of your site is not a side issue. If that first look is a cropped photograph, a menu with ten items and a form that needs twelve fields, buyers go back to the list and try the next name. Nothing in your reports tells you it happened. That is the real case for designing for the phone first: the enquiries that quietly never arrive. ## Key takeaways - Phones made up 46.11% of UK web page views in August 2026, and many B2B buyers first look at a supplier on a phone even when the contract is signed at a desk. - Google uses the mobile version of your pages for indexing and ranking, so content that is missing or hidden on phones counts for less everywhere. - The first phone screen should say who you help, prove it in a line, and offer one obvious way to get in touch. - Tap-to-call numbers, short forms and no full-screen pop-ups often do more for mobile enquiries than a visual redesign; legally required cookie banners stay, kept small. - LinkedIn and other in-app browsers can break booking widgets, forms and tracking, so test your contact route from inside the apps your buyers use. - Visits from the ChatGPT phone app often arrive with no referrer and show up as Direct, so read your device reports with care. ## How much of your traffic is actually on phones? Across all UK websites, close to half. StatCounter recorded phones at 46.11% of UK page views in August 2026. We could not find a reliable published figure for B2B sites alone, so your own analytics are the number that matters, and they take five minutes to check. In Google Analytics, look at the device category report, then break it down by landing page and by traffic source. Three patterns are common in B2B firms. The homepage and about page get a large share of phone visits, because that is where people check you out. Articles shared on LinkedIn tend to be read on phones. Pages such as pricing, detailed service descriptions and case studies see more desktop visits, because that is where a buyer who is already interested does their homework. That split matters. The phone is where people decide whether you are worth a proper look, and the desktop is where they take it. Lose them on the phone and the desktop visit never happens. ## Why do our mobile visitors convert so much worse than desktop visitors? Partly because phone visitors are earlier in their research, and partly because the phone version is often harder to use. A buyer checking you on a train may come back on a laptop to enquire, which makes the phone visit look wasted when it did its job. But a large gap can also mean the phone route is broken: a form that will not submit, a booking link that fails inside the LinkedIn app, a pop-up covering the button or a page too slow to load on 4G. Tracking adds noise too, because some phone visits are recorded as Direct or lost altogether. The only way to know which is happening is to try the route yourself on a phone, which is what the rest of this guide is about. ## Why does Google care which version is better? Because Google now treats the mobile version of your pages as the main version. Its [mobile-first indexing guidance](https://developers.google.com/search/docs/crawling-indexing/mobile/mobile-sites-mobile-first-indexing) says it "uses the mobile version of a site's content, crawled with the smartphone agent, for indexing and ranking". If content is missing, cut down or hard to reach on mobile, it counts for less everywhere. This catches out plenty of firms. A developer hides a long section on phones to keep the page short, or puts the case studies behind a tab that only loads when tapped. Google says it does not load content that needs a tap, swipe or typing to appear, so that content drops out of the version Google weighs most. AI assistants work a little differently. ChatGPT, Claude and Perplexity read the text of the page rather than viewing it on a phone, so what matters is that the full content is present in the page itself, not assembled later by scripts, which [GPTBot, ClaudeBot and PerplexityBot do not run](https://vercel.com/blog/the-rise-of-the-ai-crawler) (Vercel and MERJ, December 2024). We explain how these crawlers read a site in [how AI crawlers read your website](/insights/ai-crawlers-vs-google-spiders). Google AI Overviews draw on Google's own index, so the mobile version matters there too. The practical rule: the same words, the same proof and the same structure on the phone and the desktop. Change the layout, not the content. ## What is the difference between responsive and mobile-first? A responsive site rearranges itself to fit a smaller screen. A mobile-first site was designed for the smaller screen to begin with, then given more room on a desktop. Almost every site built in the last decade is responsive. Far fewer are mobile-first, and the difference shows. A responsive site designed on a desktop carries the desktop's habits down to the phone. The large hero image pushes the headline below the fold. The three-column section becomes three long stacks. The menu collapses into an icon that hides ten items and several sub-menus. The contact form keeps every field. A mobile-first site starts with harder questions. What does someone need to see in the first screen to stay? Which three pages matter most? What is the shortest route from arriving to getting in touch? The answers usually produce a simpler, faster site that also works better on a desktop. ## What should a buyer see on the first phone screen? Three things: who you help and with what, one line of proof, and one obvious next step. If a buyer has to scroll to find out what you do, many will not bother. For example, a first screen that works for an outsourced finance firm might have: - A headline such as "Outsourced finance teams for UK manufacturers with 50 to 500 staff". - A proof line such as "Month-end closed in five working days", if that is true and you can show it. - A single button, "Talk to a partner", which opens a short form or dials the office. The weak version is more common: a full-screen photo of a handshake, a slogan about partnership and excellence, and a menu icon. The buyer learns nothing and has to work to find out whether you are relevant. We look at the signals that earn a buyer's confidence in [what makes a buyer trust a website enough to get in touch](/insights/psychology-of-trust-website-conversions). ## How do you make your key pages readable on a phone? Write and lay them out for a narrow screen and a distracted reader: short paragraphs, frequent plain headings, and the most useful information first in every section. ### Keep the navigation short Five or six main items is plenty for most B2B firms. Name them in the buyer's words, such as "Services", "Sectors", "Our work", "About" and "Contact". If a page cannot be reached in two taps from the homepage, ask whether it belongs in the menu at all. ### Lead every section with the answer On a phone, the first sentence of a section is often the only one read. Put the point first and the detail after. When we rebuilt [ID-Pal](/work/id-pal)'s sector pages, the brief was the same: "Five industry pages rebuilt: fund administration, wealth management, legal, accountancy and estate agency. Each opens with the question a compliance lead types, names the regulator, and answers inside the first ninety words." ### Watch tables, diagrams and PDFs Wide tables and detailed diagrams break on small screens. Turn tables into short lists where you can, and never put important information only in a PDF. ### Start case studies with the result Open with the client type, the problem and the outcome in two or three lines, then tell the story for those who want it. ## How do you make it easy to get in touch from a phone? Remove every bit of effort you can. On a phone, each extra field, pop-up or hidden number costs you enquiries. - Put a tap-to-call phone number and a tappable email address in the header or footer of every page. - Cut the enquiry form to what you need to reply: name, work email, phone if they want a call, and a short message. - Use the right field types, so the phone shows a number pad for phone numbers and an email keyboard for email addresses. - Avoid full-screen pop-ups. [Google's guidance on intrusive interstitials](https://developers.google.com/search/docs/appearance/avoid-intrusive-interstitials) says not to obscure the whole page and to use banners that take up a small part of the screen. It accepts interstitials needed for legal reasons, such as consent, so keep a compliant cookie banner and make it small. - If you offer booked calls, link to a calendar that works properly on a phone. In our experience, these small fixes often do more for mobile enquiries than a full visual redesign, and they take days rather than months. ## Do LinkedIn and other in-app browsers break forms and booking links? They can. When someone taps a link inside LinkedIn, or inside many email and messaging apps, the page often opens in the app's own built-in browser rather than in Safari or Chrome. That browser may not share the visitor's usual cookies or logins, may block pop-up windows, and may handle downloads and new tabs differently. The common casualties on B2B sites are booking widgets that open in a new window, forms that rely on a third-party script to submit, calendar links that ask the buyer to sign in, and analytics that treat every in-app visit as a new visitor. Since many LinkedIn visits happen on phones, this is where a lot of first visits happen. Test it directly. Post a link to your contact page in a private message to a colleague, open it on a phone inside the LinkedIn app, and try to book a call and send the form. Do the same from your email app. Anything that fails there is failing for real buyers. ## Should a B2B website have a sticky call button on mobile? Often, yes, if it is slim and does not cover content. A small bar at the bottom of the screen with "Call" and "Book a call" keeps the next step in reach on long pages. Keep it to one line, make sure it does not hide form fields or the cookie banner's buttons, and check it does not cover whatever a keyboard user has selected. It should never behave like a pop-up. ## How fast does it need to be on a phone? Fast enough that the main content appears almost at once on an ordinary 4G connection: Google's [Core Web Vitals thresholds](https://support.google.com/webmasters/answer/9205520?hl=en) ask for the main content within 2.5 seconds, a response to taps within 200 milliseconds and a layout shift score of 0.1 or less. We cover the thresholds and the fixes in [does our website's speed and accessibility really affect enquiries?](/insights/website-accessibility-speed-matter). ## How do you test your website on a phone properly? Use real phones and real people, not only the preview mode in a web browser. Emulators are useful for developers but they hide the problems a buyer actually meets. ### Step 1: Compare visits and enquiries by device in GA4 In Google Analytics 4, open Reports, then Tech, then Tech details, and set the dimension to device category. Add your enquiry form submission as a key event and compare the key event rate for mobile and desktop, then repeat for your top ten landing pages. Read the result with one caveat: [visits from the ChatGPT phone app often lose their referrer](https://www.nicelookingdata.com/blog/ga4-ai-traffic-chatgpt-referrals) and are recorded as Direct, and GA4's AI Assistant channel, added in May 2026, does not cover Claude or Perplexity. A rise in Direct phone visits may be AI traffic you cannot see. ### Step 2: Run your key pages through PageSpeed Insights Google's free PageSpeed Insights tool shows how your pages perform on mobile, with the Core Web Vitals results and the main causes of delay. Test the homepage, your main service page and your contact page. ### Step 3: Watch someone outside the firm use it Hand your phone to someone who does not know the business, ideally someone like your buyers. Ask them to find out what you do, find a relevant case study and get in touch. Say nothing while they try, and note where they hesitate. ### Step 4: Fix the worst three problems and test again Take the three problems that stopped your tester, fix them, then repeat the test next quarter. For more on how each page should earn its place, see our [guide to websites that bring work in](/insights/pillar/website-design-performance). ## What to do this month 1. Open GA4 and compare phone and desktop visits and enquiry rates for your homepage, main service page and contact page. 2. Try to get in touch with your own firm using only your phone, including from a link opened inside LinkedIn, and time how long it takes. 3. Add a tap-to-call number and shorten your enquiry form to four fields or fewer. 4. Replace any full-screen pop-up on phones with a small banner, keeping your cookie banner compliant. 5. Run your three key pages through PageSpeed Insights and send the results to whoever looks after your site, with a date to fix the worst issue. ## Frequently asked questions ### Our contracts are signed at a desk. Does the mobile site still matter? Yes. The decision may be made at a desk, but the shortlist is often made on a phone. If a buyer's first look at your site on their phone is confusing or slow, you may not make the shortlist that leads to the desk. ### Is designing for mobile first more expensive? Not usually. It is mostly a change in the order decisions are made, not in the amount of work. Mobile-first sites often end up simpler, with fewer heavy elements to build and maintain. ### Do we need a separate mobile site or an app? No. For B2B firms a single responsive website designed for the phone first is the right answer. Separate mobile sites mean two sets of content to maintain, and very few B2B buyers want to download an app to find out about a supplier. ### What about tablets? Tablets made up 4.72% of UK page views in August 2026, according to StatCounter. A site designed for phones and desktops almost always works on a tablet, so test your key pages on one if your analytics show a meaningful share, but most firms do not need to design for them separately. ### Will a better mobile site help ChatGPT recommend us? Indirectly. ChatGPT, Claude and Perplexity read the text of your pages rather than viewing them on a phone, so layout alone will not change what they say. But Google ranks on the mobile version, Google AI Overviews draw on that index, and making a site mobile-first usually makes the content clearer for every reader, human or machine. Your buyers are already checking you on their phones, and the firms that make that first look easy win the next conversation. If you want a website designed around how buyers actually look at you, [Website That Sells](/website-that-sells) starts with the phone, or book a free [Growth Review](/growth-review) for three fixes in writing. ## Sources - [StatCounter, Desktop vs mobile vs tablet market share in the United Kingdom (August 2026)](https://gs.statcounter.com/platform-market-share/desktop-mobile-tablet/united-kingdom) - [Google Search Central, Mobile-first indexing best practices (December 2025)](https://developers.google.com/search/docs/crawling-indexing/mobile/mobile-sites-mobile-first-indexing) - [Google Search Central, Avoid intrusive interstitials and dialogs (December 2025)](https://developers.google.com/search/docs/appearance/avoid-intrusive-interstitials) - [Google Search Console Help, Core Web Vitals report (accessed September 2026)](https://support.google.com/webmasters/answer/9205520?hl=en) - [Vercel, The rise of the AI crawler (December 2024)](https://vercel.com/blog/the-rise-of-the-ai-crawler) - [NiceLookingData, AI traffic in GA4: how ChatGPT, Perplexity, Gemini and Google AI Mode referrals show up (July 2026)](https://www.nicelookingdata.com/blog/ga4-ai-traffic-chatgpt-referrals) --- Title: Does ChatGPT prefer long articles or short ones? URL: https://fortitudemedia.ai/insights/long-form-vs-short-form-ai-prefers Author: Ashley Kenerson Published: 2026-03-31 Last reviewed: 2026-09-24 Topic: Content that gets quoted Short answer: Neither on its own. Google says there is no ideal page length, and studies disagree: SE Ranking found pages over 2,900 words gathered more ChatGPT citations, while AirOps found 500 to 2,000 words did best per query. What they agree on is position: a 2026 analysis of 1.2 million ChatGPT answers found 44% of citations came from the first 30% of a page. Answer early, make each section stand alone, and write as long as the question needs. Somebody on your team has told you that nobody reads long articles any more. Somebody else has told you that ChatGPT only quotes long, detailed pages. The budget is finite, the people who could write are busy with client work, and you need to decide whether the next twelve pieces should be 600 words or 3,000. The studies do not settle it, and they disagree with each other. What they do agree on is where the answer should sit. A [2026 analysis of 1.2 million ChatGPT answers by Kevin Indig](https://searchengineland.com/chatgpt-citations-content-study-469483), reported by Search Engine Land in February 2026, found that 44.2% of citations came from the first 30% of a page, 31.1% from the middle and 24.7% from the final third. Google, for its part, says in its [guide to generative AI features](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide) (updated July 2026): "There's no ideal page length". So the decision is less "long or short" than "how much does this question need, and is the answer near the top". This guide gives you a way to decide per piece, sets out what the studies found and why they conflict, and explains how to write sections that can be lifted on their own. ## Key takeaways - There is no ideal length: Google says so, and the published studies point in different directions. - Put the direct answer and the key figure in the first third, because 44% of ChatGPT citations in one large 2026 analysis came from the first 30% of the page. - Big buying questions usually need more words than definitions, because the buyer needs reasons, trade-offs and an example, but padding does not help at any length. - Definitions, single-task how-tos and news are better short; one 2026 study found pages over 5,000 words were cited less often than pages under 500. - Assistants and Google's AI features lift single passages, so every section should answer its own question in its first two sentences. - Split a very long guide into several focused pages when it answers several distinct questions. ## Do AI assistants favour longer pages? Not on their own. The studies disagree, and Google says there is no ideal length. What counts is whether the page answers the question completely and early. Longer pages, when they are good, contain more of what an assistant can use: a clear definition, the reasoning, the exceptions, a worked example and answers to the obvious follow-up questions. For example, when a buyer asks what should be in a managed IT support contract for a 40-person accountancy firm, a page that names the things that belong in the contract, explains why each matters and notes the two that firms usually forget gives an assistant something to build on. A page that says "a good contract should cover response times, scope and costs" is not wrong, but gives it little to use. Specific sentences take more words. "Response time targets should distinguish between one user's problem and a whole-site outage" is a sentence an assistant can lift. "Make sure response times are clearly defined" is not. That is why good pages on big questions tend to be longer, and why padding, which adds words without adding specifics, does not help. ## What do the studies say about length? They disagree, mainly because they measure different things. The table sets out the three most useful recent studies and Google's position. | Source | What it measured | What it found about length and position | |---|---|---| | SE Ranking, November 2025 (216,524 pages, 20 niches) | Average number of ChatGPT citations per page | Articles over 2,900 words averaged 5.1 citations; under 800 words, 3.2. Sections of 120 to 180 words between headings did best | | AirOps, April 2026 (16,851 queries, 353,799 pages) | How often a retrieved page was cited for one query | Pages of 500 to 2,000 words did best; pages over 5,000 words were cited less often than pages under 500. Narrow answers beat broad guides | | Kevin Indig, February 2026 (1.2 million ChatGPT answers) | Where on the page cited passages sat | 44.2% from the first 30% of the page, 24.7% from the final third | | Google, July 2026 | Official guidance | "There's no ideal page length" | Why do they conflict? [SE Ranking](https://www.searchenginejournal.com/new-data-top-factors-influencing-chatgpt-citations/561954/) counted citations per page across many prompts, and a long page covers more sub-questions, so it can be cited for more of them. [AirOps](https://searchengineland.com/chatgpt-citations-ranking-precision-length-study-474538) measured how often a page was cited when it was retrieved for one specific query, and a focused page that answers that query directly wins more often than a long guide that answers it somewhere in the middle. Both are correlational, and both come from firms that sell search tools. Our reading: long pages can collect more citations overall, but each citation goes to a passage that answers one question well. Google's guidance on [creating helpful content](https://developers.google.com/search/docs/fundamentals/creating-helpful-content) makes the same point from the other side, asking whether you are writing to a particular word count because you heard Google prefers one: "No, we don't." ## How long should our articles be? As long as the question needs. The ranges below are what we work to, by type of page. They are working ranges, not rules. | Type of page | Working range in our experience | What it must contain | |---|---|---| | Glossary entry | 200 to 400 words | A one-line definition, a plain explanation, an example | | Definition or news page | Under 1,000 words | The answer, why it matters, what to do | | Single-task how-to or narrow comparison | 1,000 to 1,800 words | The steps, common mistakes, when not to do it | | Big buying question (which option, what it costs, what goes wrong) | Often 1,500 to 3,000 words | The answer first, reasons, trade-offs, an example, the follow-up questions | | Your own research or a full process guide | Longer, if every section earns its place | Method, findings, limits | The floors in our other guides describe the same pattern from different angles. [What an article that ChatGPT quotes looks like](/insights/anatomy-of-article-cited-by-ai) treats a broad B2B subject as usually under-treated below about 1,200 to 1,500 words, and our [content audit method](/insights/audit-existing-content-ai-readiness) scores a page of under about 800 words that covers a subject at one level as shallow. A 300-word glossary entry is not shallow, because its question is narrow. The test is always the question, never the count. For example, an employment law firm's page on "what is a settlement agreement" can be well under 1,000 words and be better for it. "Should we offer a settlement agreement or go through a formal process" is a big buying question, because the buyer needs the trade-offs, timings and risks before they can act. ## Does ChatGPT read the whole page or only the top? It can read the whole page, but its citations cluster near the top. In the Indig analysis, 44.2% of ChatGPT citations came from the first 30% of the page, with a sharp drop near the footer. Cited passages were also nearly twice as likely to use clear definitions ("X is", "X refers to"). So put the direct answer in the first paragraph, the key figure in the first third, and a plain definition wherever you introduce a term. ID-Pal, an identity verification firm that had bought NorthRow, rebuilt its industry pages on this pattern. Its case study records: "Five industry pages rebuilt: fund administration, wealth management, legal, accountancy and estate agency. Each opens with the question a compliance lead types, names the regulator, and answers inside the first ninety words." And: "Gemini now names ID-Pal for UK wealth managers. Claude names it for UK fund administrators, on a page rebuilt in August, and links NorthRow to ID-Pal in the same answer." The [ID-Pal case study](/work/id-pal) has the detail. ## Why does every section need to answer its own question? Because assistants and Google's AI features often lift one passage, not a whole page. Google's AI Mode and AI Overviews use query fan-out, which Google describes as "a set of concurrent, related queries generated by the model", and each related query can match a different section of your page. That is also why ranking alone is not enough. [Ahrefs found in March 2026](https://ahrefs.com/blog/ai-overview-citations-top-10) that only 38% of AI Overview citations came from pages in the top ten, down from 76% in July 2025. A section that answers a follow-up question clearly can be cited even when the page does not rank first for the main one. - Make each heading the sub-question in the buyer's words. AirOps found pages with the strongest match between headings and the query were cited 41.0% of the time, against roughly 30% for weaker matches. - Answer the heading in the first two sentences, then explain. - Name the subject at the start of a section instead of "it" or "as mentioned above", so the passage makes sense on its own. - Keep sections to a few hundred words. SE Ranking found sections of 120 to 180 words between headings performed best. Google's guide says there is no need to break content into tiny pieces for AI, and we agree. The aim is sections that stand on their own inside a complete page. ## When is shorter content the better choice? When the question is narrow, factual or time-sensitive, and extra context would only get in the way. - **Definitions.** "What is Cyber Essentials certification" needs a clear definition, a few paragraphs of explanation, one example and a note on who needs it. - **Single-task instructions.** "How to register for the Construction Industry Scheme as a contractor" is a list of steps and the common mistakes. - **News.** A change in regulation, a new appointment, a new service: say what happened, what it means and what to do. - **Tool-specific help.** Setting up one feature in one system rarely needs more than a short page. Short wins when the question has one answer and no real trade-offs. Long wins when the buyer is deciding something and needs the reasoning. Most pages that get a firm named fall into the second group, but the short pages are the supporting cast, and they link to the main pieces. ## Should we split a long guide into several pages? Usually, when it answers several distinct questions. AirOps found that pages answering the main query narrowly outperformed broader, more comprehensive guides, and a 6,000-word "ultimate guide" hides the answer to any one question somewhere in the middle. Break it into a short overview page that links to the separate pieces, keep the original address for the overview, and give each new page its own question as its title and a direct answer at the top. You keep the content and gain several quotable pages. Covering the follow-up questions buyers ask matters because Google fans out into them; [organising content so AI assistants see you as the expert](/insights/topic-clusters-ai-understands) explains how to map them before you write. ## How do we tell whether a long article is earning its length? Ask five questions of the draft. If it fails any of them, it is too long, or it is two articles. 1. Can you say what the article argues in one sentence? 2. Can you name the point of each section in a few words? 3. Would removing any section leave the reader worse off? 4. Is there a shape: the question, why it matters, how it works, where it goes wrong, what to do? 5. Does every claim have an example, a figure or a reason behind it? A piece that passes all five at 2,400 words is finished. A piece that fails two of them at 4,000 words needs an editor, not more words. ## What to do this month 1. List the ten questions that most decide whether a buyer shortlists you, and mark each as a big question or a narrow one. 2. Check the first 30% of your five most important pages: if the direct answer and the key figure are not there, move them up. 3. Read every heading on those pages and check that the first two sentences under it answer it. 4. Run the five-question test on any page over 4,000 words, and split those that answer several questions. 5. Write two short pages for narrow questions that link to your strongest big-question page. ## Frequently asked questions ### What is the ideal article length to get cited in Google AI Overviews in 2026? There is none. Google's own guide says "There's no ideal page length", and the studies disagree. What helps is a direct answer early, headings that match the questions people ask, and sections that answer the follow-up questions Google's fan-out searches for. ### Will anyone actually read a 2,500-word article? The people who matter will, if it answers their question properly. What loses readers is padding, poor structure and no answer at the top. Give the answer first and a reader, or an assistant, can take what it needs and stop. ### Is a 500-word answer page enough to get cited by Perplexity? For a narrow, factual question, it can be. AirOps found pages under 500 words were cited more often by ChatGPT than pages over 5,000; we know of no published length study for Perplexity itself. For a buying question with trade-offs, 500 words usually leaves out the reasoning, and a fuller page will serve the buyer better. ### How long should a B2B service page be for Google and AI? Long enough to say who the service is for, what problem it solves, how it works, when it is not the right fit, and what proof you have, with the answer to "what do you do" in the first paragraph. In our experience that is usually shorter than a guide, and the guides do the deeper explaining and link to it. ### We have a 6,000-word guide that took weeks to write. Should we cut it? Probably split it. A very long guide usually contains three or four questions that each deserve a page with its own title. Keep the original address for a short overview that links to them. ### Is it better to publish often or to publish long? Publish steadily, at the depth each question needs. In our experience a regular rhythm of well-structured pages builds more than an occasional very long one. [How often to publish for AI assistants](/insights/how-often-publish-content-ai) sets out a realistic cadence. ### Does word count matter for videos and podcasts? Not in the same way. A recording helps with assistants once it is written up, and the written version needs the same structure as any other page. Our piece on [turning webinars and podcasts into written content](/insights/video-podcast-transcripts-ai-content) covers the process. Write as much as the question needs, put the answer where it will be found, and stop. If you want a second opinion on which of your pages are earning their length, the free [Growth Review](/growth-review) is where we usually start: 30 minutes with David Adams and three fixes in writing. If the site's structure is the problem, that is [Website That Sells](/website-that-sells). ## Sources - [Search Engine Land, 44% of ChatGPT citations come from the first third of content: study (February 2026)](https://searchengineland.com/chatgpt-citations-content-study-469483) - [Google Search Central, Guide to optimizing for generative AI features on Google Search (July 2026)](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide) - [Search Engine Journal, New data: top factors influencing ChatGPT citations, SE Ranking study (November 2025)](https://www.searchenginejournal.com/new-data-top-factors-influencing-chatgpt-citations/561954/) - [Search Engine Land, AirOps study of ChatGPT citations, ranking and page length (April 2026)](https://searchengineland.com/chatgpt-citations-ranking-precision-length-study-474538) - [Google Search Central, Creating helpful, reliable, people-first content (December 2025)](https://developers.google.com/search/docs/fundamentals/creating-helpful-content) - [Ahrefs, AI Overview citations and top-ten rankings (March 2026)](https://ahrefs.com/blog/ai-overview-citations-top-10) --- Title: Does ChatGPT prefer newer content, and how old is too old? URL: https://fortitudemedia.ai/insights/content-freshness-recency-ai Author: Ashley Kenerson Published: 2026-03-31 Last reviewed: 2026-09-24 Topic: Content that gets quoted Short answer: Yes, on average. Ahrefs found in July 2025 that pages cited by AI assistants were 25.7% fresher than Google's ordinary results, with ChatGPT the most recency-minded, and Seer found in July 2026 that 75% of cited pages had been updated within a year. Old pages still get cited if they are maintained. Review the twenty pages tied to revenue every quarter and everything else at least yearly, show an honest Updated date that matches your markup and sitemap, and never redate an unchanged page. Most founder-led firms we talk to have a website with a hundred or so pages, half of them written years ago by someone who has since left. The worry is usually one of two things. Either "our content is old, so ChatGPT must be ignoring it", or "we cannot keep rewriting everything, so what is the point". Both worries are reasonable. Both are slightly wrong. AI assistants do lean towards recent pages, and ChatGPT leans hardest. [Ahrefs studied about 17 million citations across seven AI platforms in July 2025](https://ahrefs.com/blog/do-ai-assistants-prefer-to-cite-fresh-content) and found that the pages assistants cite were on average 25.7% fresher than the pages in Google's ordinary results. ChatGPT's citations were the newest, at an average of 958 days old. A year later, [Seer Interactive found](https://www.seerinteractive.com/insights/study-content-recencys-impact-on-ai-visibility-in-2026) that 75% of the pages AI assistants cited had been updated in the previous 12 months. Fresh does not mean published last week. Even ChatGPT's average cited page is more than two and a half years old; what assistants favour is a page that is still looked after. This guide explains what "current" means to an assistant, how often to review which pages, how to show dates, and how to keep a library current without rewriting it. ## Key takeaways - AI assistants cite newer pages than Google's ordinary results do: Ahrefs found AI citations 25.7% fresher on average in July 2025, with ChatGPT the most recency-minded of seven platforms. - Updating a good existing page usually does more than writing a new one: most pages AI cites were first published more than a year earlier and updated since. - The twenty or so pages tied to revenue deserve a review every quarter; SE Ranking found pages updated within three months averaged 6 ChatGPT citations against 3.6 for older ones. - Every other page should be reviewed at least once a year, grouped by subject, with a one-hour light update rather than a rewrite. - Show a visible "Updated" date, and make it match the dateModified value in your markup and the lastmod value in your sitemap. - Never put a new date on an unchanged page: it misleads the buyer reading it and can make Google stop trusting your sitemap dates. ## Do AI assistants prefer newer content? Yes, on average, and ChatGPT more than the others. They favour pages that are current, which is different from pages that are merely new. An assistant uses your content in two ways. The first is its memory: what the model learned from a snapshot of the web taken months or years earlier. The second is live search: ChatGPT search, Perplexity, Copilot, Gemini and Google's AI features look pages up when a buyer asks, and judge your page as it stands today. Google says its AI features ground their answers in its search index partly to improve freshness ([Google Search Central, updated July 2026](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide)). The Ahrefs study shows how much the platforms differ. Average age of the pages each one cited, in days since publication: | Platform | Average age of cited pages | |---|---| | ChatGPT | 958 days | | Microsoft Copilot | 1,056 days | | Gemini | 1,118 days | | Perplexity | 1,166 days | | Google AI Overviews | 1,432 days | | Google's ordinary results, for comparison | 1,416 days | Source: Ahrefs, July 2025, about 17 million citations. Two 2026 studies point the same way. Seer found that 88% of cited pages had been updated within two years, and that among pages showing both dates, 72% looked fresh by their last update while only 42% had first been published in the past year. Seer covered four consumer and finance industries, not B2B services, so treat it as a direction rather than a rule. An [AirOps study of 353,799 pages reported by Search Engine Land in April 2026](https://searchengineland.com/chatgpt-citations-ranking-precision-length-study-474538) found that pages published 30 to 89 days earlier were cited most by ChatGPT, pages newer than 30 days did worse, and pages more than two years old were cited less often. Subject matters too. A three-year-old page on what an audit involves is still broadly right; a three-year-old page on audit exemption thresholds is wrong. In our experience recency counts for most where the answer moves. ## How old is too old for our content? For the pages that bring in work, three months without a review is when they start to lose ground. For everything else, a page untouched for more than a year sits outside where most cited pages are, and one untouched for more than two years is in a small minority. That replaces the old rule of thumb of "twelve to twenty-four months is fine", which was looser than the data. [SE Ranking's study of 216,524 pages, reported by Search Engine Journal in November 2025](https://www.searchenginejournal.com/new-data-top-factors-influencing-chatgpt-citations/561954/), found that pages updated within three months averaged 6 ChatGPT citations, against 3.6 for outdated pages. It is a correlation, since firms that update often may be doing other things well, but it is the best case for a shorter cycle on your most important pages. | Page type | Examples | How often to review | Why | |---|---|---|---| | The twenty pages tied to revenue | Service pages, sector pages, comparison pages, the guides that bring in enquiries | Every quarter | SE Ranking: pages updated within three months averaged 6 citations against 3.6 | | Fast-moving subjects | Tax, employment law, regulation, technology, anything with a threshold or a regulator | At least yearly, and whenever the facts change | A wrong figure is worse than an old one | | Moderately changing subjects | How a service works, what to look for in a supplier, common mistakes | Yearly | Seer: 75% of cited pages updated within 12 months | | Slow subjects | Definitions, principles, how a process works in general | Yearly, marked as reviewed even when nothing changes | Seer: 88% of cited pages updated within two years | | Services you no longer offer | Old service pages, retired products | Redirect or retire now | Old, wrong pages can be the ones an assistant picks up | For example, a firm of chartered surveyors has a 2022 guide to dilapidations. The law has barely changed, but the example figures are old and two links are dead. That page needs an hour, not a rewrite: new figures, working links, a paragraph on anything new and an honest "Updated" line. ## Should we update old content or write new articles? Update first, wherever a good page already exists. In Seer's data, only 42% of cited pages showing both dates had first been published in the past year, yet 72% had been updated in it. An old page also carries history a new one lacks: links from other sites, mentions and a track record at the same address. Write new pages where a buyer question has no page that answers it. As a working balance, a firm with fifty or more solid pages should mostly keep the best ones current and fill gaps; a firm with fewer than thirty should mostly write new pages that answer buyer questions. For how much new publishing a firm like yours needs, see [how often we need to publish for AI assistants to notice us](/insights/how-often-publish-content-ai). Rebuilding existing pages is often where results show first. For [ID-Pal](/work/id-pal), an identity verification firm, the work included "Five industry pages rebuilt: fund administration, wealth management, legal, accountancy and estate agency. Each opens with the question a compliance lead types, names the regulator, and answers inside the first ninety words." The result so far: "Where a page has been rebuilt, ID-Pal appears. Where it has not yet, it does not, which tells us exactly what to build next." ## Should we republish old articles with a new date? No. Keep the original publication date, add an honest updated date, and say what changed. The old argument was that a changed date on an unchanged page is easy to spot. The evidence says otherwise. [A September 2025 study by Fang and others](https://arxiv.org/abs/2509.11353) added artificial publication dates to passages and asked seven language models, including GPT-4o, to rank them. The "fresh" passages were consistently promoted, with individual items moving by as many as 95 places. Larger models reduced the effect but none removed it. So a fake date might work on a machine for a while. The case against it is honesty: a date is a promise to the reader that someone checked the page. Google's guidance says dates "must describe the publication or update date of the page" ([Google Search Central](https://developers.google.com/search/docs/appearance/publication-dates)), and Ahrefs reports that Google's John Mueller has warned against changing publish dates without changing the content. There is a practical cost too: Google says it uses the lastmod date in your sitemap only if it is "consistently and verifiably" accurate ([Google Search Central, July 2026](https://developers.google.com/search/docs/crawling-indexing/sitemaps/build-sitemap)). Redate pages that have not changed and you risk Google ignoring the dates on the pages that have. A reviewed date is fine when the review was real: if someone read the page and confirmed it is still right, say "Reviewed September 2026". And never rewrite a strong, well-linked article at a new address just to get a new date. Update it in place and keep the URL. ## How do we show the dates correctly? Show the dates in three places and make them agree: the visible line on the page, the Article markup and the sitemap. Google recommends a user-visible date labelled "Published" or "Last updated", and says the visible and structured dates should match. Example: a guide first published on 14 March 2023 and updated on 15 September 2026. ``` Visible line under the title: Published 14 March 2023. Updated 15 September 2026: new thresholds added, examples refreshed. Article markup (JSON-LD in the page's HTML): Sitemap entry: https://www.example.co.uk/insights/dilapidations-guide 2026-09-15 ``` Change dateModified and lastmod only when the content, figures or links changed meaningfully, not for a typo or the copyright year, and check that your content system does not stamp a new date on every save. Write the update note in plain words, because that is the part a reader and a model can both see. For more on what markup does and does not do, see [whether we need schema markup for AI assistants](/insights/structured-data-schema-markup-ai). ## Does ChatGPT read the dateModified in our schema markup? OpenAI has not said that it does. Google reads datePublished and dateModified to decide which date to show, so keep the markup accurate, but do not rely on it to reach ChatGPT. The date any model can be sure to see is the one written in the page text. The Fang study put dates into the passage text, and the models reacted to them. That is why the visible "Updated" line matters most, and why all three dates should agree. ## How do we get ChatGPT to see an updated page faster? Make sure the crawlers can reach the page, then tell the indexes it has changed. Each assistant looks in a different place. | Assistant | Where it finds pages | What to do after an update | |---|---|---| | ChatGPT search | OpenAI's OAI-SearchBot, and it has drawn on Bing | Allow OAI-SearchBot in robots.txt and in your CDN settings; notify Bing through IndexNow or Bing Webmaster Tools | | Microsoft Copilot | Bing's index | Notify Bing through IndexNow or Bing Webmaster Tools | | Gemini, AI Overviews and AI Mode | Google's index | Use Request indexing in Search Console's URL Inspection tool; keep sitemap lastmod accurate | | Perplexity | PerplexityBot | Allow PerplexityBot; link the page from pages that are already crawled often | | Claude | Claude-SearchBot | Allow Claude-SearchBot in robots.txt and your CDN | [IndexNow](https://www.indexnow.org/faq) lets a site tell Bing and several other engines the moment a page is added, updated or deleted; Google does not take part. OpenAI says a site must allow OAI-SearchBot to appear in ChatGPT search ([OpenAI Help Center](https://help.openai.com/en/articles/9237897-chatgpt-search)). If you use Cloudflare, check its bot settings: it has blocked AI crawlers by default on new domains since July 2025. Our guide to [how AI crawlers read your website](/insights/ai-crawlers-vs-google-spiders) covers the robots.txt side. ## How to keep a content library current without rewriting it Schedule reviews, keep them light and batch them by subject. ### Step 1: List what you have Export every page with its publication date and its last update, and sort by age. For a firm with a hundred pages this takes an hour. ### Step 2: Mark the twenty that matter Pick the twenty or so pages that answer the questions buyers ask or already bring in enquiries, and put them on the quarterly cycle. If you are not sure which pages matter, our guide on [how to audit our existing content for AI assistants](/insights/audit-existing-content-ai-readiness) gives a way to score them. ### Step 3: Set a review calendar by subject Group the remaining pages by subject. January: everything on subject A. April: subject B. July: subject C. October: subject D. This spreads the work and means every page is seen at least once a year. ### Step 4: Do a one-hour light update Refresh figures and say which year each is from, add a recent example, fix dead links, add a paragraph on anything new, write a short update note and change the three dates together. Only rewrite when a page is wrong or a competitor's page is clearly better, and then aim for the standard in [what an article that ChatGPT actually quotes looks like](/insights/anatomy-of-article-cited-by-ai). ### Step 5: Retire what is not worth keeping A page on a service you no longer offer should be redirected to the nearest current page or removed, because old, wrong pages can be the ones an assistant picks up when it describes you. Age alone is not a reason to delete: some of your oldest pages are your most linked and most cited. ## How do we tell whether it is working? Keep a fixed list of 20 to 50 buyer questions and ask each several times a month in the main assistants, in a clean session, because answers vary from run to run. Record whether you are named and which page is cited, and count a win only when it repeats. Where an old page is still being cited, leave it alone. Where a page has stopped being cited, review it next. Google Search Console has also shown [performance reports for AI Overviews and AI Mode](https://developers.google.com/search/blog/2026/06/gen-ai-performance-reports) to all sites since 31 August 2026, which lets you see whether updated pages gain impressions there. Our guide on [tracking whether AI tools are recommending your business](/insights/track-ai-recommending-your-business) sets this up as a routine. ## What to do this month 1. Export a list of every page on your site with its publication date and last update, and sort it by age. 2. Mark the twenty pages tied to revenue, put a quarterly review in the calendar and name the person who owns it. 3. Add a visible "Published" and "Updated" line to your article template, make the markup and sitemap dates match it, and stop your content system changing dates on trivial saves. 4. Give the ten oldest pages that still matter an hour each: refresh figures, add a recent example, fix links and write an update note. 5. Redirect or retire any page that is no longer true. 6. After each update, request indexing in Search Console and notify Bing through IndexNow. ## Frequently asked questions ### How often should I update blog posts for ChatGPT and AI Overviews? Review the twenty or so pages that bring in work every quarter, and every other page at least once a year. SE Ranking found pages updated within three months were cited more by ChatGPT, and Seer found 75% of cited pages had been updated within a year. AI Overviews showed the weakest preference for new pages in the Ahrefs study. ### Does changing the date on a blog post help SEO or AI citations? Changing the date without changing the page is misleading, and we do not recommend it. A 2025 study showed language models do promote passages with newer dates, but Google's John Mueller has warned against it, readers lose trust when a "new" page quotes old figures, and Google may stop trusting your sitemap dates. Change the date when you have changed the content, and say what changed. ### Should we show the original publication date or hide it? Show it, alongside the latest update. Hiding the publication date does not make a page look newer; it makes it impossible to place in time, and Google recommends a visible, labelled date. A page maintained since 2022 is a better signal than one that appeared last month with no history. ### Do Google and the AI assistants treat freshness the same way? No. In the Ahrefs study ChatGPT cited the newest pages and Google AI Overviews cited pages of about the same age as Google's ordinary results. Honest dates and real updates serve both. ### How long after an update does an assistant notice? Assistants that search the web live, such as ChatGPT search, Perplexity, Copilot and Google's AI features, can reflect an updated page within weeks, once it has been recrawled; Google says [crawling can take from a few days to a few weeks](https://developers.google.com/search/docs/crawling-indexing/ask-google-to-recrawl). Answers drawn from a model's memory change only when the model is retrained, which takes months. For ID-Pal, the tracked count moved month by month: "UK buyer questions where AI names ID-Pal: none in May, two in August, four in September 2026." Keeping content current is mostly discipline: a review calendar, a template with honest dates and a clear note when something changes. If you want an outside view of where your site stands, [book a free Growth Review](/growth-review): 30 minutes with David Adams, and three fixes in writing afterwards. The rest of our [content strategy and authority guides](/insights/pillar/content-strategy-authority) pick up from there. ## Sources - [Ahrefs, Do AI assistants prefer to cite fresh content? (July 2025)](https://ahrefs.com/blog/do-ai-assistants-prefer-to-cite-fresh-content) - [Seer Interactive, Content recency's impact on AI visibility in 2026 (July 2026)](https://www.seerinteractive.com/insights/study-content-recencys-impact-on-ai-visibility-in-2026) - [Search Engine Journal, SE Ranking data on the top factors influencing ChatGPT citations (November 2025)](https://www.searchenginejournal.com/new-data-top-factors-influencing-chatgpt-citations/561954/) - [Search Engine Land, AirOps study of ChatGPT citations, ranking and length (April 2026)](https://searchengineland.com/chatgpt-citations-ranking-precision-length-study-474538) - [Fang and others, Do large language models favor recent content? (arXiv, September 2025)](https://arxiv.org/abs/2509.11353) - [Google Search Central, Influence your byline dates in Google Search (December 2025)](https://developers.google.com/search/docs/appearance/publication-dates) - [Google Search Central, Build and submit a sitemap (July 2026)](https://developers.google.com/search/docs/crawling-indexing/sitemaps/build-sitemap) - [Google Search Central, Optimizing your website for generative AI features on Google Search (July 2026)](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide) --- Title: Does ChatGPT trust press coverage more than our website or our ads? URL: https://fortitudemedia.ai/insights/earned-owned-paid-authority-signals Author: Ashley Kenerson Published: 2026-03-31 Last reviewed: 2026-09-24 Topic: Press, reviews and being trusted Short answer: Yes. AI assistants lean hardest on earned coverage, meaning what independent editors, judges and clients chose to say about you: it made up 84% of the links cited by ChatGPT, Claude and Gemini in Muck Rack's May 2026 study, against 0.3% for paid content. Shared content such as Reddit and reviews is also cited heavily. Your website tells assistants what you do but needs outside confirmation. Put most reputation effort into earning mentions, then make your site say the same. Look at where your marketing money went last year. For many founder-led B2B firms, a good share went on things that were bought: a sponsored feature in a trade magazine, an awards entry and a table at the dinner, a paid listing in an industry directory, a few boosted LinkedIn posts. Another share went on your own website and content. A small share, often nothing, went on getting other people to write about you because they wanted to. That split made some sense when the aim was to be seen. It makes much less sense now that buyers ask ChatGPT, Gemini, Claude, Copilot or Perplexity who to shortlist. The assistants draw overwhelmingly on what others say: in [Muck Rack's May 2026 analysis](https://www.globenewswire.com/news-release/2026/05/07/3290268/0/en/generative-pulse-earned-media-consistently-drives-ai-citations-holding-at-84.html) of more than 25 million links cited by ChatGPT, Claude and Gemini, earned media made up 84% of citations and paid or advertorial content 0.3%. So the question for your next budget meeting is which kind of reputation spend, as much as how much. This guide sorts that spend into four buckets, earned, shared, owned and paid, explains how AI assistants treat each one, and sets out a simple way to move money from the bucket that does least to the ones that do most. ## Key takeaways - Earned mentions, where an independent editor, judge or client chose to talk about you, carry the most weight: they were 84% of the links AI tools cited in Muck Rack's 2026 study. - Shared content, such as Reddit threads, LinkedIn discussions and review sites, is neither yours nor editor-chosen, yet AI assistants cite it heavily. - Your own website is where assistants learn what you do, and they trust it more when outside sources say the same thing. - Paid placements buy attention from people; clearly labelled sponsored content adds little to how assistants describe or recommend you. - ChatGPT shows labelled ads to UK users on its free plans, but OpenAI says ads do not change the answer, so a place in the answer is still earned. - Most firms can fund more earned activity by cutting the paid items nobody can link to an enquiry, a meeting or a piece of coverage. ## What counts as earned, shared, owned and paid? It comes down to one question: who decided this piece of content should exist? If an independent editor, judge or client decided, it is earned. If members of the public decided, on a platform nobody controls, it is shared. If you published it on your own channels, it is owned. If you paid someone to publish it, it is paid. ### Earned Anything an independent person or organisation chose to say about you without being paid to. A journalist quotes your managing director. A trade body shortlists you for an award judged by a panel. A client writes about the project on their own site. A directory lists you because its researchers assessed you. The common thread is that someone else put their own reputation behind the mention. ### Shared Conversation on platforms where anyone can post: Reddit threads, comments and discussions on LinkedIn, forum posts, and reviews on Google, Trustpilot, G2 or Clutch. You can take part, but you cannot control what is said, and neither can an editor. That is why the assistants value it: it reads as first-hand experience. ### Owned Everything you publish on channels you control: your website, service pages, case studies, articles, team biographies, your LinkedIn company page and your founders' own posts. It is the most complete description of what you do, and also the one with the most obvious reason to flatter you. ### Paid Space you bought. Sponsored articles and advertorials, awards where every paying entrant gets a badge, speaking slots bundled with a sponsorship package, premium directory placements, boosted posts and advertising. The test: if you stopped paying, would it exist? ### The grey areas Paying an entry fee for a properly judged award is still earned if the judging is independent and most entrants do not win, as [whether awards and accreditations make ChatGPT more likely to recommend you](/insights/awards-accreditations-ai-trust) explains. Sponsoring an event where you were already invited to speak is mostly earned. When in doubt, ask whether a sceptical buyer, reading it cold, would believe someone chose you. ## What is the PESO model, and how does it apply to AI search? PESO stands for paid, earned, shared and owned, the four buckets above, and PR teams use it to plan how the four work together. For AI search it is a useful map, because assistants weigh the buckets very differently. | Bucket | Who decides it exists | How AI assistants tend to treat it | |---|---|---| | Earned | Journalists, judges, researchers, clients | Most cited and most trusted: 84% of cited links in Muck Rack's study | | Shared | Members of the public on open platforms | Heavily cited, especially Reddit and LinkedIn, and prone to swings | | Owned | You | The reference for what you do; stated plainly when others agree | | Paid | Whoever sells the space | Rarely cited: paid and advertorial content was 0.3% of cited links | The practical point is that earned and shared carry the credibility, and owned carries the detail. Paid mostly buys reach with people. ## How do AI assistants decide what to believe about you? They gather several accounts of your firm and repeat the parts that agree. The accounts come from what the models absorbed in training and from what they find when they search the web for a question, as ChatGPT search, Perplexity, Copilot and Google's AI features do for many questions. That has three practical consequences. First, independent sources count for more than self-description. When an assistant can only find your own claims, it tends to hedge, with phrases like "according to the company". When it finds the same claim in a trade publication, a client's case study and a directory profile, it states it plainly. Why assistants hedge, and how to fix it, is covered in [why ChatGPT sounds unsure about your firm](/insights/ai-cross-references-business-sources). Second, consistency matters as much as volume: if your website says four offices and your LinkedIn page three, the assistant has to pick one or leave the detail out. Third, substance beats placement. Assistants quote sentences; they cannot quote a banner. ## Why do earned mentions carry the most weight? Because they are the only evidence that costs someone else something, and the data bears it out: earned media was 84% of the links AI tools cited in Muck Rack's study, and journalism alone 27%. The journalist risked their credibility by quoting you; the judges put their names to a shortlist; the client put your name on their own website. Earned mentions also tend to be specific. A good trade article says what you did, for whom and why it mattered, which is the raw material an AI answer is built from. And they last: a well-placed article stays online and keeps being found, although Muck Rack also found that more than half of cited journalism was published in the previous 12 months, so a steady rhythm matters more than one big hit. How that builds is covered in [how long PR takes to work and what a year should look like](/insights/compounding-effect-consistent-pr-12-months). ## Why does ChatGPT cite Reddit threads instead of our website? Because Reddit is full of first-hand answers to the exact questions buyers ask, written by people with no reason to flatter anyone. A [Peec AI study of 30 million sources](https://searchengineland.com/ai-search-engines-cite-reddit-youtube-and-linkedin-most-study-473138) (March 2026) found Reddit, YouTube, LinkedIn, Wikipedia and Forbes were the most cited domains, and [Profound's 2025 study](https://www.tryprofound.com/blog/ai-platform-citation-patterns) found Reddit made up 46.7% of Perplexity's top ten cited sources. The pattern shifts, though. [Semrush found](https://www.semrush.com/blog/most-cited-domains-ai/) the share of ChatGPT responses citing Reddit fell from about 60% to about 10% between August and mid-September 2025. So treat shared platforms as one route among several. Make sure your own site answers the buyer's question directly, earn coverage that does the same, and where your team takes part in communities, do it openly and helpfully under real names. Planted posts get spotted and damage trust. ## What does your own website still have to do? Be the reference. Your website is where an assistant learns what you do, who for, where you work and who leads the firm. Earned coverage often mentions you in passing; your site fills in the detail. In practice that means one service page per thing you sell, in buyers' words; case studies with named clients where you have permission; team pages showing your experts and where they have been published or quoted; core facts written the same way everywhere; and links to the coverage you have earned. A well-argued article on your site is often what leads to a quote in someone else's, as [what makes ChatGPT treat your content as expert](/insights/what-makes-expert-quality-content) explains. Owned pages can win recommendations on their own when they answer the question better than anyone else. For [Helm Club](/work/helm-club), a founder membership community, "We built the structured data AI engines read, then a comparison page for every question a founder asks when weighing up peer groups: Helm against YPO, Vistage and the rest." Today, "1 in 5 new members this year came through an AI recommendation." ## Where does paid activity help, and where does it not? Paid activity buys attention from people. It rarely buys credibility with AI assistants. Sponsored articles are usually labelled, for readers and for search engines. Google's spam policies say paid links are fine only when marked with a nofollow or sponsored attribute, and treat links in paid articles that pass ranking credit as link spam ([Google Search Central](https://developers.google.com/search/docs/essentials/spam-policies)). Some sponsored pieces stay online for years, but they stay labelled as paid. No assistant has said how it treats them; our inference is that assistants drawing on search results give them the same caution, and sponsored copy reads like marketing, which gives an assistant little to quote. Ads inside AI answers are now real. ChatGPT has shown labelled ads to UK users on its Free and Go plans since 11 August 2026, and OpenAI says "Ads do not influence the answers ChatGPT gives you" ([OpenAI](https://openai.com/index/testing-ads-in-chatgpt/)). You can buy a place next to the answer. Being named in it is still earned. This table maps common spend lines to their bucket and their likely effect on AI answers. | Spend line | Bucket | Likely effect on AI answers | Ask before renewing | |---|---|---|---| | Sponsored feature or advertorial | Paid | Little: labelled and rarely cited | Did it bring enquiries we can name? | | Awards entry and dinner table | Earned if independently judged; paid if every payer wins | A judged shortlist helps once covered; a bought badge adds little | Are the judges named, and do most entrants lose? | | Premium directory listing | Paid placement, with any reviews shared | The placement adds little; independent reviews and research-based rankings add more | Does it send qualified leads? | | Boosted LinkedIn post | Paid boost on an owned post | Reach with people, no added credibility | Did the right people see it? | | PR retainer | Buys earned, if it produces coverage | High when it produces coverage in titles buyers read | What named coverage came from it? | | Clutch or G2 profile | Owned profile, shared reviews, paid for any sponsored slot | Reviews are cited, and buyers trust them | How many detailed, recent reviews do we have? | | Website and content | Owned | The reference for what you do | Does it answer the questions buyers ask? | Paid activity earns its place when it puts you where earned opportunities happen, amplifies coverage you have earned, brings enquiries you can trace, or supports a trade body your clients care about. The mistake is calling it authority. Whether being named beats paying for search ads is set out in [whether being named by ChatGPT is cheaper than Google Ads](/insights/economics-ai-visibility-vs-advertising). ## How do you move budget from paid to earned? Most firms can fund a proper earned programme without increasing total spend, because the money is sitting in paid lines nobody can connect to an enquiry. This is the process we use. ### Step 1: List every reputation line from the last twelve months Pull every invoice meant to build reputation, from sponsorships and award entries to PR retainers and content, and include your team's time. ### Step 2: Put each line in a bucket Mark each one earned, shared, owned or paid, using the test above, and be strict about the grey areas. Seeing the four totals side by side is the point of the exercise. ### Step 3: Ask what each paid line produced For every paid item, ask whether anyone can name an enquiry, a meeting, a client or a piece of coverage that came from it. If nobody can, it is a candidate to cut. ### Step 4: Move the freed money into steady earned activity Earned authority rewards consistency. A regular rhythm of expert comment, contributed articles and relationships with the few journalists who cover your buyers' sector does more over a year than one big campaign. Which coverage matters most now is compared in [digital PR or traditional PR: which gets a firm named by ChatGPT](/insights/digital-pr-vs-traditional-pr). ### Step 5: Bring your own site into line As coverage builds, make sure your website says the same thing, in the same words, and links to the coverage. ## What to do this month 1. List last year's reputation spend and put every line into earned, shared, owned or paid. 2. Pick the two paid items you can least justify and decide whether to renew them. 3. Identify three journalists or trade titles that cover your buyers' sector, and read their last month of articles. 4. Check that your website, LinkedIn page and main directory and review profiles state the same core facts about your firm. 5. Ask five recent clients for a detailed review on the platform your buyers use most. 6. Ask ChatGPT, Gemini and Claude to name the leading providers in your field, and note which sources they cite and where they hedge. ## Frequently asked questions ### What share of our marketing budget should go on earned media rather than ads? There is no right percentage, and we would be wary of anyone who quotes one. Start from what each line does: fund earned activity first with money from paid lines that produce nothing you can name, keep advertising that brings traceable enquiries, and keep investing in your site wherever it cannot yet answer buyers' questions. [How to get your board to approve budget for AI search](/insights/business-case-ai-optimisation-board) shows how to frame the change. ### Do sponsored articles help with SEO or AI visibility at all? A little, and mostly indirectly. Links in them should be marked as sponsored, so they pass little or no ranking credit, and paid or advertorial content made up only 0.3% of the links AI tools cited in Muck Rack's study. If a sponsored piece contains useful facts, an assistant may still draw on them. Judge sponsored content on the enquiries and relationships it brings, not on authority. ### Is a paid profile on Clutch or G2 earned or paid media? Both, in parts. The profile you fill in is owned, any sponsored placement is paid, and the reviews on it are shared, written by clients. The reviews are the valuable part: in [G2's April 2026 survey of 1,076 software buyers](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html), 45% said citations of review sites were the most confidence-inspiring signal in an AI answer. ### Are our founders' LinkedIn posts earned or owned? Posts on your own profiles are owned. The discussion underneath, and shares with other people's comments, are shared, and they become a route to earned coverage when a journalist asks for a quote or an event invites the author to speak. Paying to boost a post moves it into the paid bucket, which helps reach but not credibility. ### How long does earned coverage take to show up in AI answers? Assistants that search the web live, such as ChatGPT search, Perplexity, Copilot and Google's AI features, can reflect a new article within weeks. Answers drawn from a model's own memory change only when it is retrained, which takes months. A steady earned programme keeps improving answers for as long as it runs. ### Should we cancel all our sponsorships? No. Cancel the ones you cannot connect to an outcome, and keep those that put you in front of buyers or give you a speaking role, measured as relationship and reach spend. If you want a clear view of what independent sources say about your firm, and a plan to add to it every month, [Reputation That Wins](/reputation-that-wins) is built for that. A free [Growth Review](/growth-review) will show you where to start. ## Sources - [Muck Rack, Generative Pulse: earned media consistently drives AI citations (May 2026)](https://www.globenewswire.com/news-release/2026/05/07/3290268/0/en/generative-pulse-earned-media-consistently-drives-ai-citations-holding-at-84.html) - [Search Engine Land, AI search engines cite Reddit, YouTube and LinkedIn most: study (March 2026)](https://searchengineland.com/ai-search-engines-cite-reddit-youtube-and-linkedin-most-study-473138) - [Profound, AI platform citation patterns (June 2025)](https://www.tryprofound.com/blog/ai-platform-citation-patterns) - [Semrush, The most-cited domains in AI: a 3-month study (November 2025)](https://www.semrush.com/blog/most-cited-domains-ai/) - [Google Search Central, Spam policies for Google web search (August 2026)](https://developers.google.com/search/docs/essentials/spam-policies) - [OpenAI, Testing ads in ChatGPT (updated August 2026)](https://openai.com/index/testing-ads-in-chatgpt/) - [G2, New G2 research: half of B2B software buyers now start their research with AI chatbots (April 2026)](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html) --- Title: Does our website platform affect whether ChatGPT can read our site? URL: https://fortitudemedia.ai/insights/cms-choice-affects-ai-visibility Author: David Adams Published: 2026-03-31 Last reviewed: 2026-09-24 Topic: Websites that bring work in Short answer: Somewhat, but less than how the site is built. GPTBot, ClaudeBot and PerplexityBot do not run JavaScript, so your text must be in the page's HTML when it is delivered. WordPress, Webflow, HubSpot, Wix and Squarespace can all do that; sites built with JavaScript frameworks or AI builders need server rendering or prerendering. Check robots.txt, your platform's AI crawler switch and Cloudflare, and test what crawlers see before you switch. At some point in the life of every website, someone suggests changing platform. Your developer says WordPress has become a headache. A new agency swears by Webflow. Your sales director wants everything in HubSpot. Someone has built a prototype in an afternoon with an AI site builder. And now there is a new question in the mix: will the choice affect whether ChatGPT, Claude, Gemini and Perplexity can read our site and recommend us? It is a sensible question, because a migration is expensive and disruptive, and you only want to do it once. The platform matters less than most agencies suggest and more than most founders assume. The [HTTP Archive's 2025 Web Almanac](https://almanac.httparchive.org/en/2025/cms) found WordPress runs roughly 64% of sites built on a CMS, yet only 45% of WordPress sites passed Google's Core Web Vitals on mobile, against 74% for Wix. The same platform can be fast or slow, readable or not, depending on how it is built. This guide explains what AI assistants need from your website, how the common platforms compare, what to check on sites made with AI builders, how to test whether your current site is holding you back, and when switching is worth it. ## Key takeaways - Your website platform matters less than how the site is built, but a poor set-up can hide your content from the crawlers AI assistants rely on. - GPTBot, ClaudeBot and PerplexityBot do not run JavaScript, so your main text must be in the page's HTML when it is first delivered; Googlebot does run it, which helps Gemini and Google's AI features. - Since 1 July 2025 Cloudflare has blocked AI crawlers by default on new domains, and Webflow and Squarespace have their own AI crawler switches, so check the settings. - WordPress, Webflow, HubSpot, Wix and Squarespace can all deliver readable pages; they differ in control over robots.txt, structured data and code. - Sites made with AI builders or modern JavaScript frameworks need server rendering or prerendering, and you should test what a crawler receives before launch. - Before switching platforms, check what your current site shows crawlers and time a simple edit, because many problems can be fixed without moving. ## What is a CMS, and why does it matter for AI search? A CMS, or content management system, is the software your team uses to create and edit website pages without writing code. WordPress, Webflow, HubSpot, Framer, Squarespace and Wix are all examples. It matters for AI search in two ways. First, the CMS, together with how your developer set it up, decides what code a crawler receives when it asks for a page. If that code contains your words, headings and key facts, the assistants can use them. If the words only appear after scripts have run, or are locked inside images, the main AI crawlers see very little. Second, the CMS decides how easy it is for your team to keep the site current. A platform that makes updates slow tends to produce an out-of-date site, and out-of-date sites produce out-of-date AI answers. ## What do AI assistants need from your website? Before comparing platforms, be clear about what you are asking the platform to deliver. ### Your words in the page itself Your main text, headings and key facts should be in the page's code when it is first delivered, not added afterwards by scripts in the visitor's browser. [Vercel and MERJ found in December 2024](https://vercel.com/blog/the-rise-of-the-ai-crawler) that GPTBot, ClaudeBot and PerplexityBot fetch JavaScript files but do not run them, so content that depends on scripts is invisible to them. Googlebot renders JavaScript, which helps Gemini and Google's AI features, but that does not help with ChatGPT, Claude or Perplexity. We explain how the crawlers differ in [how AI crawlers read your website](/insights/ai-crawlers-vs-google-spiders). ### A clear structure One main heading per page, sensible subheadings beneath it, and one clear topic per page. Headings should be real headings in the code, not ordinary text styled to look large. ### Accurate page titles and descriptions Every important page should have its own title and short description that say what the page is about. ### Structured data Structured data is a small block of labelled information in a page's code that tells machines facts such as your firm's name, address, services and the author of an article. It helps search engines read your facts correctly, and you want a platform that lets you add it. Our guide to [whether you need schema markup](/insights/structured-data-schema-markup-ai) explains what is worth adding. ### Access for AI crawlers Your robots.txt file, a short text file that tells crawlers which parts of the site they may visit, should not block the crawlers of the assistants you want to be named by. Check your security and hosting settings too. [Cloudflare changed its default on 1 July 2025](https://www.cloudflare.com/press-releases/2025/cloudflare-just-changed-how-ai-crawlers-scrape-the-internet-at-large/): new domains now start with AI crawlers blocked unless the owner chooses to allow them, and many firms never realise the setting is on. ### Easy editing Your team should be able to update a service page, add a case study or change a team profile in minutes, without calling a developer. This is the requirement most often forgotten when platforms are chosen. ## How do the common platforms compare? Each of the main platforms can meet those needs. They differ in how much control you have and where the risks lie. The table reflects each platform's help pages where we could check them in September 2026; features change often, and a lot depends on how your particular site is built. For running and build costs, see [how much a B2B website costs in the UK](/insights/real-cost-of-a-website). | Platform | Text in the page by default | Edit robots.txt | Custom structured data | AI crawler switch | |---|---|---|---|---| | WordPress | Yes, pages are built on the server | Yes, by file or plugin | Yes, by plugin or by hand | Through robots.txt or plugins | | Webflow | Yes | Yes, in site settings | Yes, through custom code on a paid plan | Yes, traffic control toggles for named AI bots | | HubSpot CMS | Yes, depending on the template | Yes, in settings | Yes, through head code or templates | Through robots.txt | | Framer | Generally yes, but test yours | Check current settings | Yes, through custom code | Check current settings | | Squarespace | Yes | No direct editing | Limited, through code injection | Yes, "Block known artificial intelligence crawlers" | | Wix | Yes | Yes, as an advanced feature | Yes, up to five JSON-LD blocks per page | Through robots.txt | | Headless or custom build | Only if pages are server-rendered or generated in advance | Yes | Yes | Yes | ### WordPress Self-hosted WordPress gives you the most control. You can edit almost anything, add structured data through plugins such as Yoast SEO or Rank Math or by hand, and move your content elsewhere later. The risks come from that same freedom: heavy page builders, too many plugins and poor hosting make many WordPress sites slow, and someone has to keep the software updated for security. ### Webflow Webflow is a hosted visual builder popular with design-led agencies. It handles hosting and security, lets editors manage articles and case studies through content collections, and its [site settings](https://help.webflow.com/hc/en-us/articles/33961368603539-Disable-search-engine-indexing) let you edit robots.txt and switch named AI bots such as GPTBot, ClaudeBot and PerplexityBot on or off. The trade-offs are cost, reliance on one company and the work involved in moving away later. ### HubSpot CMS Many B2B firms already use HubSpot for contacts and pipeline, and its website platform connects directly to that, which makes it easier to see which pages lead to enquiries. It is hosted, so updates and security are handled for you. How well a HubSpot site reads to crawlers depends on the template, so run the checks below. ### Squarespace and Wix These are the simplest to use, and they handle hosting, security and mobile layouts for you. They are more capable than they were: [Wix lets you edit robots.txt](https://support.wix.com/en/article/editing-your-sites-robotstxt-file) and [add your own JSON-LD structured data](https://support.wix.com/en/article/adding-structured-data-markup-to-your-sites-pages), and [Squarespace has a setting to block known AI crawlers](https://support.squarespace.com/hc/en-us/articles/360022347072), which you should check is off if you want to be read. The limits appear as the site grows: less control over code, and templates that make detailed service and sector pages harder. We look at the wider trade-offs in [template or custom website: which does a B2B firm need?](/insights/template-vs-custom-website-b2b). ### Headless and custom builds A headless set-up stores content in one system and has developers build the website separately, often with a JavaScript framework such as React. It can be fast and flexible. The main risk is a site that assembles its pages in the visitor's browser, which leaves the main AI crawlers looking at an almost empty page. Insist that pages are rendered on the server or generated in advance. ## Can ChatGPT read a website built with Lovable, Framer or another AI site builder? Usually, if the builder delivers full HTML, and you should check before launch. Framer sites generally arrive as HTML, but test yours. Lovable's [documentation](https://docs.lovable.dev/tips-tricks/seo-geo) says apps created from 13 May 2026 use server-side rendering, while older React and Vite apps get pre-rendered pages served only to verified search and AI crawlers, and other tools see the ordinary single-page app. Other AI builders, such as Bolt and v0, generate code in various frameworks, and whether a page arrives as full HTML depends on the framework and settings used. Treat any site made this way like a custom React build: confirm it is server-rendered or prerendered, then test what a crawler receives, as below. ## How do I check whether my React website is readable by AI crawlers? Look at the HTML the server sends before any script runs. If your headings and body text are in it, AI crawlers can read the page; if you see an almost empty page with a script tag, they cannot. Three simple checks: open the page, right-click and choose "View page source", then search for a sentence from the page; turn off JavaScript in your browser's developer tools and reload; or ask your developer to fetch the page from the command line with a crawler's user agent. Do this for the homepage, a service page and an article, because different templates can behave differently. ## How to check whether your platform is holding you back You do not need to switch platforms to find out whether yours is a problem. These checks take a morning. ### Step 1: See what a crawler sees Use the page source check above on your homepage and one service page. If the source is mostly code with little of your text in it, that is the first thing to fix. ### Step 2: Check your crawler settings Open your robots.txt file by adding /robots.txt to your web address, and look for lines that block GPTBot, OAI-SearchBot, ClaudeBot, Claude-SearchBot or PerplexityBot. Then check your platform's AI crawler setting and any Cloudflare or hosting setting. Decide deliberately which crawlers you allow. ### Step 3: Check headings and titles On your five most important pages, check that each has one main heading, clear subheadings and its own page title and description. ### Step 4: Check whether you can add structured data Ask your developer whether the platform lets you add structured data for your organisation, services and articles, and whether any is there already. ### Step 5: Time a simple edit Ask the person who manages the site to update a sentence on a service page and add a new team member. If it takes more than a few minutes or needs a developer, the site will struggle to stay current. For a fuller list, see [the technical checklist for an AI-ready website](/insights/technical-checklist-ai-ready-website). ## When is it worth switching platforms? Switch when the platform stops you fixing problems, not when a new agency prefers a different tool. Moving platform almost always changes page addresses and puts search history, content and senior time at risk. Good reasons include: - Your team cannot update pages without a developer, and the platform cannot be set up to change that. - Your main content cannot be delivered in the page itself on the current set-up. - You cannot control crawler access, page titles or structured data at all. - The platform, or your version of it, is no longer supported, or the site is too slow to fix without rebuilding. If you do switch, list every existing page, redirect every old address to its new one, and check afterwards that nothing was dropped. When we rebuilt [arum](/work/arum)'s site, "Every old address redirected, so years of rankings carried over." We discuss when a rebuild is justified in [whether to redesign your website or keep improving it](/insights/website-redesign-vs-continuous-iteration). ## What matters more than the platform? What your site says, and whether the rest of the web agrees with it. A perfectly built site will not be recommended if it describes your firm vaguely or contradicts your LinkedIn page and directory listings. Many fixes need no migration at all. For [ID-Pal](/work/id-pal), the fix was "About four hours of structured data work declaring NorthRow part of ID-Pal on both domains. Live in week one, with no copy changes and no risk to the redirect." In our experience, most firms that are not named by AI assistants have a content and clarity problem, not a platform problem. We look at how structure affects visitors and AI in [how to organise your website so buyers and AI understand it](/insights/website-architecture-ux-and-ai). ## What to do this month 1. Check the page source of your homepage and main service page for your actual text. 2. Check your robots.txt file, your platform's AI crawler setting and any Cloudflare setting, and allow the crawlers you want to be read by. 3. Fix headings and titles on your top five pages. 4. Time a simple edit, and if updating a page needs a developer, fix that before anything else. 5. If your site was made with an AI builder or a JavaScript framework, confirm it is server-rendered or prerendered. 6. Decide on switching only after these checks. ## Frequently asked questions ### Is Webflow or WordPress better for SEO and AI search in 2026? Neither is better in itself. WordPress gives more control and more structured data options but depends heavily on how well it is built and maintained. Webflow needs less upkeep and has built-in AI crawler controls, but ties you to one company. Both deliver pages as HTML that AI crawlers can read, so choose the one your team can edit easily and your developer can support properly. ### Does HubSpot CMS let us add custom schema markup? Yes. Structured data in JSON-LD format can be added to a page's head code or built into the templates your developer creates, so organisation, service and article markup are all possible. Check what the template already outputs first, so you do not end up with two conflicting versions. ### Can a Squarespace or Wix site be recommended by ChatGPT? Yes. What an assistant says about you depends mainly on what your site says and what other sources say about you, and a clear, accurate Squarespace or Wix site can be read well. For a small, simple firm, it may be enough. ### How much does it cost to move a website from WordPress to Webflow? It depends far more on the site than on the platforms. The main drivers are the number of pages and templates, how many articles and case studies must move, redirects for every old address, forms and integrations, and who rewrites the content. We do not quote prices here: ask for a fixed quote that lists each of those items, and see [how much a B2B website costs in the UK](/insights/real-cost-of-a-website) for published market ranges. ### Should we block AI crawlers to protect our content? That is a business decision, but for most B2B service firms the answer is no. If you block the crawlers ChatGPT, Claude and Perplexity use, those assistants have less of your own material to work from. Some firms block training crawlers such as GPTBot and ClaudeBot but allow search crawlers such as OAI-SearchBot and Claude-SearchBot. Whatever you choose, choose it deliberately. The right platform is the one that shows your words clearly to crawlers and lets your team keep them current without a fight. To have a site built with both in mind, see [Website That Sells](/website-that-sells), or book a free [Growth Review](/growth-review) to find out what crawlers and assistants see today. ## Sources - [HTTP Archive, Web Almanac 2025: CMS (January 2026)](https://almanac.httparchive.org/en/2025/cms) - [Vercel, The rise of the AI crawler (December 2024)](https://vercel.com/blog/the-rise-of-the-ai-crawler) - [Cloudflare, Cloudflare just changed how AI crawlers scrape the internet at large (July 2025)](https://www.cloudflare.com/press-releases/2025/cloudflare-just-changed-how-ai-crawlers-scrape-the-internet-at-large/) - [Lovable documentation, search and AI visibility best practices (accessed September 2026)](https://docs.lovable.dev/tips-tricks/seo-geo) - [Webflow Help Center, Disable search engine indexing (accessed September 2026)](https://help.webflow.com/hc/en-us/articles/33961368603539-Disable-search-engine-indexing) - [Wix Help Center, Editing your site's robots.txt file (accessed September 2026)](https://support.wix.com/en/article/editing-your-sites-robotstxt-file) - [Wix Help Center, Adding structured data markup to your site's pages (accessed September 2026)](https://support.wix.com/en/article/adding-structured-data-markup-to-your-sites-pages) - [Squarespace Help Center, Request that AI models exclude your site (accessed September 2026)](https://support.squarespace.com/hc/en-us/articles/360022347072) --- Title: Does publishing our own research help ChatGPT recommend us? URL: https://fortitudemedia.ai/insights/original-research-data-ai-trust Author: Ashley Kenerson Published: 2026-03-31 Last reviewed: 2026-09-24 Topic: Content that gets quoted Short answer: Yes, usually. A specific figure only your firm has published gives an assistant a reason to name you, and a Princeton and IIT Delhi study found adding statistics raised visibility in AI answers by about 34%. It is not guaranteed: assistants often credit whoever repeated the figure. Publish each study on one findings page, state the method and margin of error, write the headline figure with your firm's name in it, and get it covered by the trade press. Your firm has been publishing for years. There are articles on the website, a newsletter and a few LinkedIn posts a week from the founders. It is decent work. And yet when a buyer asks ChatGPT or Perplexity a question in your field, the answer quotes a trade body, a large consultancy and a software vendor, and you are nowhere. The difference is usually not the writing. Those three organisations publish numbers of their own, and when an assistant wants to say something specific, it needs a source it can name. In a [Princeton and IIT Delhi study presented at KDD 2024](https://arxiv.org/abs/2311.09735), adding statistics to a page raised its visibility in AI-generated answers by about 34%, and adding quotations from relevant people by about 41%. It measured adding figures to existing pages rather than publishing new research, and it ran on a research engine, so treat it as a direction. But the direction is clear: specific numbers get used. This guide explains why assistants treat original data differently from commentary, what counts as research for a founder-led firm, how many respondents you need, how to keep the credit when others repeat your finding, and how to run one credible study a quarter. Most of it needs a spreadsheet you already have. ## Key takeaways - A figure only you have published gives an assistant a reason to name you, but it does not guarantee the credit: assistants often cite the site that repeated the figure, or state it with no source. - Original data does not mean academic studies; your own job records, client outcomes, a short survey or a new reading of public data all count. - Fifty respondents gives a margin of error of about 14 points either way, so a small survey supports big differences and clear majorities, not decimal points. - The findings that get quoted are specific, a little surprising, dated and published with a plain description of how you got them. - Put every study on one findings page on your own site, write each headline figure as a sentence that includes your firm's name, and ask anyone who reports it to link there. - One solid piece of research a quarter, repeated yearly, builds the track record that makes your firm the obvious source on its subject. ## Why do AI assistants reach for original data? Because a specific finding cannot be passed off as common knowledge. An assistant can say "outsourcing payroll reduces admin" in its own words without naming anyone, but a finding only one firm has published needs a source. For example, suppose a payroll firm publishes: "in our review of 140 payroll transfers, most delays were caused by missing pension data, not by the new provider". An assistant that uses that point has a reason to say where it came from. It is a reason to name you, not a promise. In our experience assistants often credit the news site or blog that repeated a figure, or state the number with no source at all, which is why the section below on keeping the credit matters as much as the research itself. There is a second effect. Google's own [guide to its AI search features](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide), last updated in July 2026, says content with first-hand expertise that could not come from anyone else counts for more than commodity content. We explain the wider idea in [what E-E-A-T is and why AI assistants care about it](/insights/eeat-why-ai-cares-more). Buyers respond the same way. In the [2025 Edelman and LinkedIn B2B Thought Leadership Impact Report](https://www.edelman.com/expertise/Business-Marketing/2025-b2b-thought-leadership-report), a survey of US executives, 64% of "hidden" decision makers such as finance and procurement said they trust thought leadership more than marketing materials when judging a supplier's capabilities. ## What kind of research do AI assistants cite most? Research that other people have reported, that is recent and that carries specific numbers. Your own study is the start; coverage of it is what spreads it. [Muck Rack's May 2026 analysis](https://www.globenewswire.com/news-release/2026/05/07/3290268/0/en/generative-pulse-earned-media-consistently-drives-ai-citations-holding-at-84.html) of more than 25 million links cited by ChatGPT, Claude and Gemini found that earned media, meaning coverage by others, made up 84% of cited links and paid content 0.3%. More than half of the journalism cited had been published in the previous 12 months, and citations dropped sharply after six months. [Profound's 2025 study of 680 million citations](https://www.tryprofound.com/blog/ai-platform-citation-patterns) found the analyst firm Gartner among Perplexity's most cited sources and the review site G2 among ChatGPT's. So publish the finding on your own site, get it into the trade press your buyers read, and refresh it every year. How to pitch it is covered in [turning your own data into trade press coverage](/insights/leverage-data-research-press-coverage). ## What counts as research for a firm your size? Far more than most founders assume. Research means a new finding with a clear method, not a five-hundred-person survey or a peer-reviewed paper. For most founder-led B2B firms the richest source is their own operations. | Method | What it looks like | Typical effort | Best for | |---|---|---|---| | Your own records | Analysis of jobs, projects, cases or support tickets you already hold | A week of someone's time and a day with a founder | Findings nobody else can publish | | A short survey | Eight to ten plain questions to the people you serve | Three weeks to gather, a week to analyse, a week to write | What buyers expect, plan or worry about | | Public data, read in a new way | Job adverts, planning applications, published accounts, regulator data | One to two weeks | Market trends where you add the reading | | Structured expert interviews | Ten recognised people asked the same questions | Two to four weeks | Views on where a field is going | | An annual benchmark | The same measure, repeated every year | The first edition's effort, then less | Trend data that becomes the reference | ### Your own records If you deliver work, you have data. For example, a commercial fit-out contractor with two hundred completed projects knows the average overrun, the most common cause of delay and the difference between landlord-led and tenant-led jobs. "Across 200 office fit-outs since 2020, jobs where the landlord's consent arrived after design sign-off took a third longer on average" is a sentence an assistant can quote, and nobody else can publish it. ### A short survey of the people you serve A recruitment firm that asks sixty finance directors what they expect from a search partner has a finding that a general survey of a thousand office workers does not. Ask eight to ten questions, keep them plain, and include at least one where the answer might surprise you. Be honest about what the sample size supports, which the next section sets out. ### Public data, expert views and benchmarks The data can be public if the analysis is new: assembling and reading it is what makes the result yours. Ten structured interviews, asked the same questions and written up with care, is research too. And the most valuable study is the one you repeat: after three editions of an annual benchmark, a buyer asking how your field is changing has one obvious source. ## How many survey respondents do we need for credible B2B research? Fifty to eighty well-chosen respondents is enough for a credible B2B study, as long as you publish the margin of error and only claim what it supports. At fifty, a result of 60% could really be anywhere from about 46% to 74%. The margin of error shrinks slowly as the sample grows. These figures use the standard formula for a simple random sample at the 95% confidence level, the basis [Pew Research Center](https://www.pewresearch.org/short-reads/2016/09/08/understanding-the-margin-of-error-in-election-polls/) uses when it says a random sample of 1,067 gives a margin of about 3 points. | Respondents | Margin of error (95% confidence) | What you can fairly claim | |---|---|---| | 50 | About 14 points either way | Clear majorities and large differences only | | 80 | About 11 points either way | Clear majorities and large differences | | 150 | About 8 points either way | Most headline findings; avoid small subgroups | | 400 | About 5 points either way | Comparisons between groups, such as sector or firm size | Two cautions. First, these figures assume a random sample; a survey of your own clients is not random, so treat the margin as the best case and say who you asked. Second, subgroups have their own, wider margins: if 50 respondents split into three sectors, each sector's result rests on fewer than 20 people on average. Report "about half" rather than "48.7%", and let the sample decide how bold the headline can be. ## What makes a finding quotable? A quotable finding is specific, attributable and dated. It tells the reader who found what, among whom and when, in one sentence. A live example: [Helm Club](/work/helm-club), a membership community for founders, can say "1 in 5 new members this year came through an AI recommendation." It is one number, it belongs to one organisation, and it says exactly what was measured. That is the shape to aim for. - Be specific. "Most delays are caused by missing pension data" is quotable. "Delays have many causes" is not. - Be slightly surprising. A finding that confirms the obvious is rarely repeated. - Lead to an action. "Firms that agreed a governance owner before choosing software finished sooner" tells the reader what to do. - Carry the method. "Based on 140 transfers we managed between 2023 and 2025 for firms with 50 to 500 staff" is credible. "Based on our experience" is not a source. - Sit on a page an assistant can read cleanly, with the finding near the top and the method close by. The structure is set out in [what an article that ChatGPT actually quotes looks like](/insights/anatomy-of-article-cited-by-ai). Avoid any number bigger than the evidence behind it. ## How do we make sure ChatGPT credits our data and not the site that republished it? You cannot force it, but you can make your firm's name travel with the figure and make your page the easiest one to cite. Five habits do most of the work. ### Publish one canonical findings page first Put every study on its own page on your site, at a permanent address, before the press release or LinkedIn post goes out: findings at the top, then method, sample, dates, limits and a named contact. ### Write each headline figure as a sentence that includes your name Example: "Across 200 office fit-outs since 2020, Example Ltd found that jobs with late landlord consent took a third longer." When a journalist or a blog lifts that sentence, the name comes with it. A bare "a third longer" travels without you. ### Put the figures in text, not only in charts or a PDF Google does [index text-based PDFs](https://developers.google.com/search/docs/crawling-indexing/indexable-file-types), but a web page is easier to cite, link to and update. Keep the key numbers in the page text, with each chart described in a sentence. ### Add Dataset markup If you publish the underlying table, [Dataset structured data](https://developers.google.com/search/docs/appearance/structured-data/dataset) tells Google its name, description, creator and formats, and makes it findable in Google's Dataset Search. It helps search engines read your facts correctly; OpenAI and Anthropic have not said they read it, so treat it as good housekeeping rather than a ranking trick. See [whether we need schema markup for AI assistants](/insights/structured-data-schema-markup-ai). ### Ask everyone who reports it to link to the findings page Give trade titles, partners and associations the findings page address and ask them to link to it rather than to your home page or the PDF. When coverage appears, ask politely for a link where one is missing. ## How to run a light research programme The aim is one solid piece a quarter, not a study a week. ### Step 1: Pick four questions buyers actually ask Choose four questions your sales team hears and competitors cannot answer with data, one per quarter, where you have the data or could gather it in a few weeks. [Building content around the questions customers ask](/insights/content-around-customer-questions) is the place to start if the list is not obvious. ### Step 2: Match each question to the cheapest honest method Rotate through the methods in the table: records, survey, public data, interviews. Varying the method stops any single dataset carrying the whole year. ### Step 3: Do the work in a fixed window Put a four-week window in the diary, with a named owner, or it will not happen. The effort for each method is in the table above. ### Step 4: Write the findings page, then the shorter versions The findings page carries the results, the method, the limits and what the reader should do. Then produce a one-page summary, a chart or two, three short articles that each pick out a finding, and a pitch to your trade title with the most surprising number in the first line. ### Step 5: Date it, and promise the next one Put the year in the title and say you will run it again. When the next edition comes, link back and describe what changed. Recent findings are cited more, so refreshing an existing study is often better value than starting a new one; see [how old is too old for AI assistants](/insights/content-freshness-recency-ai). ## How do we make research credible? State the method plainly: who, how many, when, how chosen and how analysed. Acknowledge the limits: if the sample is your own clients, say so and say what that might bias. Credit anything you built on, for example "this extends the trade body's 2024 survey". And name a person who will answer questions about the data. A small sample with an honest description beats a large sample with none. "We asked 60 finance directors at UK firms in March 2026; the margin of error is about 13 points" is a source. "A survey found" is not. ## What to do this month 1. List the ten questions your sales team hears most, and mark the three where a specific number would change the conversation. 2. Open the records you already hold, such as your job list, CRM, support log or timesheets, and see which of those three questions the data could already answer. 3. Draft one finding as a single sentence that includes your firm's name, the sample and the date. 4. Create the findings page template: headline findings, method, sample, margin of error, limits and a named contact. 5. Book a four-week window to produce the first study, with a named owner and a date for the trade press pitch, and book the next quarter's window now. ## Frequently asked questions ### How much does it cost to run a B2B survey in the UK? Cost depends mostly on who you need to reach. Surveying your own clients and contacts costs staff time. Buying responses from a research panel costs more, because senior B2B respondents are scarce, so ask panel providers for a price per completed response for your exact audience before you commit. An analysis of your own records is usually the cheapest credible option. Our own fees are on [what it costs](/what-it-costs). ### Can we publish anonymised client data under UK GDPR? Yes, if it is anonymised properly. The [ICO says](https://ico.org.uk/for-organisations/uk-gdpr-guidance-and-resources/data-sharing/anonymisation/introduction-to-anonymisation/) data protection law does not apply to anonymous information, but pseudonymised data, where someone could still be identified using other information, is still personal data. Small numbers and specific details can make people or firms recognisable, so publish totals and averages across enough cases, check your client contracts for confidentiality terms, and take advice if in doubt. ### Do we need a designer or a PDF? No. A clear web page with the findings near the top, a simple chart and the method underneath is more useful. Google indexes text-based PDFs, but a page is easier to cite, link to and update. Produce a designed PDF for people who want one, and publish the page first. ### What if our findings are not surprising? Publish them anyway if they are specific and useful. A confirmatory finding still gives an assistant a number to attach to your name. Then look harder at the data for the one result that is not what people assume, and make it the headline. ### Can we outsource the research? Yes, an analyst or research firm can gather and analyse the data, and the findings are yours to publish if your agreement says so. Keep a founder in the room for the judgement about which questions matter and what the results mean for a buyer. ### How is this different from a white paper? A white paper is usually an argument dressed as a report. Research is a finding with a method someone could check. If your white paper contains a number nobody else has and explains where it came from, it is research; if it contains a point of view and borrowed statistics, it is not. If you want help choosing the first study and getting it into the trade press, [Reputation That Wins](/reputation-that-wins) is built around this work. To find out where your data already gives you an edge, start with a [free Growth Review](/growth-review): 30 minutes with David Adams and three fixes in writing. ## Sources - [Aggarwal and others, Princeton and IIT Delhi, study of what raises visibility in AI-generated answers (KDD 2024)](https://arxiv.org/abs/2311.09735) - [Muck Rack, What is AI reading? (May 2026)](https://www.globenewswire.com/news-release/2026/05/07/3290268/0/en/generative-pulse-earned-media-consistently-drives-ai-citations-holding-at-84.html) - [Profound, AI platform citation patterns (June 2025)](https://www.tryprofound.com/blog/ai-platform-citation-patterns) - [Edelman and LinkedIn, 2025 B2B Thought Leadership Impact Report (June 2025)](https://www.edelman.com/expertise/Business-Marketing/2025-b2b-thought-leadership-report) - [Pew Research Center, Understanding the margin of error in election polls (September 2016)](https://www.pewresearch.org/short-reads/2016/09/08/understanding-the-margin-of-error-in-election-polls/) - [Google Search Central, Dataset structured data (September 2026)](https://developers.google.com/search/docs/appearance/structured-data/dataset) - [ICO, Introduction to anonymisation (April 2025)](https://ico.org.uk/for-organisations/uk-gdpr-guidance-and-resources/data-sharing/anonymisation/introduction-to-anonymisation/) - [Google Search Central, Optimizing your website for generative AI features on Google Search (July 2026)](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide) --- Title: How can a private equity firm tell whether ChatGPT recommends its portfolio companies? URL: https://fortitudemedia.ai/insights/ai-optimisation-private-equity-vc Author: Ross Williams Published: 2026-03-31 Last reviewed: 2026-09-24 Topic: Sector guides Short answer: Run a fixed audit. For each portfolio company, write 20 to 50 questions its buyers would type, run each several times a month in ChatGPT, Google, Copilot and Gemini in a clean session, and count a win only when it repeats. Record share of answers against named competitors, whether the description is right and the main cause of any gap. Report it on one page to the board. Buy-and-build platforms should check first whether acquired brands are still named as rivals. The value creation plan for a portfolio company usually assumes more inbound demand: a stronger brand, a better website, a sales team with more to work with. Meanwhile the company's buyers have started asking ChatGPT, Copilot and Google's AI answers which suppliers to shortlist. If the assistant names three competitors and not your company, or names a brand you acquired and retired two years ago as a rival, part of that plan is failing where nobody on the board can see it. This matters because the shortlist now forms early. In [6sense's 2025 Buyer Experience Report](https://6sense.com/newsroom/the-timeline-for-influencing-b2b-buyers-is-shrinking-insights-from-6senses-2025-buyer-experience-report/), buying groups had already ranked their preferred vendors before first contact in 94% of cases, and bought from that first choice 77% of the time. In [G2's April 2026 survey of 1,076 software buyers](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html), 51% said they now start research in an AI chatbot. And the businesses private capital backs are mostly the kind of firm this affects: [UK Private Capital](https://www.ukprivatecapital.co.uk/about-us.html) says the industry backs 13,000 UK businesses, nine in ten of them small or medium-sized. This guide is for operating partners and portfolio chief executives who want a plain way to check whether AI assistants recommend their companies, to report it to the board, and to fix what is missing. A shorter section at the end covers the fund's own visibility when founders ask assistants for investors. ## Key takeaways - Buyers now ask AI assistants for a shortlist before they contact anyone, so whether a portfolio company is named is a pipeline question, not a marketing detail. - The fastest way to know is a fixed audit: 20 to 50 buyer questions per company, run several times a month in a clean session, with a win counted only when it repeats. - Buy-and-build platforms have a particular problem: acquired brands, redirects and group names that assistants cannot connect, so they name the pieces or a rival. - Report share of answers against named competitors on one page per quarter, alongside the cause and the next fix, so the board can track it like any other plan metric. - Nobody can guarantee what an assistant says, but a provider can guarantee the work, the reporting and outcomes it controls, such as meetings held. - The fund faces the same test when founders ask which investors back businesses like theirs, and most fund websites give an assistant too little to match. ## Why should an operating partner care what ChatGPT says? Because it shapes who gets the meeting. When a buyer asks an assistant which firms to consider, the answer is a shortlist, and the 6sense figures above suggest most deals go to a firm already on it. A portfolio company that is not named loses deals it never knew about, and the cost shows up only as slower pipeline growth than the plan assumed. There is a second reason. Assistants now describe companies to anyone who asks: prospective customers, candidates and, at exit, anyone researching the business. A company described wrongly, or described but never recommended, carries that into every conversation. We saw this with [ID-Pal](/work/id-pal), an identity verification firm: "AI could describe ID-Pal. It never recommended it." ## Why do good portfolio companies go missing from AI answers? Usually for reasons that come with ownership changes, not with the quality of the business. These are the patterns we see most in PE-backed and founder-led B2B firms: | Cause | What the assistant sees | What fixes it | |---|---|---| | Acquired brands folded in | Two or more names with no statement that they are one company, so the old brand may be named as a rival | Structured data and pages that declare the relationship on every domain | | Group and brand confusion | A group name, several trading brands and a parent fund, each described differently | One agreed description, repeated on the group site, each brand site and every listing | | Rebrand after the deal | Years of press and reviews under the old name, little under the new one | Redirects plus clear "formerly known as" wording, and updated directory profiles | | Thin website | A logo wall and a contact form, with nothing to match to a buyer's question | Pages that answer the questions buyers ask, one question per page | | Founder-led sales | Expertise that lives in the founder's head and sales calls, not in public | Named experts publishing the answers they give on calls | The buy-and-build pattern is the one we would check first. When ID-Pal bought NorthRow and folded it into id-pal.com, "The redirect from northrow.com was technically correct, but nothing on either site told Google or the AI engines that NorthRow and ID-Pal were the same company." Tracking showed both the cost and the recovery once it was fixed: "AI answers still recommending NorthRow as a rival: six in May, one in September 2026." [TCC Group](/work/tcc-group), a compliance group with three brands, had the group version of the same problem: "Three brands, one group. AI could only see the pieces." ## How do you audit a portfolio's AI visibility? Use a fixed method, so the numbers mean the same thing each quarter. Here is the template we would give an operating partner. ### Step 1: Write the buyer questions for each company For each portfolio company, write 20 to 50 questions its buyers would type, in their words. Cover four kinds: category questions ("best payroll outsourcing firms for UK retailers"), problem questions ("how do we cut debtor days"), comparison questions ("X or Y for mid-sized manufacturers") and checking questions ("is [company] any good?"). Include the region and sector where buyers would. Agree the list with the company's chief executive and head of sales, then fix it for at least a year. ### Step 2: Name the competitors to measure against List the three to five competitors the sales team meets most often, plus any acquired or retired brands. You are measuring share of answers against these names, and whether an old brand still appears as a rival. ### Step 3: Run each question in a clean session, several times a month Start with ChatGPT and Google's AI Overviews and AI Mode, then Microsoft Copilot and Gemini, then Perplexity and Claude. Copilot matters in B2B: [Microsoft 365 Copilot passed 30 million paid business seats](https://www.microsoft.com/en-us/investor/earnings/fy-2026-q4/press-release-webcast) in July 2026, and many buyers use it at work. Answers vary from run to run, so run each question several times and count a win only when it repeats. ### Step 4: Record the same fields every time | Field | What it records | How to count it | |---|---|---| | Named | Whether the company is recommended, not only mentioned | Wins that repeat across runs, per question | | Share of answers | How often the company is named against each competitor | Repeat wins divided by questions tracked | | Description | Whether the assistant describes the company correctly | Correct, partly correct or wrong, with the error noted | | Old brands | Whether an acquired or former name is still named, and how | Count of answers naming the old brand as a rival | | Sources cited | Which pages and sites the answer leans on | The three most frequent, per company | | Main cause | Why the company is missing, from the table above | One cause per company, the largest first | ### Step 5: Put it on one page for the board One row per company: share of answers now and last quarter, whether the description is correct, the main cause, the next fix, who owns it and when it will be done. Keep losses on the page as well as wins. ID-Pal's measurement followed the same rule: "The same buyer questions run every month in ChatGPT, Gemini and Claude, in a fresh chat. A win only counts when it repeats, and losses are reported too." ## Which fixes move the numbers, and how fast? Start with the cheapest fixes to the largest causes. Assistants that search the web live, such as ChatGPT search, Copilot and Google's AI features, can reflect a new page or a corrected listing within weeks. Answers drawn from a model's own memory change only when it is retrained, which takes months. | Fix | Cause it addresses | When it can show | |---|---|---| | Structured data declaring acquired brands part of the company | Acquired brands named as rivals | Weeks, in assistants that search live | | Consistent descriptions across group, brand and directory pages | Group and brand confusion | Weeks to a few months | | Pages answering each buyer question, with the answer first | Thin website, founder-led sales | Weeks for live search, months for memory | | Coverage, reviews and awards in the buyer's own sector | Too little independent evidence | Months, and it compounds | The ID-Pal fix shows how small the first step can be: "About four hours of structured data work declaring NorthRow part of ID-Pal on both domains. Live in week one, with no copy changes and no risk to the redirect." The dated result: "UK buyer questions where AI names ID-Pal: none in May, two in August, four in September 2026." And the pattern that tells you what to build next: "Where a page has been rebuilt, ID-Pal appears. Where it has not yet, it does not, which tells us exactly what to build next." ## How can portfolio companies get recommended to their buyers? By giving assistants specific, checkable evidence that matches the questions buyers ask. In practice that means a website organised around those questions rather than internal teams, named experts, consistent facts across every listing, reviews on the sites buyers read, and coverage in the trade press the buyers trust. Our sector guides go deeper for [B2B technology companies](/insights/ai-optimisation-technology-companies) and [consultancies](/insights/ai-optimisation-management-consultancies), and [how to get ChatGPT to recommend your firm](/insights/what-is-llm-optimisation) covers the general method. Being named is one channel. [Researched outbound](/pipeline-that-fills), where a person writes and approves every message, reaches buyers before any assistant is asked, and works alongside it. When a portfolio board asks for the business case, our guide to [getting board approval for AI search budget](/insights/business-case-ai-optimisation-board) sets out the argument. ## The fund's own visibility: when founders ask for investors The same test applies to the fund. Founders and advisers ask assistants questions such as "which private equity firms back founder-led B2B services businesses with £1m to £3m EBITDA in the UK?" and "which PE firms are known for being founder-friendly in the lower mid-market?". An assistant can only answer with firms whose criteria and record it can read. - **Criteria in plain text.** Sectors, EBITDA range, equity cheque size, deal types, regions and what you do not do, on one page, not in a graphic. - **A portfolio with a paragraph per company.** What it does, when you invested, the deal type and what has happened since, agreed with each company. - **Founders' own accounts.** Founder-friendly is a claim assistants check against what founders say in the press, on podcasts and on their own sites. - **Independent records that agree.** Claim and correct your profiles on Crunchbase and Dealroom, check PitchBook and Beauhurst, and make sure deal coverage in Real Deals, Sifted and regional business press uses your correct name. Awards such as the Real Deals Private Equity Awards, the Growth Investor Awards and the UK Private Capital Awards add independent evidence. - **The trade body's new name.** The BVCA formally became [UK Private Capital](https://www.ukprivatecapital.co.uk/about-us.html) on 27 January 2026, so update any page or profile that still describes you as a BVCA member. Keep performance and fundraising material behind a professional-investor gate, reviewed by compliance. The [FCA's rules on promoting non-mass market investments](https://www.handbook.fca.org.uk/handbook/COBS/4/12B.html) (COBS 4.12B) restrict promotions to retail clients, with limited exemptions. Regulatory status also varies: some smaller managers, including managers of qualifying venture capital funds, can [register with the FCA as small registered UK AIFMs](https://www.fca.org.uk/firms/aifmd/apply-registered-aifm) rather than become authorised, and in July 2026 the FCA consulted on [a new three-tier AIFM regime](https://www.fca.org.uk/publications/consultation-papers/cp26-28-uk-aifm-regime), with implementation envisaged for 2028. Describe your status accurately, and take legal advice before publishing anything that could be a promotion. ## What to do this month 1. Pick the three portfolio companies where inbound growth matters most to the plan, and write 20 to 50 buyer questions for each with their chief executives. 2. Run those questions in ChatGPT, Google, Copilot and Gemini, several times each in a clean session, and record the fields in the table above. 3. For every buy-and-build platform, check whether any acquired brand is still named as a rival, and fix the structured data first. 4. Agree one description for each group and brand, and correct the company's website, directory profiles and LinkedIn pages to match. 5. Add a one-page AI visibility summary to the next board pack, with share of answers, the main cause and the next fix per company. ## Frequently asked questions ### Who should own AI visibility: the operating partner or the portfolio chief executive? The chief executive owns the fixes, because they sit in the company's website, sales and marketing. The operating partner owns the method, so every company is measured the same way and the board sees comparable numbers. A single fixed question list per company, agreed by both, keeps the discussion on results rather than anecdotes. ### How often should we run the audit? Run the questions several times a month, because single answers vary. Report quarterly to the board, since most fixes take weeks to show in live search and months to reach answers drawn from a model's memory. Keep the question list fixed so quarter-on-quarter changes mean something. ### Can anyone guarantee a portfolio company will be named? No. Nobody can guarantee what an AI assistant will say, and anyone promising a place in ChatGPT's answers should be treated with caution. A provider can guarantee effort, reporting and outcomes it controls, such as meetings held. ChatGPT has shown labelled ads to UK users on its Free and Go plans since 11 August 2026, and OpenAI says the ads do not change the answer itself. ### Does this matter if the company sells through referrals and outbound? Yes. A buyer who hears about a company through a referral or an outbound message can check it with an assistant before replying, so a wrong or missing description can cost meetings in those channels too. Researched outbound and being named work best together. ### Which PE firms are known for being founder-friendly in the UK lower mid-market? An assistant answers this from what founders and the press say, not from fund websites. Firms that publish how they work after completion, and whose founders describe the experience in their own words in public, give it something to repeat. A fund that wants to be named here needs those independent accounts more than a new tagline. ### What should a founder ask a private equity firm before accepting a term sheet? In our view, start with what happens after completion: who sits on the board, how the value creation plan is set, what happens if the business misses the plan, and what role the founder keeps. Ask to speak to founders the firm has backed, including one where things went badly. Ask how decisions are made and how quickly. A fund that answers these plainly on its own website is also the fund an assistant can describe accurately. For a portfolio company that needs a plan its own team can deliver, our [Paid Discovery](/discovery) sets out every finding, ranked, with fixes written so the company's team can ship them. To start with a quick read on one company, [book a free Growth Review](/growth-review). ## Sources - [UK Private Capital, About us (2026)](https://www.ukprivatecapital.co.uk/about-us.html) - [6sense, 2025 Buyer Experience Report (2025)](https://6sense.com/newsroom/the-timeline-for-influencing-b2b-buyers-is-shrinking-insights-from-6senses-2025-buyer-experience-report/) - [G2 via PR Newswire, Half of B2B software buyers now start their research with AI chatbots (April 2026)](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html) - [Microsoft, FY26 Q4 earnings press release (July 2026)](https://www.microsoft.com/en-us/investor/earnings/fy-2026-q4/press-release-webcast) - [FCA, CP26/28 UK AIFM regime (July 2026)](https://www.fca.org.uk/publications/consultation-papers/cp26-28-uk-aifm-regime) - [FCA, Apply to be a small registered UK AIFM (accessed September 2026)](https://www.fca.org.uk/firms/aifmd/apply-registered-aifm) - [FCA Handbook, COBS 4.12B Promotion of non-mass market investments (accessed September 2026)](https://www.handbook.fca.org.uk/handbook/COBS/4/12B.html) --- Title: How can we tell which of our new leads came from ChatGPT? URL: https://fortitudemedia.ai/insights/attribution-ai-era-track-leads Author: Ashley Kenerson Published: 2026-03-31 Last reviewed: 2026-09-24 Topic: Costs, returns and the business case Short answer: Mostly by asking. Add a required "How did you first hear about us?" question with an AI option to every form, ask on every first call and carry the answer into your CRM. Since May 2026 GA4 has an AI Assistant channel for ChatGPT, Gemini and Copilot, but Perplexity and Claude need a custom rule, and clicks from Google's AI Overviews count as Organic Search. Many buyers read an answer and then search for your name, so report a floor, not a total. Your board wants to know how many leads came from AI. You open Google Analytics and find a handful of visits from ChatGPT last month, and no enquiries you can tie to them. Meanwhile your sales director says new prospects keep mentioning that "ChatGPT suggested you". Both are telling the truth. They are measuring different things. The reason is the way buyers use AI assistants. A finance director asks ChatGPT, Gemini or Copilot for a shortlist, reads your name, then searches for you on Google. Or types your address the next morning. Or forwards the answer to a colleague, who finds you a week later on LinkedIn. In each case your analytics records the last step, and the AI answer that started it is invisible. This matters because the habit is now common. [Forrester reported in January 2026](https://www.forrester.com/blogs/b2b_buyers_make_zero_click_buying_number_one/) that 94% of business buyers use AI in their buying process. Google Analytics 4 (GA4) has caught up only partly: since 13 May 2026 it has a default [AI Assistant channel](https://support.google.com/analytics/answer/9756891?hl=en), but it leaves out Perplexity and Claude, and it cannot separate clicks from Google's own AI Overviews and AI Mode. This guide explains what GA4 now shows, what it misses, and the form, call and CRM questions that fill the gap, so you can give your board an honest figure. ## Key takeaways - Most leads influenced by AI assistants do not show up as AI in analytics, because buyers often search for you by name or type your address afterwards. - Asking every new enquiry how they first heard of you, on the form and on the first call, is the most reliable way to measure AI-driven leads. - GA4's AI Assistant channel covers ChatGPT, Gemini, Copilot, Grok and DeepSeek, but Perplexity and Claude still need a custom channel rule. - Clicks from Google AI Overviews and AI Mode count as Organic Search in GA4 and cannot be separated from ordinary Google clicks. - ChatGPT links tagged with a source but no medium can fall into GA4's Unassigned channel, so check it before concluding AI sends you nothing. - Report AI-sourced leads as a floor built from several signals, and tell your board what is measured and what is estimated. ## Why does your analytics undercount AI? Analytics credits the last place a visitor came from before landing on your site. AI assistants are often the first place, not the last. There are four common journeys, and only one shows up clearly. ### The buyer clicks a link in the answer ChatGPT, Perplexity, Copilot, Gemini and Claude often show links to their sources. A click from most of them now lands in GA4's AI Assistant channel, and a click from Perplexity or Claude lands in Referral. A click from a Google AI Overview or AI Mode is different: Google's own documentation says it counts as Organic Search, so it looks exactly like an ordinary Google click. This is the journey you can partly see, and for most firms it is the smallest. ### The buyer searches for your name Many answers name firms without a link, or the buyer prefers to check you on Google. They search for your firm's name and click your listing. GA4 records organic search, and the visit looks like someone who already knew you. ### The buyer types your address, or the source is lost Some buyers note your name and visit later by typing your address. Others click a link in the ChatGPT phone app, which opens it in the phone's browser; in that hand-off the referrer often does not survive, according to a [July 2026 analysis by Nice Looking Data](https://www.nicelookingdata.com/blog/ga4-ai-traffic-chatgpt-referrals). Both show up as Direct. ### The buyer comes back through another channel A buyer reads your name in an AI answer, does nothing for a fortnight, then sees your founder's post on LinkedIn and clicks through. GA4 credits LinkedIn. These visitors often arrive knowing who you are and get to the point quickly, which we explore in [what your website should do now visitors arrive pre-sold](/insights/death-cold-website-visit-ai-funnel). The same shift explains why many firms see [traffic falling while their rankings hold](/insights/zero-click-search-your-business). ## What does GA4 show about AI traffic now? Less than the channel name suggests. Here is where each kind of AI visit usually ends up, based on Google's channel definitions and the Nice Looking Data analysis. | Where the visit came from | Where GA4 usually puts it | What to do | |---|---|---| | A link in ChatGPT, Gemini, Copilot, Grok or DeepSeek | AI Assistant channel (default since 13 May 2026) | Nothing extra, but see the next rows | | A link in Perplexity or Claude | Referral | Add a custom channel rule (Step 4) | | A ChatGPT link carrying a source tag but no medium | Unassigned | Catch it with a custom channel on session source | | A link opened from the ChatGPT phone app | Often Direct | Ask the buyer (Steps 1 to 3) | | A Google AI Overview or AI Mode link | Organic Search, not separable | Use Search Console's AI report for impressions (Step 5) | | Any AI answer followed by a search or typed address | Organic Search or Direct | Ask the buyer | The third row catches many firms out. ChatGPT adds its own name to some of the links it gives, as a tag that reads chatgpt or chatgpt.com depending on where in ChatGPT the link came from. Because the tag often carries a source but no medium, GA4 stops reading the referrer, the visit matches none of its channel rules and it falls into Unassigned. To check, open Reports, then Acquisition, then Traffic acquisition, find the Unassigned row and add Session source / medium as a secondary dimension. ## How to set up AI attribution The set-up takes a few hours and no new software. Most of the value comes from the first three steps, because asking people catches the journeys that analytics cannot. ### Step 1: Add the question to every enquiry form Add "How did you first hear about us?" as a required field, with options such as search engine, AI assistant (for example ChatGPT, Gemini, Copilot, Claude or Perplexity), LinkedIn, recommendation from someone, press or article, event, and other. Put a short free-text box beside it. People often write something more useful than the option they tick. ### Step 2: Ask on every first call Train whoever takes first calls to ask an open question first: "How did you come across us?" Then, if AI has not come up, a gentle follow-up: "Did you use ChatGPT or any other AI tool while you were looking at firms like ours?" The order matters. Asking about AI first plants the idea and inflates the answer. ### Step 3: Record it in your CRM as a required field Create two fields: where the buyer first heard of you, and whether they used an AI assistant at any point, with the assistant named if they remember. Make both required before an enquiry can move to the next stage, because optional fields get left blank in busy weeks. In HubSpot, for example, that means: - A contact property called "How did you first hear about us?", as a dropdown with the options from Step 1, added to your forms as a required field. - A second dropdown, "Used an AI assistant in research?", with the values yes, no and not asked, plus a text field for which assistant. Sales fills these in after the first call. - Matching deal properties, filled in when the deal is created (by a workflow if your plan includes them), and set as required when a deal moves to its first stage, so every proposal and win carries its source. Other CRMs work the same way. The point is that the answer travels with the deal, so you can later count proposals and wins, not just enquiries. ### Step 4: Set up GA4 to catch every AI source Create a custom channel group in GA4 (Admin, then Data display, then Channel groups), copying the default group so everything else stays the same. Add a channel called "AI assistants" with the condition Session source matches regex, using: `.*(chatgpt|openai|perplexity|claude|gemini|copilot).*` Move it above Referral in the list, so it catches Perplexity and Claude visits and the tagged ChatGPT visits that would otherwise land in Unassigned. Nice Looking Data reports that custom channel groups also reclassify past sessions, so you can see history as well. To link those visits to enquiries, build an exploration: Explore, then Free-form, with Session source / medium as rows and Sessions and Key events as values, filtered to the key event your enquiry form fires (for example generate_lead). That shows which assistants sent visits that went on to enquire. ### Step 5: Watch searches for your name, and Google's AI impressions In Google Search Console, filter the Performance report for queries containing your firm's name. A steady rise with no campaign, event or press coverage to explain it suggests people are hearing about you somewhere, and AI answers are among the likeliest places. Search Console also has [generative AI performance reports](https://developers.google.com/search/blog/2026/06/gen-ai-performance-reports), live for all sites since 31 August 2026. They show impressions, pages, countries and devices for AI Overviews and AI Mode, though not clicks or the searches behind them. Use them to see which pages Google's AI features show, alongside the Organic Search figures in GA4. ### Step 6: Bring the signals together monthly Once a month, put the form answers, call notes, AI channel visits, branded search trend and Search Console AI impressions side by side, with what the assistants currently say about you. Our guide to [tracking whether AI assistants recommend your firm](/insights/track-ai-recommending-your-business) explains that last check. ## How to read the signals together Read them the way a detective would. Each has a blind spot, but when they point the same way you can be reasonably confident. For example, say a firm receives 40 enquiries in a month. Five people pick "AI assistant" on the form. First calls reveal three more who used ChatGPT or Perplexity during their research but ticked "search engine", because that is how they finally arrived. GA4 shows a dozen visits in the AI assistants channel, two of which became enquiries already counted. And searches for the firm's name rose, with nothing else to explain it. A fair summary is that at least 8 of the 40 enquiries involved an AI assistant, and the true number is probably higher, because some buyers will not remember or mention it. That figure rests on evidence and gives you a trend to follow. Then track how many AI-involved enquiries become proposals and clients compared with the rest, rather than assuming they are better or worse. ## What share of leads should we expect from AI search? There is no reliable benchmark for B2B services, so judge your own trend rather than someone else's average. The share depends on how many of your buyers use assistants to shortlist and whether you are named when they do. In [G2's April 2026 survey](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html), 51% of software buyers said they start research in an AI chatbot, but that is one sector. Once a firm is named, the share can be meaningful. For Helm Club, a founder membership community we work with, "1 in 5 new members this year came through an AI recommendation." It is one organisation in one market, so treat it as a sign of what is possible, not a benchmark. The detail is in the [Helm Club case study](/work/helm-club). ## How to report it to your board Report AI-sourced leads as a floor, explain the method and show the trend. Boards are comfortable with estimates when the method is clear, and lose trust when a precise-looking number turns out to be a guess. Using the example above, a monthly line might read: "At least 8 of 40 enquiries this month involved an AI assistant, based on form answers and first calls. Visits in our AI assistants channel rose again, and searches for our name are up on last quarter." Put next to it what the assistants say for your buyers' questions, so the board sees cause and effect side by side. Over time, add how many AI-involved enquiries became proposals and clients. Our guides to [building an AI visibility dashboard and board report](/insights/build-ai-visibility-dashboard) and [measuring the return from your website](/insights/measure-website-roi-b2b) show how to fit it into a wider report. ## Mistakes that distort the numbers - **Counting only GA4's AI Assistant channel.** It misses Perplexity, Claude, Unassigned ChatGPT visits and every buyer who searched for your name. - **Asking leading questions.** "Did ChatGPT recommend us?" invites a polite yes. Ask openly first. - **Leaving fields optional.** Partial data can look like a trend when it is really a change in who bothered to fill in the form. - **Crediting all branded search to AI.** Press, events, referrals and your own sales activity drive searches for your name too. Only the unexplained rise is a clue. - **Building a complicated model too early.** Weighted models that split credit across every touchpoint rest on guesses. Get the basic questions answered consistently for a few months first. - **Changing the method every month.** A rough measure used the same way for a year tells you more than a perfect one changed every quarter. ## What to do this month 1. Add "How did you first hear about us?" with an AI assistant option to every enquiry form, as a required field. 2. Brief whoever takes first calls on the open question and the gentle follow-up. 3. Create the source and AI fields in your CRM and make them travel with each deal. 4. Set up the GA4 custom channel group and check last month's Unassigned sessions for ChatGPT. 5. Put a monthly half hour in the diary to bring the signals together and write one line for the board. ## Frequently asked questions ### Can GA4 show us leads from ChatGPT? Partly. With the AI Assistant channel, or a custom channel group, it can show visits that came from ChatGPT and which of those visitors went on to enquire. It cannot show buyers who read an AI answer and then searched for you or typed your address, which is why asking buyers matters so much. ### Why does ChatGPT traffic show as Direct in GA4? Usually because the referrer was lost. Links opened from the ChatGPT phone app often reach your site with no referrer, and buyers who copy your name or address from an answer arrive by typing it. GA4 then has nothing to go on and records Direct. Asking on the form and first call is the only way to recover these. ### Does GA4's AI Assistant channel include Perplexity and Claude? No. Google's definition, checked in September 2026, lists ChatGPT, Gemini, DeepSeek, Copilot and Grok, and excludes Google's own AI Overviews and AI Mode. Perplexity and Claude visits appear as Referral unless you add them to a custom channel group, as in Step 4. ### How do I see Google AI Mode traffic in Search Console? You can see impressions, not clicks. The generative AI performance reports show how often your pages appeared in AI Overviews and AI Mode, which pages, and by country and device. Clicks from those features are counted in GA4 as Organic Search and cannot be split out. ### Do people remember that ChatGPT recommended us? Many do, especially if you ask soon after they get in touch and in conversation rather than on a form. Some will not, and some think of it as research rather than a recommendation. That is why the follow-up asks whether they used an AI tool while looking, and why the answers are a minimum. ### Is a rise in searches for our name proof that AI is working? No, it is a clue. Searches for your name also rise after press coverage, events, referrals and sales campaigns. If you can rule those out, and your tracking shows assistants naming you more often over the same period, the link is reasonable. Present it as supporting evidence, not proof. ### What if almost nobody mentions AI? Then you have learned something. Either buyers in your market are not yet using assistants to shortlist, or they are and you are not being named. Check what ChatGPT, Gemini, Copilot, Claude and Perplexity say when asked your buyers' questions. If competitors are named and you are not, a low number is a warning rather than a reassurance. If you would like a second pair of eyes on how you track where enquiries come from, [book a free Growth Review](/growth-review): 30 minutes with David Adams, and three fixes in writing afterwards. ## Sources - [Google Analytics Help, Default channel group (September 2026)](https://support.google.com/analytics/answer/9756891?hl=en) - [Nice Looking Data, AI traffic in GA4: how ChatGPT, Perplexity, Gemini and Google AI Mode referrals show up (July 2026)](https://www.nicelookingdata.com/blog/ga4-ai-traffic-chatgpt-referrals) - [Google Search Central, Introducing Search generative AI performance reports in Search Console (June 2026)](https://developers.google.com/search/blog/2026/06/gen-ai-performance-reports) - [Forrester, B2B buyers make zero-click buying number one (January 2026)](https://www.forrester.com/blogs/b2b_buyers_make_zero_click_buying_number_one/) - [G2, New G2 research: half of B2B software buyers now start their research with AI chatbots (April 2026)](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html) --- Title: How do AI crawlers read my website, and are we letting them in? URL: https://fortitudemedia.ai/insights/ai-crawlers-vs-google-spiders Author: David Adams Published: 2026-03-31 Last reviewed: 2026-09-24 Topic: Getting named by AI assistants Short answer: AI crawlers such as GPTBot, OAI-SearchBot, ClaudeBot and PerplexityBot fetch your pages but, unlike Googlebot, do not run JavaScript, so your key text must be in the HTML your server sends. Each obeys robots.txt by name, though fetchers acting on a user's question, such as ChatGPT-User, may not. Check robots.txt, your Cloudflare or firewall settings and a text-only view of your key pages. Most B2B firms should let the search crawlers in. Most founders have never looked at their robots.txt file and could not say what a crawler is. That was fine when the only crawler that mattered was Google's and your web agency handled it. It matters now, because a different set of crawlers decides whether ChatGPT, Claude, Copilot and Perplexity can read your site at all, and firms regularly shut them out without knowing. A crawler is a program that visits web pages, reads them and follows links, so that a search engine or an AI assistant knows what is there. Google's crawler, Googlebot, has been visiting your site for twenty years. The AI crawlers, with names like GPTBot, ClaudeBot and PerplexityBot, are newer and behave differently. The most important difference is JavaScript: when [Vercel and MERJ studied AI crawler traffic in December 2024](https://vercel.com/blog/the-rise-of-the-ai-crawler), they found that "none of the major AI crawlers currently render JavaScript". Googlebot does. They are not rare visitors either. In the same study, GPTBot, Claude's crawler, AppleBot and PerplexityBot together made nearly 1.3 billion fetches a month across Vercel's network, a little over 28% of Googlebot's volume. This guide explains how the two kinds of crawler differ, which names to know in 2026, how to check whether you are letting them in, and how to decide whether you should. ## Key takeaways - Googlebot runs JavaScript; the main AI crawlers fetch pages but do not, so text that only appears after scripts load is invisible to them. - Each AI company now runs separate crawlers for training, for search and for fetching a page when a user asks, and you can allow or block each by name. - Blocking a training crawler such as GPTBot does not remove you from ChatGPT search; blocking OAI-SearchBot does. - robots.txt is not a complete opt-out: OpenAI says its rules may not apply to ChatGPT-User, and Perplexity says Perplexity-User generally ignores them. - A crawler follows only the most specific group in robots.txt, so private folders must be listed under every crawler you name, not just under the asterisk. - Cloudflare and other security services can block AI crawlers whatever robots.txt says, so check the firewall as well as the file. - For most B2B firms, letting the search crawlers in costs little and is the precondition for being named. ## How does Googlebot read your site differently from AI crawlers? Googlebot renders your pages like a browser, including the JavaScript, and reads them to rank them. The main AI crawlers read only the HTML your server sends, and they read it so an assistant can describe you, cite you or learn from you. | | Googlebot | AI crawlers (GPTBot, ClaudeBot, PerplexityBot and others) | |---|---|---| | Runs JavaScript | Yes, in a headless browser | No: they fetch script files but do not run them (Vercel and MERJ, December 2024) | | Purpose | Rank pages for Search, including AI Overviews and AI Mode | Train models, index for AI search, or fetch a page for one user's question | | How often it visits | Often, weighted to important pages | Varies by crawler; some now crawl at very high volume | | Controlled by | robots.txt, Search Console | robots.txt by name, except some user-triggered fetchers | Because Googlebot renders JavaScript, a script-built site can look fine in Google, and in Gemini and Google's AI features that draw on Google's index, while ChatGPT, Claude and Perplexity see a near-empty page. That gap is the most common technical reason a firm that ranks well is missing from AI answers. Crawl frequency varies widely by crawler. Some AI crawlers visit rarely; others are heavy. [Cloudflare reported in August 2025](https://blog.cloudflare.com/perplexity-is-using-stealth-undeclared-crawlers-to-evade-website-no-crawl-directives/) that Perplexity's declared crawler made 20 to 25 million requests a day across its network. What matters more for your firm is that assistants which search the web live, such as ChatGPT search and Perplexity, can fetch a changed page within days or weeks, while answers drawn from a model's training change only when the model is retrained, which takes months. ## Which AI crawlers should you know by name? Know the search crawlers and user-triggered fetchers first, because they decide whether you appear in live answers. The training crawlers decide only whether your content helps train future models. | Company | Name | What it does | Follows robots.txt? | If you block it | |---|---|---|---|---| | OpenAI | GPTBot | Collects content that may be used to train models | Yes | Content not used for training; ChatGPT search unaffected | | OpenAI | OAI-SearchBot | Indexes pages for ChatGPT search | Yes | You are not shown in ChatGPT search answers | | OpenAI | ChatGPT-User | Visits a page when a user's request needs it | OpenAI says rules "may not apply" | A block may not stop it | | Anthropic | ClaudeBot | Collects content for model training | Yes | Future content excluded from training | | Anthropic | Claude-SearchBot | Indexes pages to improve Claude's search results | Yes | May reduce your visibility in Claude's search | | Anthropic | Claude-User | Fetches pages when a user asks Claude a question | Yes | Claude cannot retrieve your page for users | | Perplexity | PerplexityBot | Indexes pages for Perplexity's search; not used for training | Yes | You are less likely to appear in Perplexity | | Perplexity | Perplexity-User | Visits a page when a user asks a question | Perplexity says it "generally ignores" them | A block may not stop it | | Google | Googlebot | Crawls for Search, including AI Overviews and AI Mode | Yes | You leave Google Search, AI features included | | Google | Google-Extended | Not a crawler: a token controlling use in Gemini training | Yes | No effect on Search, rankings or AI Overviews | | Microsoft | Bingbot | Crawls for Bing, which feeds Copilot | Yes | You drop out of Bing and Copilot | | Apple | Applebot-Extended | Not a crawler: controls use in training Apple's models | Yes | Pages can still appear in Apple's search features | | Meta | Meta-ExternalAgent | Crawls for AI model training and indexing | Yes | Your content is not collected by it | The sources are each company's own documentation: [OpenAI](https://developers.openai.com/api/docs/bots), [Anthropic](https://support.claude.com/en/articles/8896518-does-anthropic-crawl-data-from-the-web-and-how-can-site-owners-block-the-crawler) (April 2026) and [Perplexity](https://docs.perplexity.ai/guides/bots), plus Google's, Apple's and Meta's crawler pages. If your robots.txt still names older Anthropic user agents copied from a 2023 template, replace them: Anthropic now documents only ClaudeBot, Claude-SearchBot and Claude-User. ## Does ChatGPT follow robots.txt? Its crawlers do. GPTBot and OAI-SearchBot obey robots.txt, but OpenAI's documentation says that because ChatGPT-User visits are "initiated by a user, robots.txt rules may not apply". So a robots.txt block controls training and ChatGPT search, but is not a complete opt-out from ChatGPT reading your page. The same split exists elsewhere. Perplexity says Perplexity-User "generally ignores robots.txt rules", and Cloudflare's August 2025 report found that Perplexity had used undeclared crawlers to reach sites that had blocked it; Cloudflare removed Perplexity from its verified bots list. In the same tests, Cloudflare found ChatGPT-User fetched the robots.txt file and stopped when it was disallowed. Anthropic says all three of its bots respect robots.txt. If you need content kept away from every AI tool, put it behind a login. ## How do you check whether you are letting AI crawlers in? Read your robots.txt file, check your Cloudflare or firewall settings, and look at a key page the way a crawler sees it. The five steps below take under an hour. ### Step 1: Read your robots.txt file Type your domain followed by /robots.txt into a browser. Look for "User-agent:" lines naming GPTBot, OAI-SearchBot, ChatGPT-User, ClaudeBot, Claude-SearchBot, Claude-User, PerplexityBot, Perplexity-User, Google-Extended, Applebot-Extended or Meta-ExternalAgent, and read the lines beneath. "Disallow: /" means that crawler is shut out of the whole site. Then check the structure. A crawler obeys only the group with the most specific user agent that matches it, and [Google's robots.txt documentation](https://developers.google.com/search/docs/crawling-indexing/robots/robots_txt) is explicit that specific groups and the "*" group "are not combined". If you name GPTBot in its own group with "Allow: /", it ignores any "Disallow: /admin/" you put under "User-agent: *". ### Step 2: Check Cloudflare, your host and your firewall Security services can block AI crawlers before robots.txt is ever read. Cloudflare [changed its default to block AI crawlers in July 2025](https://blog.cloudflare.com/content-independence-day-no-ai-crawl-without-compensation/), and its [documentation](https://developers.cloudflare.com/bots/additional-configurations/block-ai-bots/) says that from 15 September 2026 new domains block training and agent crawlers on pages that show ads, while search crawlers stay allowed. In the Cloudflare dashboard, open your domain and go to Security Settings, then Configure AI bot policies (older accounts show a Block AI bots switch). You can set search, agent and training crawlers separately to block on all pages, block on pages with ads, or allow. If you use another host, CDN or security plugin, ask whoever manages it whether any AI or bot protection is on and which names it blocks. ### Step 3: See what a crawler actually receives Open a key page, right-click and choose "View page source", or ask your developer to fetch it with JavaScript off. Search the source for your main heading and first paragraph. If they are missing, AI crawlers are not seeing them. Example: an illustrative services page built entirely by script. | What a visitor sees | What a crawler that does not run JavaScript receives | |---|---| | Heading, three paragraphs on the service, two case studies, team photos, contact form | Page title, menu links, footer, and an empty space where the content should be | The fix is to have the server send the page text in the HTML, with JavaScript adding interaction on top. Our guide to [how your website platform affects whether ChatGPT can read it](/insights/cms-choice-affects-ai-visibility) covers the common builders. ### Step 4: Look at your server logs Your logs record every visit and its user agent. If GPTBot, OAI-SearchBot, ClaudeBot and PerplexityBot never appear, they are blocked or have not found you. If they appear with error codes such as 403, something is refusing them. ### Step 5: Decide and write it down Record which crawlers you allow and why, so the next developer does not quietly reverse it. Review it every quarter and after any rebuild, change of host or CDN. ## Should you block GPTBot but allow OAI-SearchBot? You can, and it is a sensible middle ground if you object to training use. Blocking GPTBot keeps your content out of OpenAI's model training, while allowing OAI-SearchBot keeps you eligible for ChatGPT search answers, which is where recommendations increasingly come from. For most B2B firms we recommend allowing both. Your content is written to win clients, and a training block means future models know less about you. Whatever you choose, repeat the private folders under every named group. This example allows the main AI crawlers everywhere except two private folders: ``` User-agent: GPTBot Disallow: /admin/ Disallow: /private/ User-agent: OAI-SearchBot Disallow: /admin/ Disallow: /private/ User-agent: ClaudeBot Disallow: /admin/ Disallow: /private/ User-agent: Claude-SearchBot Disallow: /admin/ Disallow: /private/ User-agent: PerplexityBot Disallow: /admin/ Disallow: /private/ User-agent: * Disallow: /admin/ Disallow: /private/ ``` To block training but stay in ChatGPT search, change only the GPTBot group to "Disallow: /". The same pattern works for ClaudeBot (training) and Claude-SearchBot (search). ## Do AI crawlers slow down your website or add to hosting costs? For a typical B2B site of a few hundred pages, rarely enough to notice, in our experience. The heavy volumes in the studies are totals across millions of sites. If your logs show one crawler making thousands of requests a day, slow it rather than block it. Anthropic's bots support the Crawl-delay line in robots.txt, and Cloudflare and most CDNs can rate-limit a named bot. Blocking a search crawler to save a little bandwidth removes you from the answers your buyers read. ## What should you build so both kinds of crawler read you well? Put the page text in the HTML, use one clear heading hierarchy, and write sections that make sense on their own. Almost everything that helps AI crawlers also helps Google. Structured data helps search engines read your facts correctly, but Google says its AI features need no special markup, and OpenAI and Anthropic have not said they read it; keep it accurate rather than treating it as a ranking trick. Our guide to [schema markup and AI assistants](/insights/structured-data-schema-markup-ai) explains what is worth adding, and the full list of checks is in our [technical checklist for an AI-ready website](/insights/technical-checklist-ai-ready-website). ## What to do this month 1. Open your robots.txt file and list every AI crawler it names, with what each is allowed to do. 2. Replace any older Anthropic names with ClaudeBot, Claude-SearchBot and Claude-User, and repeat your private folders under every named group. 3. Check the AI bot settings in Cloudflare, or ask your host or IT provider what their bot protection blocks. 4. View the source of your three most important pages and confirm the main text is there without JavaScript. 5. Search your server logs for GPTBot, OAI-SearchBot, ClaudeBot and PerplexityBot, and note any error codes. 6. Write down your decision on each crawler and put a quarterly review in the diary. ## Frequently asked questions ### If we block GPTBot, will we disappear from ChatGPT? Not from ChatGPT search, if OAI-SearchBot is still allowed. Blocking GPTBot stops your content being used for training, so future models may know less about you. Blocking OAI-SearchBot is what removes you from ChatGPT search answers. ### Will blocking AI crawlers protect our Google rankings? It makes no difference to them. Googlebot is separate, and Google says Google-Extended "does not impact a site's inclusion in Google Search nor is it used as a ranking signal". Google-Extended controls use in Gemini training, not AI Overviews, which draw on Google's normal index. ### Is our Webflow or Wix site readable by ChatGPT? Usually, but check rather than assume. Hosted builders generally send the main page text in the HTML, but text inside embedded widgets, tabs, sliders or third-party apps may only appear after scripts run. The view-source test in Step 3 settles it in two minutes, page by page. ### How quickly will ChatGPT notice a change to our site? Assistants that search the web live, including ChatGPT search, Perplexity, Copilot and Google's AI features, can reflect a changed page within weeks. Anything the model learned in training changes only when the model is retrained, which takes months. Fix wrong facts early, as set out in [why ChatGPT describes your firm wrongly](/insights/ai-conflicting-information-online). ### Our site is an app-style build. Do we have to rebuild it? Not necessarily. Most modern frameworks can render public pages on the server or generate them in advance, so the HTML arrives complete. Ask your developer about server-side rendering or static generation for the marketing pages. If the platform cannot do it, that is a strong case for moving the marketing site. ### How often should we review our crawler settings? Every quarter, and after any rebuild or change of host. New crawlers appear, names change, CDNs change their defaults, and developers copy robots.txt files from old projects. You cannot be recommended by an assistant that cannot read you, and ten minutes with robots.txt and a page's source will tell you whether that is your problem. For a check of how your site reads to AI crawlers and buyers, [book a free Growth Review](/growth-review); if the site needs rebuilding so the content is in the HTML, that is [Website That Sells](/website-that-sells). ## Sources - [Vercel and MERJ, The rise of the AI crawler (December 2024)](https://vercel.com/blog/the-rise-of-the-ai-crawler) - [OpenAI, Overview of OpenAI crawlers (September 2026)](https://developers.openai.com/api/docs/bots) - [Anthropic, Does Anthropic crawl data from the web, and how can site owners block the crawler? (April 2026)](https://support.claude.com/en/articles/8896518-does-anthropic-crawl-data-from-the-web-and-how-can-site-owners-block-the-crawler) - [Perplexity, Perplexity crawlers (September 2026)](https://docs.perplexity.ai/guides/bots) - [Google Search Central, How Google interprets the robots.txt specification (August 2026)](https://developers.google.com/search/docs/crawling-indexing/robots/robots_txt) - [Cloudflare, Content Independence Day: no AI crawl without compensation (July 2025)](https://blog.cloudflare.com/content-independence-day-no-ai-crawl-without-compensation/) - [Cloudflare Docs, Block AI bots and AI bot policies (July 2026)](https://developers.cloudflare.com/bots/additional-configurations/block-ai-bots/) - [Cloudflare, Perplexity is using stealth, undeclared crawlers to evade website no-crawl directives (August 2025)](https://blog.cloudflare.com/perplexity-is-using-stealth-undeclared-crawlers-to-evade-website-no-crawl-directives/) --- Title: How do I know whether our website is actually working? URL: https://fortitudemedia.ai/insights/evaluate-whether-website-working Author: David Adams Published: 2026-03-31 Last reviewed: 2026-09-24 Topic: Websites that bring work in Short answer: Your website is working if it brings in qualified enquiries, helps buyers who already know you decide in your favour, and gives Google and AI assistants an accurate picture of your firm. To check, count last quarter's website enquiries, read the site on a phone as a buyer would, ask ChatGPT, Google and Copilot who they recommend, and test your contact form. Then score the site from 1 to 5 on six questions and fix the two lowest scores first. Every month a report arrives. Sessions are up, bounce rate is down, and a chart shows a line going roughly the right way. Meanwhile your sales team says new clients almost never mention the website, and the contact form brings in mostly suppliers trying to sell you something. You are left wondering whether the site does anything at all. A B2B website is working if it brings in enquiries from the right people, helps buyers who already know you decide in your favour, and gives search engines and AI assistants an accurate picture of what you do. Traffic on its own proves none of those. The last job matters more each year. In [G2's April 2026 survey of 1,076 B2B software buyers](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html), 51% said they now start their research in an AI chatbot, up from 29% a year earlier. G2's sample is software buyers, but we see the same pattern in client work: at [Helm Club](/work/helm-club), "1 in 5 new members this year came through an AI recommendation." This guide shows you how to judge your own site in an afternoon, without specialist tools. ## Key takeaways - A B2B website is working if it brings in qualified enquiries and helps win deals, whatever the traffic line does. - Start with results: count last quarter's website enquiries and ask sales how new clients used the site. - Analytics undercounts AI visits, because many app visits land in Direct, so add a GA4 AI channel and ask every enquiry how they found you. - Half of software buyers now start research in an AI chatbot (G2, April 2026), so check what the assistants say about you. - Read your site on a phone as a stranger would, and fix any page that does not say who you help and prove it. - Score the site from 1 to 5 on the same six questions every quarter. ## What does "working" mean for a B2B website? It means four jobs done well: bringing in qualified enquiries at a steady rate, helping buyers who already know you decide in your favour, giving Google and AI assistants an accurate picture of what you do and for whom, and being somewhere your team is proud to send people. Traffic, page views and bounce rate help you diagnose problems with those jobs, but they are not the jobs. A site with modest traffic that brings in two good enquiries a month beats a busy one that brings in none. To put a financial value on the site, see [how to measure the return you get from your website](/insights/measure-website-roi-b2b). ## How to evaluate your website, step by step ### Step 1: Count the enquiries that came through the site Start with the result. List every enquiry from the last three to six months that came through the website: forms, calls to the number on the site, emails to the address listed, meeting bookings. Note which were the right kind of client and which became work. Then ask sales how many clients won in that period looked at the site first, and what they said. From now on, ask every new enquiry how they found you, with "ChatGPT or another AI assistant" among the options, and record it in your CRM. We cover combining those answers with analytics in [how to tell which new leads came from ChatGPT](/insights/attribution-ai-era-track-leads). ### Step 2: Check where visitors come from, including AI assistants Look for visits from chatgpt.com, claude.ai, perplexity.ai, gemini.google.com and copilot.microsoft.com. Three details trip people up. - **GA4's AI channel has gaps.** GA4 gained a default "AI Assistant" channel on 13 May 2026, and [Google's channel definitions](https://support.google.com/analytics/answer/9756891?hl=en) say it covers ChatGPT, Gemini, DeepSeek, Copilot and Grok, while clicks from AI Overviews and AI Mode count as Organic Search. Claude and Perplexity are not on the list, so their visits sit in Referral. - **ChatGPT tags its links.** [OpenAI says](https://help.openai.com/en/articles/12627856-publishers-and-developers-faq) ChatGPT "automatically includes the UTM parameter utm_source=chatgpt.com in referral URLs", so filter on that source too. - **App visits often show as Direct.** Links tapped in the ChatGPT app often lose their referrer on the way to the phone's browser ([Nice Looking Data, July 2026](https://www.nicelookingdata.com/blog/ga4-ai-traffic-chatgpt-referrals)). Direct visits to a deep service page nobody would type from memory are often unlabelled AI traffic. To see the AI visits you can identify in one line, go to Admin, then Data display, then Channel groups in GA4. Copy the default group, add a channel called "AI assistants" with the condition "Session source matches regex", and paste `chatgpt\.com|chat\.openai\.com|claude\.ai|perplexity\.ai|gemini\.google\.com|copilot\.microsoft\.com`. Move it above Referral and save. Treat the result as a minimum. For Google's own AI features, the generative AI reports [Search Console rolled out to all sites by 31 August 2026](https://developers.google.com/search/blog/2026/06/gen-ai-performance-reports) show how often your pages appear in AI Overviews and AI Mode. ### Step 3: Read the site as a buyer would This is the most useful step and the one most firms skip. Open your homepage on a phone and give it ten seconds. Can you tell who the firm helps, what it does and where? Then check your main service page: does it describe the problem in the buyer's words, show proof such as case studies with outcomes, show who you would work with, and make the next step obvious? If a new salesperson could not explain your firm from those two pages, neither can a buyer or an AI assistant. We cover this in [what makes a buyer trust your website enough to get in touch](/insights/psychology-of-trust-website-conversions). ### Step 4: Follow the path to an enquiry From each of your five most visited pages, count the clicks to a way of getting in touch. Then fill in your own contact form on your phone. Does it work, and how quickly does anyone reply? In our experience this step turns up broken forms and unchecked inboxes far more often than founders expect, and none of it shows in a traffic report. ### Step 5: Ask AI assistants what they know about you Write five questions a buyer would ask, such as "best IT support for accountancy practices in the North West". Ask each in ChatGPT and Google's AI Mode first, then Copilot and Gemini, then Perplexity and Claude, in a fresh chat each time. Record who is named and how you are described. Then ask each assistant what your firm does. A vague or wrong description means your site is not stating the facts clearly, or other sources contradict it. A correct description with no recommendation means the assistants know you and have no reason to name you. That was [ID-Pal](/work/id-pal)'s position: "Across 36 recorded AI answers to the questions ID-Pal's buyers actually ask, ID-Pal was named three times, and only for its brand or its Salesforce integration. AI knew who it was, and had no reason to recommend it." Answers vary between runs, so for tracking keep a fixed list of 20 to 50 buyer questions, ask each several times a month in a clean session, and count a win only when it repeats, as set out in [finding out whether ChatGPT recommends your firm](/insights/track-ai-recommending-your-business). ### Step 6: Check the technical basics Technical problems rarely win work but can stop a site working. [Google's Core Web Vitals guidance](https://developers.google.com/search/docs/appearance/core-web-vitals) sets three targets: main content within 2.5 seconds, a response to a tap or click in under 200 milliseconds, and a layout shift score under 0.1. Also check the site on a real phone, confirm https with no warnings, fix "not found" pages, and ask your developer to confirm that neither robots.txt nor a Cloudflare or firewall setting blocks the crawlers used by ChatGPT, Claude and Perplexity. The full list is in [whether your website is technically ready for AI search](/insights/technical-checklist-ai-ready-website). ### Step 7: Compare yourself with three competitors Look at three competitors you meet in pitches the same way, and compare who gets named in Step 5. Look for the gaps a buyer would notice with both sites open side by side. ## How do you score the result? Score the site from 1 to 5 on six questions, with one sentence of evidence beside each. The questions follow the lenses of our free [Growth Review](/growth-review), which scores a firm out of ten on five: Found, Named by AI, Website, Authority and Pipeline. Here Website is split into three, because it is the part you control, Enquiries comes first as the result, and Authority and Pipeline, which sit outside the site, are left out. | Question | Growth Review lens | A 5 looks like | A 1 looks like | |---|---|---|---| | 1. Enquiries: qualified enquiries at a steady rate? | The result | Qualified enquiries monthly, some becoming work | None, or only suppliers | | 2. Found: does Google show you for buyer questions? | Found | Service and sector pages appear | Only your own name ranks | | 3. Named by AI: named and described accurately? | Named by AI | Named for several buyer questions | Not named, or described wrongly | | 4. Clarity and proof: clear in ten seconds, with current proof? | Website | Named sectors, recent results, real people | Generic claims, old testimonials | | 5. Path to contact: easy from every key page, and working? | Website | Clear next step, tested form, same-day reply | Buried, broken or unanswered | | 6. Technical health: fast, secure, readable by crawlers? | Website | Meets Core Web Vitals, crawlers allowed | Slow, blocked or empty without JavaScript | The maximum is 30, and the bands are a rule of thumb. **24 or more** is an average of at least 4, mostly "yes": the site is broadly working. **15 to 23** means it works in parts; start with the two lowest scores. **Below 15** is an average under 2.5, more "no" than "yes": the site is probably costing you work. Whatever the total, a 1 or 2 on Enquiries outweighs the rest. Repeat the score quarterly, with the same person scoring. ### A worked example Example: a fictional 40-person engineering consultancy. The firm and the figures are illustrative. | Question | Score | Evidence | |---|---|---| | Enquiries | 2 | Six website enquiries last quarter, one qualified, none became work. | | Found | 3 | Ranks for its city and trade, but for none of the sector questions buyers ask. | | Named by AI | 1 | Not named for any of five buyer questions; ChatGPT lists a service dropped two years ago. | | Clarity and proof | 2 | The homepage says "engineering excellence"; the newest case study is five years old. | | Path to contact | 4 | Contact button on every page; the form worked and got a reply within two hours. | | Technical health | 4 | Meets Core Web Vitals; robots.txt allows the main AI crawlers. | | **Total** | **16** | Working in parts. | It chose two fixes from the lowest scores. First, rewrite the homepage and two main service pages to name the sectors served, each with a recent case study that leads with its result. Second, correct the old service list everywhere it appears, on the site, LinkedIn and directories, so AI assistants have one consistent description to draw on. ## Why does our website get traffic but no enquiries? Usually for one of four reasons: the visitors are not buyers, the pages they land on do not speak to them, the route to contact is broken, or enquiries arrive and are not credited to the site. Check them in that order. | Cause | How to spot it | First fix | |---|---|---| | The visitors are not buyers | Visits land on old blog posts, or come from job seekers and other countries | Point effort at service, sector and case study pages | | The pages do not speak to buyers | General claims, no sectors, results or named people | Rewrite the top five landing pages around one buyer each | | The route to contact is broken | The Step 4 test fails, or nobody replies the same day | Fix the form, add a phone number and a next step | | Enquiries are not credited | Clients read the site, then phoned or came through a referral | Ask every enquiry how they found you | Falling numbers need a cause before a redesign. At ID-Pal, "Traffic to the pages that bring in enquiries had fallen by around 40 per cent since the NorthRow move, and nobody could say why." The cause was one no traffic report could show: "The redirect from northrow.com was technically correct, but nothing on either site told Google or the AI engines that NorthRow and ID-Pal were the same company." ## When should you fix the site, and when should you rebuild it? Fix it if the problems are about content and clarity. Rebuild only if the platform or structure stops you fixing them. Rewriting service pages, adding case studies and simplifying forms can be done on the existing site, faster than a rebuild and without risking your search history. Consider a rebuild when your team cannot edit pages without an agency, the platform is unsupported, or the structure no longer matches how you sell. We compare the options in [template or custom website: which does a B2B firm need?](/insights/template-vs-custom-website-b2b), and explain why a site left alone drifts in [whether to redesign or keep improving](/insights/website-redesign-vs-continuous-iteration). ## What to do this month 1. **List last quarter's website enquiries.** Note which were qualified and which became work. 2. **Add the AI channel to GA4.** Use the pattern in Step 2, then compare it with Direct visits to your service pages. 3. **Read your homepage and main service page on your phone.** Write down what a stranger would understand in ten seconds. 4. **Ask the main AI assistants five buyer questions.** Record who is named and how your firm is described. 5. **Test your own contact form** on a phone, and time the reply. 6. **Score the site on the six questions.** Fix one thing in each of the two lowest before next month. ## Frequently asked questions ### How many enquiries a month should a B2B website generate? There is no reliable benchmark, because client values and markets vary too much. Work backwards instead. For example, if you want 12 new clients a year, win one proposal in three, and half of qualified enquiries reach a proposal, you need about 72 qualified enquiries a year from all sources. If the website should supply a third, that is about two a month. ### How do I set up a channel for ChatGPT traffic in GA4? The default AI Assistant channel misses Claude and Perplexity. Add a custom channel under Admin, Data display, Channel groups, matching session source against the regex in Step 2, placed above Referral. Many AI app visits still arrive as Direct, so pair it with a "how did you find us" question. ### What should a free website audit include? It should start from your enquiries and your buyers' questions, not only an automated technical score. A useful audit covers what the site brings in, how it reads to a buyer on a phone, what Google and AI assistants say, whether the contact route works, the technical basics and two or three competitors. Our free Growth Review does this in 30 minutes: five scores, the buyer questions where you are not named, and three fixes in writing. ### How do I know if my website is putting buyers off? Listen to the sales process. Prospects who ask questions the site should answer, who did not realise you offer something you have done for years, or who need a PDF after every call are telling you the site is not doing its job. In analytics, look for visitors who reach a service page and leave without clicking anything. The quickest test is to ask five recent clients what they thought when they first looked. ### Our traffic is falling but enquiries are steady. Should we worry? Not necessarily. Many buyers now get answers from Google's AI Overviews or ChatGPT without clicking. If enquiries hold, the site may be doing its job with fewer, better-informed visitors. Check what AI assistants say about you and whether AI visits, including unlabelled Direct ones, are growing. ### How often should we evaluate our website? Run the full score once a quarter and check enquiries and AI answers monthly. Assistants that search the web live can reflect a changed page within weeks, while answers drawn from a model's training change only when it is retrained, which takes months. Most B2B websites that are not working have a handful of fixable problems rather than a fatal flaw. If you would like a second opinion, our free [Growth Review](/growth-review) scores your firm on the same lenses, shows the buyer questions where you are not named, and sends the three fixes we would make first, in writing. ## Sources - [G2, New G2 research: half of B2B software buyers now start their research with AI chatbots (April 2026)](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html) - [Google Analytics Help, Default channel group (September 2026)](https://support.google.com/analytics/answer/9756891?hl=en) - [Nice Looking Data, AI traffic in GA4: how ChatGPT, Perplexity, Gemini and Google AI Mode referrals show up (July 2026)](https://www.nicelookingdata.com/blog/ga4-ai-traffic-chatgpt-referrals) - [OpenAI Help Center, Publishers and developers FAQ (August 2026)](https://help.openai.com/en/articles/12627856-publishers-and-developers-faq) - [Google Search Central, Understanding Core Web Vitals and Google search results (December 2025)](https://developers.google.com/search/docs/appearance/core-web-vitals) - [Google Search Central Blog, Generative AI performance reports in Search Console (June 2026)](https://developers.google.com/search/blog/2026/06/gen-ai-performance-reports) --- Title: How do I organise our content so AI assistants see us as the expert? URL: https://fortitudemedia.ai/insights/topic-clusters-ai-understands Author: David Adams Published: 2026-03-31 Last reviewed: 2026-09-24 Topic: Getting named by AI assistants Short answer: Organise your content into topic clusters: one main guide on a subject you want to be known for, plus four or five articles that each answer one follow-up question, all using the same framework and terms. Google's AI features split each question into several related searches, so a cluster that answers the follow-ups gives them more to cite. Make every page quotable on its own, remove contradictions between pages, and say what each link means. You have forty or fifty articles on the website, some good, some written by an agency years ago and never read since. When a buyer asks ChatGPT or Gemini who the best specialists in your field are, a competitor with half your experience gets named and you do not. In our experience this is rarely a volume problem. It is an organisation problem. If forty pieces point in forty directions, disagree with each other in places and answer half of each question, an assistant has no single page or passage it can rely on to say what your firm knows. There is also a mechanical reason organisation matters more than it did. Google's AI features split one question into several related searches, a technique Google calls query fan-out, and cite pages that answer the parts. [Ahrefs found in March 2026](https://ahrefs.com/blog/ai-overview-citations-top-10) that only 38% of AI Overview citations came from pages ranking in Google's top ten for the original search, down from 76% in July 2025. The rest ranked lower or not at all for it, a shift that fits Google running several related searches for each question. A group of pages that covers a subject and its follow-up questions gives an assistant more to cite. This guide explains how to arrange what you publish into topic clusters that buyers and assistants can follow, using a worked example from employment law and the changes we made to our own guides. ## Key takeaways - A topic cluster is one main guide on a subject plus four or five connected articles that go deeper, all using the same framework and the same terms. - Google's AI features run several related searches for one question (query fan-out), so a cluster that answers each follow-up question gives them more to cite. - Assistants usually retrieve passages one at a time, so every page, and every section, should make sense on its own. - Pages in a cluster must not contradict each other: two versions of the same fact give an assistant a reason to trust neither. - Say what each link means in the sentence ("this covers stage two of the process in our redundancy guide") rather than relying on a bare "related reading" list. - Do not spin a page for every variation of a question: Google says doing so to manipulate AI answers breaks its scaled content abuse policy. - Measure with a fixed list of 20 to 50 buyer questions, and count a cluster as working when several of its pages are cited, not just one. ## What is a topic cluster? A topic cluster is a main guide on one subject, linked to a handful of articles that each cover part of that subject in more depth. The main guide is usually called the pillar page; the others are child pages. The pillar links to every child, every child links back, and children link to each other where it makes sense. The idea was popularised by HubSpot and built for Google, which reads the link structure as a sign that you cover a subject in depth. AI assistants need a stricter version, because they quote passages rather than weighing links alone. ## What is query fan-out, and how does it change content planning? Query fan-out is Google's term for its AI model issuing "a set of concurrent, related queries" to gather more material for one question. [Google's guide to its AI features](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide), updated in July 2026, gives the example of "how to fix a lawn that's full of weeds" fanning out to "best herbicides for lawns", "remove weeds without chemicals" and "how to prevent weeds in lawn". [Google says AI Mode](https://blog.google/around-the-globe/google-europe/united-kingdom/ai-mode-search-uk/), available in the UK since July 2025, uses this technique. For a B2B firm, a buyer's single question, "how do we handle redundancies without ending up at tribunal?", is answered from pages on consultation, selection, pay and settlement agreements. A clear page for each gives you several chances to be cited; one long page that touches each briefly gives you one. There is a limit. The same Google guide warns that creating "separate content for every possible variation of how people might search", including fan-out queries, "primarily to manipulate rankings or generative AI responses" breaks its scaled content abuse spam policy. A cluster answers the real sub-questions of a subject once each. It is not a page for every phrasing. ## How do AI assistants read a cluster? Mostly one passage at a time. When an assistant answers a question it retrieves the passages that best match each part of it, often from different pages and different sites, and composes an answer. It does not read your cluster from start to finish. So each page, and ideally each section, should open with a direct answer that makes sense without the rest: a [2026 analysis of 1.2 million ChatGPT answers](https://searchengineland.com/chatgpt-citations-content-study-469483) by Kevin Indig found 44% of citations came from the first 30% of a page. The pages must agree, because an assistant that finds two versions of the same fact on your site has a reason to use neither; our guide to [why ChatGPT describes firms wrongly](/insights/ai-conflicting-information-online) covers what happens then. And links should say what they mean, for readers and crawlers alike. Language models handle synonyms well, so "buying committee" on one page and "stakeholders" on another are unlikely to confuse them. Consistent terms still matter for a simpler reason: a reader and an assistant can quote your framework accurately only if you describe it the same way everywhere. ## What does a cluster that assistants can use look like? Every page is on the same subject, the children develop the framework the pillar sets out, and every page could be quoted on its own. Here is the example we use. A firm of employment lawyers wants to be known for handling redundancies. The pillar sets out a four-stage process; each child covers part of it. - **Pillar:** How to run a redundancy process that stands up at tribunal (sets out the four stages) - **Child 1:** Collective consultation: when it applies and how to run it (stage one) - **Child 2:** Selection criteria and scoring: what is fair and what is not (stage two) - **Child 3:** Redundancy pay and settlement agreements (stage three) - **Child 4:** Redundancy in a business acquisition (a situation that changes every stage) A loose cluster would add "how to write an employee handbook" or "managing poor performance". Those are employment law topics, but they say the firm writes about employment law in general, not that it is the firm to call when redundancies are coming. The children should develop the pillar's framework, not start their own. If the pillar says there are four stages, the best children are one deep article per stage. "Ten redundancy tips", "a redundancy case study" and "redundancy software", each with its own structure, read as three unrelated approaches rather than one considered method. ## How should the pages link to each other? Use three kinds of link, and make every anchor say what the linked page is. Each type does a different job. | Link type | Where it goes | Redundancy example of the anchor text | |---|---|---| | Hierarchical | Pillar to every child, every child back to the pillar | "This covers stage two, selection, of the four-stage process in our redundancy guide." | | Concept-based | Between any pages that share an idea, at the point it comes up | In the settlement article: "how you scored selection, as set out in our guide to selection criteria, affects what a settlement should cover." | | Sequential | From each stage to the next, where there is a natural order | At the end of the consultation article: "Once consultation has started, the next step is choosing who is at risk: see our guide to selection criteria." | Link mid-sentence where the idea comes up, link in both directions, and only link where the connection means something. A block of "related reading" at the bottom tells nobody why the pages belong together, and padding a page with links makes the cluster look less focused. How the rest of your site is organised matters too; our guide to [organising a website so buyers and AI understand it](/insights/website-architecture-ux-and-ai) covers navigation and page hierarchy. ## How do I find the follow-up questions buyers ask about my service? Start with the questions buyers already ask you, then check what the assistants and search engines add. Each follow-up that has a real answer becomes a child page or a section of one. ### Step 1: Mine your own conversations List the questions from sales calls, enquiry emails and proposals. The second and third questions a prospect asks are usually the follow-ups. ### Step 2: Check Google's suggestions Type your main question into Google and note the autocomplete suggestions and the "People also ask" questions. They show the related searches people run. ### Step 3: Ask the assistants what comes next Put the buyer's question into ChatGPT, Gemini and Google AI Mode, note which sub-topics the answers cover, and ask "what would a buyer need to know next?". ### Step 4: Map each question to one page Give each real follow-up question one home: a child page if it needs depth, a section of the pillar if not. Merge any two pages that answer the same question. Our guide to [writing content around the questions customers actually ask](/insights/content-around-customer-questions) covers the research in more detail. ## How should the pillar and the children be written? The pillar defines the subject, names the three to five parts of the framework, gives a short section on each that points to the child going deeper, and covers common mistakes and how to tell it is working. It does not go deep on everything. Each child opens by placing itself in the framework ("stage two of the four-stage process"), answers its question in the first two sentences, goes deep and links to its neighbours. Comparisons, step-by-step guides and worked examples make good children; label any illustrative example as such. Keep a short glossary of the cluster's terms, with one definition each, so every writer uses the same words; our guide to [building a glossary AI assistants use as a reference](/insights/glossary-knowledge-base-ai-references) explains how. ## Worked example: what we changed in our own guides We applied the same test to our own Insights section. The topic on [being named by AI assistants](/insights/pillar/ai-llm-optimisation) has a topic page acting as the pillar and a dozen guides as children. Read in one sitting, they disagreed on points a buyer cares about: one guide said a B2B firm should start with Perplexity and Claude while another said to check every assistant, and guides gave different answers on whether Google rankings carry over and how many questions to track. We settled one answer for each and rewrote every guide to give it. | Question the guides disagreed on | The one answer every guide now gives | |---|---| | Do Google rankings carry over to AI answers? | Ranking helps but does not guarantee a mention; because Google runs several related searches per question, covering the follow-ups matters as well | | Which assistants matter most? | ChatGPT and Google first, then Microsoft Copilot and Gemini, then Perplexity and Claude | | How many buyer questions should we track? | A fixed list of 20 to 50, each asked several times a month in a clean session | | Does schema markup help AI assistants? | It helps search engines read your facts; Google says its AI features need no special markup | Each child now does one job: [how to get ChatGPT to recommend your firm](/insights/what-is-llm-optimisation) is the foundation, [SEO versus AI search](/insights/seo-vs-llm-optimisation) the comparison, and [how AI crawlers read your site](/insights/ai-crawlers-vs-google-spiders) a technical piece. ## What does this look like on a larger site? The same principles hold at scale: a structure that follows how a buyer thinks, and facts that agree across every page. When we rebuilt the website for arum, which advises banks, governments and major brands on collections technology, the result was "310 pages rebuilt across three regions, in weeks", in "3 regional versions: UK and Europe, North America, Australia and New Zealand." The structure followed the buyer: "Services organised the way a client buys them: five stages from advice to optimisation, plus three specialist services." And "Each of the 12 case studies leads with a number, and each of the nine sector pages names the regulation in play." Underneath sat "Proper structure and markup on every page, a plain summary AI assistants can read, facts stored once so two pages never disagree, and every old address redirected so years of rankings carry over." The [arum case study](/work/arum) has the detail. ## How do you tell whether a cluster is working? Run a fixed list of 20 to 50 buyer questions on the subject, each several times a month in a clean session, and record who is named and which of your pages are cited. Answers vary from run to run, so count a win only when it repeats. Our guide to [finding out whether ChatGPT recommends your firm](/insights/track-ai-recommending-your-business) sets out the method. Watch four signs: how often you are named against your closest competitors; whether several cluster pages are cited, not always the same one; answers that repeat your framework back ("there are four stages to a fair redundancy process"); and whether enquiries that start from cluster pages are better qualified. On timing, assistants that search the web live (ChatGPT search, Perplexity, Copilot and Google's AI features) can reflect new or rebuilt pages within weeks. Answers drawn from a model's own memory change only when the model is retrained, which takes months. ## What to do this month 1. Choose one subject you want to be the first name for: one where you have real depth and the work is profitable. 2. List every page you have on it and decide which is the pillar, which are children, and which should be merged, moved or retired. 3. Write down the framework the pillar will set out, and a one-page glossary of the terms every page will use. 4. Collect the follow-up questions from sales calls, Google's suggestions and the assistants, and give each one a single home. 5. Read the cluster in one sitting and fix every contradiction between pages before you write anything new. 6. Record a baseline with 20 of your buyer questions in ChatGPT, Google AI Mode and Copilot, and repeat monthly. ## Frequently asked questions ### How many articles do I need in a cluster? Start with a pillar and four or five children. That covers the main follow-up questions and stays easy to keep consistent. If a subject needs ten or more, split it into two clusters. ### Should each of our service pages be a pillar page? Usually not. A service page explains what you sell and asks for an enquiry; a pillar explains the subject to someone still researching. Link them both ways. A firm with one core service can use its service page as the hub, provided it answers the subject and does not only sell. ### Do topic clusters still work for Google in 2026? Yes. Google says SEO best practices "continue to be relevant because our generative AI features on Google Search are rooted in our core Search ranking and quality systems", and its AI features run several related searches per question, which suits sites that cover the follow-ups. What does not work is a page for every phrasing, which Google treats as scaled content abuse. ### Should our guides sit on a blog or in a separate resources hub? A hub organised by topic, with one page per topic acting as its pillar, is easier for readers and crawlers to follow than a date-ordered blog. The folder name in the address matters far less than the links between topic pages and guides. ### What if my subject does not break down into a pillar and children? Then it may not need a cluster. A narrow subject is often better served by one thorough article, and forcing it into five short pages that repeat each other is worse. Our guide to [why ChatGPT ignores sites full of thin pages](/insights/ai-penalises-thin-content) explains why. ### How often should I update cluster articles? Review each quarter: refresh examples and figures, and date each real change. [Seer Interactive found in July 2026](https://www.seerinteractive.com/insights/study-content-recencys-impact-on-ai-visibility-in-2026) that 75% of pages cited by AI assistants had been updated in the last 12 months, in a study of four consumer and finance industries. Once a year, read the whole cluster together and check the framework still holds. A well-organised cluster is one of the most reliable ways we know to move a firm from sometimes mentioned to named first on a subject. To see how your content reads to buyers and AI assistants today, [book a free Growth Review](/growth-review): 30 minutes with David Adams and three fixes in writing. Rebuilding a site around the questions buyers ask is the work of [Website That Sells](/website-that-sells). ## Sources - [Ahrefs, AI Overview citations from top-ten pages drop from 76% to 38% (March 2026)](https://ahrefs.com/blog/ai-overview-citations-top-10) - [Google Search Central, Optimizing your website for generative AI features on Google Search (July 2026)](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide) - [Google, AI Mode in Search arrives in the UK (July 2025)](https://blog.google/around-the-globe/google-europe/united-kingdom/ai-mode-search-uk/) - [Search Engine Land, ChatGPT citations content study by Kevin Indig (February 2026)](https://searchengineland.com/chatgpt-citations-content-study-469483) - [Seer Interactive, Content recency's impact on AI visibility in 2026 (July 2026)](https://www.seerinteractive.com/insights/study-content-recencys-impact-on-ai-visibility-in-2026) --- Title: How do we audit our existing content for AI assistants? URL: https://fortitudemedia.ai/insights/audit-existing-content-ai-readiness Author: Ashley Kenerson Published: 2026-03-31 Last reviewed: 2026-09-24 Topic: Content that gets quoted Short answer: First check that AI crawlers can reach and read your pages, then record which pages assistants already use. Score each page from one to five on five reasons pages are ignored: specificity, evidence of expertise, structure, point of view and authorship. A total under 10 is fine, 10 to 15 needs targeted fixes, over 15 a rewrite. Start with your twenty most important pages and work through a six-month plan: access and quick wins, then fixes, then rewrites and gaps. The website has been gathering pages for years: service pages, blog posts, news items, a few guides someone commissioned in a good year. Nobody can say which of them do any work. Then a prospect mentions that they asked ChatGPT for firms like yours and the answer named two competitors, and the first instinct is to write something new. Resist it for a month. Ranking on Google is no longer enough on its own: [Ahrefs found in March 2026](https://ahrefs.com/blog/ai-overview-citations-top-10) that only 38% of the pages cited in Google's AI Overviews ranked in the top ten for the search, down from 76% in July 2025, because Google now runs several related searches for one question. Most of what an assistant needs is probably already somewhere on your site, but out of date, badly arranged, unreadable to AI crawlers or buried under pages that say nothing. Before you write, find out what you have, which pages AI assistants already use, which are worth fixing and which should go. That audit needs a spreadsheet, a scoring method and, in our experience, about a day for every twenty pages. Below are our method, a template with a scored example, and a six-month plan for a firm without a content team. ## Key takeaways - Check that AI crawlers can reach and read your pages before you score a single one; a blocked crawler or text that only appears after JavaScript runs makes content quality irrelevant. - Record which pages AI assistants already use, from repeated prompt checks, analytics referrals and server logs, so the audit measures real use and not only quality. - Score each page from one to five on five reasons pages are ignored: specificity, evidence of expertise, structure, point of view and authorship, where five is severe. - A total under 10 is fine, 10 to 15 needs attention on the worst one or two reasons, and over 15 needs a rewrite; these are our working bands, not an industry standard. - Fix point of view and specificity first, because content nobody else could have written counts for most, then expertise, structure and authorship. - Start with the twenty pages that matter most, and work through the results in a six-month plan with one job per month. ## Why should we audit before writing anything new? Because the pages you already have are usually closer to working than a blank page, and weak pages on a subject blur which page is yours. Content written to rank on Google tends to fail in the same ways when an assistant reads it: it says nothing specific, shows no sign of who wrote it, repeats what competitors say and is out of date. Google's [guide to its AI search features](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide), last updated in July 2026, says content with first-hand expertise that could not come from anyone else counts for most, and warns that creating separate pages for every variation of a question can break its scaled content abuse policy. So an audit that merges and deepens usually does more than one that adds. [Why ChatGPT never quotes our blog](/insights/why-ai-ignores-most-blog-posts) explains the five failures in detail. Scale is no reason to skip it. For [TCC Group](/work/tcc-group), a regulatory compliance group with three brands, the audit meant "624 pages reviewed, each given a decision: migrate, merge or retire." What it found was not weak writing: "What failed was underneath: a site organised around internal teams rather than the questions buyers ask, pages that half-answered the same question so neither was cited, and machine-readable data that declared a name, a logo and nothing else." ## What does the scoring method look like? Score each page from one to five on the same five reasons we use in [why ChatGPT never quotes our blog](/insights/why-ai-ignores-most-blog-posts), where one means no problem and five means the problem is severe. Add the five scores for a total between 5 and 25. | Reason | Score 1: no problem | Score 3: partly | Score 5: severe | |---|---|---|---| | Specificity | Mechanisms, trade-offs, dated figures and real examples | Some examples, mostly general | Definitions and generalities anyone could write | | Evidence of expertise | Practitioner examples, cited sources, stated limits, your own data | A few loose references | No sources and no examples | | Structure | Direct answer in the first paragraph, question-led headings, short sections, tables where options are compared | Clear title, uneven headings | Long unbroken text, no clear headings | | Point of view | A position or data competitors could not reproduce | One distinctive angle among familiar material | Says what the top competitors say | | Authorship | Named author with role and experience, a profile page, consistent across related pages | Named author, little context | No byline, or "Admin" | Then read the total against our working bands: | Total | What it means | What to do | |---|---|---| | 5 to 9 | Fine | Keep current: refresh dates and figures on its review cycle | | 10 to 15 | Needs attention | Fix the worst one or two reasons | | 16 to 25 | Needs a rewrite | Rewrite in place, or merge into a stronger page on the same question | These bands are our judgement from audits we have run, not a published standard. Where two pages tie, fix point of view and specificity first, because they move the result most, then evidence of expertise, then structure, then authorship. Two more columns sit alongside the score. Currency records when the page was last reviewed; see [how old is too old for AI assistants](/insights/content-freshness-recency-ai) for review cycles. AI use records whether assistants already draw on the page: a poor page that AI already cites is fixed carefully and in place, and a good page that nothing uses may have an access or linking problem. ## How to run the audit, step by step ### Step 1: List every page with its subject, traffic and importance Export the site's pages from your content system or from Google Search Console. Add columns for the subject in plain words, the last twelve months' traffic and strategic importance, scored high, medium or low by asking three questions: do buyers ask about this, is it central to what you sell, and do we know something about it that competitors do not? This list is the audit's spine. ### Step 2: Check that GPTBot and other AI crawlers can reach your pages An assistant cannot use a page its crawler cannot reach or read. Open yoursite.com/robots.txt and look for rules naming these crawlers: | Crawler | Company | What blocking it does | |---|---|---| | GPTBot | OpenAI | Keeps your pages out of OpenAI's model training | | OAI-SearchBot | OpenAI | Keeps you out of ChatGPT search results | | ClaudeBot and Claude-SearchBot | Anthropic | Keep you out of Claude's training and its search | | PerplexityBot | Perplexity | Keeps you out of Perplexity's index | | Google-Extended | Google | Controls use in Gemini training; it does not affect AI Overviews | [OpenAI says](https://help.openai.com/en/articles/9237897-chatgpt-search) a site must allow OAI-SearchBot to appear in ChatGPT search. Then check your CDN: [Cloudflare changed its default to block AI crawlers in July 2025](https://blog.cloudflare.com/content-independence-day-no-ai-crawl-without-compensation/), so a site behind Cloudflare may be blocking them without anyone having decided to. Finally, view the source of three key pages (not the rendered page) and search for the main heading and first paragraph. [Vercel and MERJ found](https://vercel.com/blog/the-rise-of-the-ai-crawler) that GPTBot, ClaudeBot and PerplexityBot fetch JavaScript files but do not run them, so text that only appears after scripts run is invisible to them. Our [technical checklist for an AI-ready website](/insights/technical-checklist-ai-ready-website) covers the rest. ### Step 3: Find out which pages AI assistants already use Use three sources, because none is complete on its own. - Prompt checks. Keep a fixed list of 20 to 50 buyer questions and ask each several times a month in ChatGPT, Gemini, Claude, Copilot and Perplexity, in a clean session. Record which of your pages is cited. Answers vary from run to run, so count a page as used only when it appears repeatedly. - Analytics. Since 13 May 2026, GA4 has a default "AI Assistant" channel covering ChatGPT, Gemini, DeepSeek, Copilot and Grok. [Claude and Perplexity are not included](https://www.nicelookingdata.com/blog/ga4-ai-traffic-chatgpt-referrals), so add a report filtered on referrers such as perplexity.ai and claude.ai. Clicks from AI Overviews and AI Mode are counted as Organic Search; Search Console's [generative AI performance reports](https://developers.google.com/search/blog/2026/06/gen-ai-performance-reports), available to all sites since 31 August 2026, show those impressions by page. - Server logs. Ask whoever hosts the site for a month of logs and count hits by user agent: GPTBot, OAI-SearchBot, ChatGPT-User, ClaudeBot, Claude-User, PerplexityBot and Perplexity-User. A ChatGPT-User or Claude-User hit means a person's assistant fetched the page to answer a question. Put the result in the AI use column as a short note, for example "cited 3 of 4 runs; 12 referrals; crawled by OAI-SearchBot". ### Step 4: Score the five reasons Read each page against the rubric above and score it one to five on each reason. An AI assistant can draft these scores if you give it the page, the rubric and the top competitor pages, which saves time. A person who knows the subject and the competitors must check every score, especially point of view, which is the hardest to judge and the most valuable. ### Step 5: Add the currency check Record the last reviewed date and flag any page tied to revenue that is more than three months past review, and any other page not reviewed in a year. Judge the content, not only the date: a three-year-old explanation of a principle that has not changed can be fine, while a one-year-old page quoting a superseded regulation is not. ### Step 6: Decide what happens to each page Combine the total with strategic importance, using the decision table below. Write one decision per page: maintain, fix, rewrite, merge or retire. ## Is there a free content audit template for AI search? Yes: copy these columns into a spreadsheet. The row below is a fully scored example for an imagined firm of chartered surveyors, with obviously illustrative figures. | Column | What to enter | Example: /services/dilapidations | |---|---|---| | URL and subject | Address and subject in plain words | /services/dilapidations: what a dilapidations claim involves | | Importance | High, medium or low | High | | Access | Crawlers allowed, text in HTML, indexed by Bing and Google | Yes, yes, yes | | AI use | Prompt checks, referrals, crawler hits | Not cited in 4 runs; 0 referrals; crawled by OAI-SearchBot | | Last reviewed | Date, and flag if overdue | March 2022, overdue | | Specificity (1 to 5) | 5 is severe | 4: defines the term, no worked example | | Evidence of expertise (1 to 5) | 5 is severe | 3: cites one guidance note, no practitioner example | | Structure (1 to 5) | 5 is severe | 3: clear title, answer buried in paragraph four | | Point of view (1 to 5) | 5 is severe | 5: says what the top three competitors say | | Authorship (1 to 5) | 5 is severe | 4: signed "the team" | | Total (5 to 25) | Sum of the five | 19 | | Decision | Maintain, fix, rewrite, merge or retire | Rewrite in place under a named partner; merge in the two older dilapidations posts | ## How do we turn scores into decisions? Put the total on one axis and strategic importance on the other. The time estimates are from our own work and will vary with the subject. | | High importance | Medium importance | Low importance | |---|---|---|---| | Total 5 to 9 | Maintain: refresh figures and dates, a day or two per page | Maintain on the yearly cycle | Leave, or merge if it overlaps | | Total 10 to 15 | Fix the worst one or two reasons, three to five days per page including practitioner time | Fix when capacity allows | Merge or retire | | Total 16 to 25 | Rewrite or rebuild at the same address, five to ten days per page | Rewrite or merge | Retire or redirect | A good page on a subject nobody buys from you is still not worth maintaining. Merge where two pages answer the same question, and redirect the old address to the one you keep. A site with fewer, stronger pages reads as more expert than one with many weak ones, for reasons set out in [why ChatGPT ignores our website when we have hundreds of pages](/insights/ai-penalises-thin-content). The audit will also show gaps: important subjects with no page at all. Those go on the new content list, and they should be the only things on it until the rebuilds are done. ## What does a realistic six-month plan look like? Assume around a hundred pages and no content team. Each month has one job, so nothing overlaps. | Month | Focus | What gets done | |---|---|---| | 1 | Access, baseline and quick wins | Fix crawler, CDN and JavaScript problems; record the AI use baseline; score the top twenty; on every high-importance page, add dates, current figures, a direct answer in the first paragraph and a named author | | 2 and 3 | Targeted fixes | High-importance pages scoring 10 to 15, worst reason first; book practitioner interviews in month one, because their diaries are the bottleneck | | 4 and 5 | Rewrites and gaps | High-importance pages scoring over 15, plus new pages for the most important gaps, linked to each other and to the pages already fixed | | 6 | Medium-importance pages and review | Fix or merge medium-importance pages, then re-run the prompt checks against the baseline and re-score the top twenty | | Every month | Retirement | One or two days merging overlaps and redirecting or removing low-importance pages | [How to organise our content so AI assistants see us as the expert](/insights/topic-clusters-ai-understands) explains how the rebuilt pages should link together, and [how to find out whether ChatGPT is recommending our firm](/insights/track-ai-recommending-your-business) turns the baseline into a monthly routine. ## What to do this month 1. Export the list of pages and add subject, traffic and strategic importance for each. 2. Check robots.txt, your CDN's bot settings and the page source of three key pages, and fix anything that blocks AI crawlers. 3. Record the AI use baseline: run your fixed list of buyer questions, set up the AI referral report in GA4 and ask your host for a month of server logs. 4. Score the twenty most important pages on the five reasons and the currency check, using the template above. 5. Do the quick wins on every high-importance page: dates, figures, a direct opening answer and a named author. 6. Book practitioner time now for the first targeted fixes next month. ## Frequently asked questions ### How long does a content audit take for a 200-page B2B website? In our experience, about twelve working days for one person: a day to export the pages and check access, a day to set up the AI use checks, and about a day for every twenty pages scored. Start with the twenty that matter most, which takes a day or so and shows the patterns that make the rest faster. ### Should I delete or noindex weak pages before optimising for AI search? Decide page by page, and do not do it blind. Check AI use and inbound links first: some weak-looking pages are still cited or linked. Merge overlapping pages and redirect the old address; delete pages that are wrong and have no links or traffic; use noindex only for pages visitors need but searchers do not, since it keeps a page out of Google's index without removing it from your site. On [arum's](/work/arum) rebuild, "Every old address redirected, so years of rankings carried over." ### Can a tool or an AI assistant do the scoring for us? Partly. Tools measure structure, length, dates and crawler access reliably. An AI assistant can draft the five scores if you give it the page, the rubric and the competitor pages, and it is often a good first pass. A person who knows the subject and the competitors must check every score, because judging point of view needs knowledge of what others have published. ### Do we need to score every page, or can we start with some? Start with the twenty pages that matter most: your core services and the pages that still bring in traffic or enquiries. Scoring those shows the site's patterns and often changes the priorities. A full audit can follow over a quarter, but there is nothing to gain by waiting for it before fixing the obvious pages. ### What if we have no data of our own for the point of view score? Point of view can come from judgement as well as data. A page that explains why the standard approach fails in a common situation, or compares options nobody has compared, is distinctive without a single figure. Ask your practitioners what they believe that competitors would argue with. ### How often should we repeat the audit? A full pass once a year, and a quarterly check of the twenty most important pages for currency and AI use. Score every new page against the same five reasons before it is published, so the standard holds. If you would rather have the audit run for you, the [Paid Discovery](/discovery) takes one week and ranks every finding, with fixes written so your own team can ship them; for TCC Group it produced "1 week to a plan TCC's own team built the new site from." Or start with a [free Growth Review](/growth-review): 30 minutes with David Adams and three fixes in writing. ## Sources - [Ahrefs, AI Overview citations from top 10 pages (March 2026)](https://ahrefs.com/blog/ai-overview-citations-top-10) - [Google Search Central, Optimizing your website for generative AI features on Google Search (July 2026)](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide) - [Vercel, The rise of the AI crawler (December 2024)](https://vercel.com/blog/the-rise-of-the-ai-crawler) - [OpenAI Help Center, ChatGPT search (accessed September 2026)](https://help.openai.com/en/articles/9237897-chatgpt-search) - [Cloudflare, Content Independence Day: no AI crawl without compensation (July 2025)](https://blog.cloudflare.com/content-independence-day-no-ai-crawl-without-compensation/) - [Nice Looking Data, GA4 AI traffic and ChatGPT referrals (July 2026)](https://www.nicelookingdata.com/blog/ga4-ai-traffic-chatgpt-referrals) - [Google Search Central, Generative AI performance reports in Search Console (June 2026)](https://developers.google.com/search/blog/2026/06/gen-ai-performance-reports) --- Title: How do we build a simple dashboard to track AI recommendations? URL: https://fortitudemedia.ai/insights/build-ai-visibility-dashboard Author: Ashley Kenerson Published: 2026-03-31 Last reviewed: 2026-09-24 Topic: Costs, returns and the business case Short answer: Use a shared spreadsheet with four tabs: an answers log, enquiries, a work log and a summary. Each month, ask the same 20 to 50 buyer questions in ChatGPT, Gemini, Copilot, Claude and Perplexity, several times each in a fresh chat, and count a win only when it repeats. Keep eight to twelve numbers on the main view, with enquiries and wins at the top and GA4 visits as supporting evidence. Give it one owner, and turn it into a one-page board report each month. You have started the work. The website has new pages answering buyer questions, someone is pitching comment to the trade press, and now and then a prospect mentions they found you through ChatGPT. Then the leadership meeting asks whether it is working, and the evidence is scattered across five places: the CRM, Google Analytics, a spreadsheet of AI answers someone keeps when they remember, an agency PDF and the sales team's memory. Your usual reports will not settle it. A ranking report no longer tells you who AI assistants name: [Ahrefs found in March 2026](https://ahrefs.com/blog/ai-overview-citations-top-10) that only 38% of Google AI Overview citations came from pages ranking in the top ten, down from 76% in July 2025. And the numbers that do matter, such as how often assistants name you and how many enquiries mention them, sit nowhere by default. A dashboard fixes that. It is not software you need to buy. It is one place where a handful of numbers are recorded the same way every month, so you can see the trend and decide what to do. This guide shows what to track, how to build it in a spreadsheet in an afternoon, with the formulas, and how to turn it into a one-page monthly report the board will read. ## Key takeaways - A useful AI search dashboard tracks four things: enquiries and wins, how often you are named, accuracy, and the work done. - Start in a shared spreadsheet, because most founder-led firms never need anything more elaborate for this job. - Fix a list of 20 to 50 buyer questions, ask each several times a month in a fresh chat, and count a win only when it repeats. - Keep eight to twelve numbers on the main view, with results at the top, and show each as a count with a three-month average. - Treat GA4's AI Assistant channel and Search Console's AI impressions as supporting evidence, because most AI-influenced buyers never click a link. - Turn the dashboard into a one-page board report every month, and report a bad month plainly with a cause, a fix and a date. - Give the dashboard one named owner and a fixed update day, or it will go stale by the third month. ## What is the dashboard for, and how is it different from the board report? The dashboard is the working record: the numbers behind the story, including the question-by-question log. The board report is the one page you take to the meeting, picking out what matters this month. Keep them separate but built from the same data. The dashboard has three jobs. It shows whether you are being named more often by AI assistants. It shows whether that is turning into enquiries and clients. And it records what was done, so you can connect cause and effect when something moves. If a number does not help you decide what to do next month, leave it off. ## What should the dashboard track? Four groups of numbers, in this order of importance, plus two supporting figures. | Group | Numbers to record | Why it matters | |---|---|---| | Results | Enquiries that mentioned an AI assistant, how many were qualified, their pipeline value, clients won | These justify the budget, though they may be zero for the first few months | | Recommendations | Tracked questions each assistant names you for, questions where you are named first, the same count for your leading competitor | These move before enquiries do | | Accuracy | Answers that describe your firm wrongly | One wrong answer repeated to every buyer can cost more than being absent | | Work done | Pages published, coverage earned, listings corrected, new reviews and case studies | Inputs that explain movement above, so they sit at the bottom | | Supporting | Visits from AI assistants in GA4, and AI Overview and AI Mode impressions in Search Console | Useful context, never the headline | Our guide to [why ChatGPT describes a firm wrongly](/insights/ai-conflicting-information-online) explains where accuracy errors come from and how to correct them. ## How to build it, step by step You can build a working version in an afternoon. The discipline is in recording the same things the same way every month. ### Step 1: Fix the buyer questions you will track Choose 20 to 50 questions your buyers ask, in their words, taken from sales calls and enquiry emails. Mix shortlist questions ("best employment lawyers for tech firms in Leeds") with problem questions ("how do we handle a grievance against a director?"), and never include your own firm's name. Once the list is set, do not edit it; add new questions in a separate section. [Tracking whether AI tools recommend your business](/insights/track-ai-recommending-your-business) has more on choosing them. ### Step 2: Set up four tabs - **Answers log.** One row per question, per assistant, per month, with columns for month, question, assistant, run 1, run 2, run 3 (Y or N for named), named on repeat, named first, accurate, and competitors named. - **Enquiries.** One row per enquiry: date, source, whether AI was mentioned, qualified or not, pipeline value and outcome. Link to your CRM if it holds this. - **Work log.** One row per piece of work: what, where and when. - **Summary.** The main view, calculated from the other three. ### Step 3: Add the formulas In Google Sheets or Excel, give the month, assistant and named-on-repeat columns the names Month, Assistant and Repeat (in Google Sheets, Data, then Named ranges; in Excel, Formulas, then Define Name) so formulas stay readable. Three formulas do most of the work: - **Named on repeat**, in each row: `=IF(COUNTIF(D2:F2,"Y")>=2,"Y","N")`. A question counts only if you were named in at least two of three runs, because answers vary from run to run. - **Share of answers** for one assistant in one month: `=COUNTIFS(Month,"2026-09",Assistant,"ChatGPT",Repeat,"Y")`, shown beside the number of questions asked, never as a bare percentage. - **Three-month average** on the summary tab: `=AVERAGE(B5:D5)` across the last three monthly columns. That is the whole template. You can copy the column list above into a blank sheet and be recording within the hour. ### Step 4: Record the answers each month Put every tracked question to ChatGPT, Gemini, Copilot, Claude and Perplexity, and check whether Google shows an AI Overview or AI Mode answer that names you. Use a fresh chat or private session each time. This is how we track client results: for ID-Pal, "The same buyer questions run every month in ChatGPT, Gemini and Claude, in a fresh chat. A win only counts when it repeats, and losses are reported too." By hand this takes a few hours a month; a tracking tool cuts it to minutes. ### Step 5: Capture where enquiries come from This is the step most firms skip and the one the board cares about most. Add "How did you first hear about us?" with an AI assistant option to your enquiry form, ask the same on the first call and record the answer in the CRM. Our guide to [tracking which leads came from ChatGPT](/insights/attribution-ai-era-track-leads) covers the detail, including AI-sourced enquiries that arrive looking like Direct traffic. ### Step 6: Pull the website figures Since 13 May 2026, GA4 has had a default [AI Assistant channel](https://support.google.com/analytics/answer/9756891?hl=en) covering ChatGPT, Gemini, Copilot, Grok and DeepSeek, as [Nice Looking Data reported in July 2026](https://www.nicelookingdata.com/blog/ga4-ai-traffic-chatgpt-referrals). Perplexity and Claude still appear in Referral, so add them with a custom channel group or note them separately. Clicks from Google's AI Overviews and AI Mode count as Organic Search. For those, use Search Console's [generative AI performance reports](https://developers.google.com/search/blog/2026/06/gen-ai-performance-reports), live for all sites since 31 August 2026, which show impressions and pages, though not clicks. Keep each to one line. ### Step 7: Build the summary view Results at the top, recommendations beneath, then accuracy, work done and supporting figures. Show this month, the three-month average and the year to date. Add a short box for commentary: what changed and why. Colour a cell green, amber or red only where there is an agreed target. ## What should the main view look like? Here is an example for a professional services firm six months in. Every figure is illustrative. | Measure (example figures, 30 fixed questions) | This month | 3-month average | Year to date | |---|---|---|---| | Enquiries mentioning AI | 2 | 1.7 | 9 | | of which qualified | 2 | 1.3 | 6 | | Pipeline from those enquiries | £35,000 | n/a | £140,000 | | Clients won | 0 | n/a | 2 | | Named by ChatGPT | 9 of 30 | 8 | n/a | | Named by Gemini | 7 of 30 | 6 | n/a | | Named by Copilot | 5 of 30 | 5 | n/a | | Named by Claude | 6 of 30 | 5 | n/a | | Named by Perplexity | 11 of 30 | 10 | n/a | | Named in Google AI Overviews | 4 of 30 | 4 | n/a | | Named first, any assistant | 4 of 30 | 3 | n/a | | Leading competitor named, any assistant | 20 of 30 | 20 | n/a | | Answers with a wrong fact about us | 1 (old service line, correction under way) | 1 | n/a | | Work: pages, press comments, listings, reviews | 2, 1, 3, 4 | n/a | n/a | | Supporting: visits from AI assistants in GA4 | 46 | 38 | n/a | Notice what it leaves out: word counts, social impressions and scores from SEO tools. ## Which tool should we use? Start with Google Sheets or Excel. For most founder-led firms with £1m to £20m+ revenue, a well-kept spreadsheet is all this dashboard needs to be. | Option | Good for | Watch out for | |---|---|---| | Google Sheets or Excel | The dashboard itself; free or already paid for | Needs one owner and a fixed update day | | Looker Studio | Charts that update from the sheet and GA4 | Worth it only once the sheet is stable | | AI visibility trackers, such as Otterly, Peec AI, or the AI features in Semrush and SE Ranking | Running the questions for you across assistants and competitors | Check which assistants each covers, how many runs per question, and that you can export the raw answers | | Search Console and GA4 | Google AI impressions and AI referral visits | Supporting evidence only | The hard part is collecting honest data every month, not drawing charts. ## How do we turn the dashboard into a board report? Take one page from it each month, in six short sections: status, results, leading indicators, what we did, what did not go to plan, and next month. Results come first because that is what the board came for; activity comes last because it only explains the rest. Using the example above: | Section | What it says (example) | |---|---| | Status | Amber. Named more often; enquiries steady. | | Results | 2 enquiries mentioned an AI assistant (last month 1), both qualified; £35,000 of pipeline; 2 clients won from AI-sourced enquiries this year | | Leading indicators | Perplexity names us for 11 of 30 fixed questions and ChatGPT for 9 (three-month averages 10 and 8); leading competitor named for 20; one answer lists an old service line, correction under way | | Supporting evidence | 46 visits from AI assistants in GA4 (three-month average 38) | | What we did | 2 pages published, 1 trade press comment, 3 listings corrected, 4 new reviews | | What did not go to plan | Manufacturing case study delayed by client approval; now due next month | | Next month | Publish the case study; pitch two trade titles; no board decision needed | Helm Club's weekly numbers work the same way: "Qualified calls a week, reported on one page every Monday and checked against Helm's own member records rather than a platform dashboard." ([Helm Club case study](/work/helm-club)) One page, your own records, the same measures every time. ### How to report a bad month Plainly, early and with a plan, using the same four lines every time (example): - **What happened.** "We were named for 7 of 30 questions, down from 11." - **Why.** "Two answers now cite a competitor's new salary survey, which replaced our older guide." - **What we are doing.** "Updating our guide with this year's figures and pitching the findings to a trade title." - **When we will know.** "By the December report. If not, we will bring a revised plan." If a number is down three months running, that is not noise. It is a trend, and the board should hear it from you. ### How to run the monthly review Twenty minutes on the leadership agenda is enough. Send the page two days before. Spend two minutes on status and results, and ask the sales lead whether the AI-sourced enquiries were good ones. Discuss one issue, not ten. Record who does what by the next report. Every third month, add the full competitor comparison from [benchmarking your AI share of voice](/insights/benchmarking-ai-visibility-competitors) and decide whether to keep, change or grow the budget. ## Mistakes that make a dashboard useless - **Too many numbers.** If the main view does not fit on one screen, cut it. - **A changing question set.** Edit the questions and every trend line becomes meaningless. - **Percentages without counts.** "Named in 30 per cent of answers" hides whether that is 3 of 10 or 15 of 50. - **Counting single runs.** One good answer is not a win until it repeats. - **Effort dressed as results.** Pages published is not a result. Put it at the bottom. - **No owner or fixed day.** Dashboards die from neglect, usually around month three. Name one person and put the update in the diary. ## What to do this month 1. Agree 20 to 50 buyer questions with your sales team and lock the list. 2. Add an AI assistant option to the "How did you first hear about us?" question on your form and in the CRM. 3. Set up the four tabs and formulas, and record this month's answers as your baseline. 4. Name one owner and a fixed monthly update day. 5. Take the first one-page report to your next leadership meeting, even if most results are still zero. ## Frequently asked questions ### Is there a free template for tracking AI visibility? This guide gives you one: the four tabs, the answers log columns in Step 2 and the formulas in Step 3. Build it in a blank Google Sheet or Excel file. Keep it in a file your firm owns, not in an agency's tool, so the history stays with you if you change provider. ### Can we build an AI visibility dashboard in Looker Studio? Yes. Looker Studio is Google's free reporting tool and can read a Google Sheet and GA4 directly, so the sheet stays the source and Looker Studio draws the charts. Do it once the spreadsheet has run for two or three months, because charts built on an unstable sheet break. ### Which KPIs matter most for AI search? Enquiries and clients that came through AI answers, then the share of your fixed questions where you are named on repeat, compared with your leading competitor. Accuracy comes next. Visits from AI assistants and Search Console AI impressions are supporting figures. ### What is a good AI share of voice for a B2B firm? There is no published benchmark we trust for B2B services, and it depends on how crowded your market is. Compare yourself with your leading competitor on the same questions, and judge your own three-month trend. A rising share against a steady competitor is a good sign. ### How do we track Google AI Overviews for our key searches? Search each tracked question in Google and record whether an AI Overview appears and whether it names or cites you. Search Console's generative AI reports then show which of your pages appeared in AI Overviews and AI Mode, and how often. Some trackers also check AI Overviews for you. ### How much time does the dashboard take each month? By hand, a few hours: most of it running the questions and recording the answers, then half an hour on enquiries and commentary. A tracking tool cuts the answers work to minutes. Enquiry data takes little time if your team records the source when each enquiry arrives. ### Should we track Google rankings on the same dashboard? One line is fine if your team already watches rankings, but keep it below the AI measures. Rankings and AI answers can move independently, so a steady ranking can hide falling visibility in AI answers, and the reverse. The first number on any dashboard is your starting point. If you want a baseline before you build the spreadsheet, [book a free Growth Review](/growth-review): 30 minutes with David Adams, and three fixes in writing afterwards. ## Sources - [Ahrefs, AI Overview citations from top-ten pages (March 2026)](https://ahrefs.com/blog/ai-overview-citations-top-10) - [Google Analytics Help, Default channel group (September 2026)](https://support.google.com/analytics/answer/9756891?hl=en) - [Google Search Central, Introducing Search generative AI performance reports in Search Console (June 2026)](https://developers.google.com/search/blog/2026/06/gen-ai-performance-reports) - [Nice Looking Data, AI traffic in GA4: how ChatGPT, Perplexity, Gemini and Google AI Mode referrals show up (July 2026)](https://www.nicelookingdata.com/blog/ga4-ai-traffic-chatgpt-referrals) --- Title: How do we choose a firm to help us get recommended by ChatGPT? URL: https://fortitudemedia.ai/insights/how-to-compare-ai-optimisation-providers Author: David Adams Published: 2026-03-31 Last reviewed: 2026-09-24 Topic: Costs, returns and the business case Short answer: Choose a provider that shows you what ChatGPT, Gemini, Copilot, Claude and Perplexity say about you before any work starts, and can name exactly what it will change on your website, in your content and in independent coverage. Ask who does the work and for a sample monthly report that repeats the same buyer questions. Nobody can guarantee what an assistant says, so accept guarantees of effort, reporting and ownership only, and avoid long lock-ins. If you run a B2B firm, your inbox has probably filled up with offers to "get you into ChatGPT". Your SEO agency has added a paragraph about AI to its proposal. A software company wants to sell you a dashboard. A PR firm says coverage is what AI assistants read. Some call the work GEO (generative engine optimisation) or AEO (answer engine optimisation); others call it AI SEO. They all mean the same thing: getting your firm named when buyers ask ChatGPT, Gemini, Copilot, Claude or Perplexity who to hire. The labels matter less than they seem. Google's own [guide to optimising for its generative AI features](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide) (May 2026, updated July 2026) names both GEO and AEO and says that, from Google's side, the work is still SEO. UK searchers mostly use the plainer term: over the 12 months to 24 September 2026, "AI SEO" drew about nine times the UK search interest of "generative engine optimization" on [Google Trends](https://trends.google.com/trends/explore?date=today%2012-m&geo=GB&q=AI%20SEO,generative%20engine%20optimization,answer%20engine%20optimization,AI%20visibility,LLM%20SEO). Whatever a provider calls it, judge what it will do. Most founders we speak to have been here before with other agencies: a retainer, a monthly report full of charts, and no customer anyone could trace to it. We sell this kind of work ourselves, so read what follows with that in mind. It is the guide we would want on your side of the table, and it applies to any provider, including us. ## Key takeaways - GEO, AEO and AI SEO are different names for the same work, and Google says it is still SEO, so judge providers on what they will change, not on the label. - Being named by AI assistants depends on your website, your content and what others say about you, so check a provider covers all three or says who will. - Ask every provider to show a baseline of what ChatGPT, Gemini, Copilot, Claude and Perplexity say about you before any work starts. - Nobody can guarantee what an AI assistant will say, but a provider can guarantee effort, reporting and outcomes it controls, such as meetings held. - Good reporting asks the same buyer questions every month, counts a win only when it repeats, and shows the sources the assistants cited. - Short notice periods, a defined first phase and named people doing the work protect you better than a long contract. ## What are you actually buying? Three kinds of work and a way of measuring them: - **A website AI assistants can read and understand.** Pages that load properly for automated readers, say plainly what you do and for whom, and state key facts consistently, including in structured data (a small block of code that labels facts for machines). - **Content that answers your buyers' real questions.** Service pages, guides and case studies that a buyer, and an assistant, could quote. - **Independent mentions.** Trade coverage, directory profiles, reviews and awards that confirm what your website says. This is where most citations come from: [Muck Rack's May 2026 study](https://www.globenewswire.com/news-release/2026/05/07/3290268/0/en/generative-pulse-earned-media-consistently-drives-ai-citations-holding-at-84.html) of more than 25 million links cited by ChatGPT, Claude and Gemini found 84% were earned media. On top of that you need a consistent way of checking what assistants say about you. Our guide on [whether you still need SEO now buyers ask ChatGPT](/insights/seo-vs-llm-optimisation) explains how the old work and the new overlap. ## Which kind of provider covers which part? Most providers grew out of one of the three, which is fine as long as nobody sells their part as the whole job. | Provider type | Usually strong on | Usually misses | How they usually charge | |---|---|---|---| | SEO agency | Website structure, technical fixes, rankings | Earned coverage and reviews; reporting on AI answers | Monthly retainer | | PR firm | Trade coverage and journalist relationships | Website fixes; measuring AI answers | Monthly retainer or project | | Specialist GEO or AI search agency | Baselines of AI answers, content built around buyer questions | Varies widely: some are rebadged SEO, some have no PR capability | Audit, then monthly retainer | | Content agency | Volume of articles and pages | Independent mentions; measurement | Per piece or monthly | | Tracking software | Asking the questions and recording answers at scale | Doing any of the work | Monthly subscription | | Freelancer | Specialist skill at lower overhead | Breadth and cover when they are away | Day rate or monthly | The first test of any proposal is whether it covers all three strands, or says clearly who else will. For what each of these typically costs in the UK, see [how much AI search optimisation costs](/insights/ai-search-optimisation-cost-uk). ## What questions should you ask every provider? Ask these seven in the first meeting, or copy them into an email to each firm on your shortlist. The quality of the answers will tell you more than any credentials slide. ### 1. What do AI assistants say about us today? A capable provider will have checked before the meeting, or will offer to. They should show you the actual answers to questions your buyers ask, who was named instead of you, and which sources were cited. A good baseline looks like the one we built for [TCC Group](/work/tcc-group): "36 recorded answers from ChatGPT, Gemini and Claude, every conversation kept and reproducible." If a provider cannot tell you where you stand now, it cannot show you progress later. ### 2. Which buyer questions will you track, and who chose them? The questions matter more than the tool. They should be the ones your buyers type, such as "best employment lawyers for tech firms in Bristol", not questions containing your own name. Agree a list of 20 to 50 together and keep it stable, so that month-to-month comparisons mean something. ### 3. What will you change on our website, and why? Expect specifics: which pages, what is missing, whether AI crawlers can reach the site, which facts disagree between pages. Vague answers about "improving content" suggest a template. ### 4. How will you earn independent mentions? Ask which publications, directories and review sites matter in your sector, who writes the pitches, and for linked examples of coverage won for similar firms. Paid placements and bulk outreach are a different thing, as our guide to [earned, owned and paid coverage](/insights/earned-owned-paid-authority-signals) explains. ### 5. Who will actually do the work? Ask for names, experience and hours a month on your account, whether any work is subcontracted, and how AI writing tools are used. Using them to help draft is normal. Publishing their output without an expert's input is not. ### 6. How will we know it is working, and when? A good answer separates early signs (fixes made, coverage secured) from later results (being named more often, buyers mentioning AI when they get in touch). Assistants that search the web live can reflect a new page or article within weeks; answers drawn from a model's memory change only when it is retrained, which takes months. ### 7. What happens if we leave? You should own your content, website changes, accounts and tracking data. Check the notice period and whether anything stops working when the contract ends. The best outcome is work your own team can carry on: for TCC Group, "TCC chose to build it with its own team, and the new tcc.group is live, built to the plan, with nothing about it depending on Fortitude." ## What can a provider fairly guarantee? Only what it controls. Nobody controls what ChatGPT or Gemini says, so be wary of a guarantee of a set number of citations, rankings inside AI answers or AI-sourced leads: it usually means the provider misunderstands the work or plans to measure something easy. A provider can fairly guarantee effort (named people, hours, pages and pitches delivered), reporting (the same questions, every month, losses included), your ownership of everything produced, and outcomes it controls directly. Our outbound service, [Pipeline That Fills](/pipeline-that-fills), carries a guarantee tied to meetings held, because we run that work end to end. We make no guarantee about AI answers, and neither should anyone else. ## Which red flags should make you wary? - **A guaranteed place in AI answers.** Ask exactly what is guaranteed and how it will be measured. Guarantees of effort, reporting and meetings held are fair; nobody can promise what an assistant will say. - **"We'll get you into ChatGPT's training data."** Models are trained on large snapshots of the web taken before release. You can influence future snapshots by being well covered and easy to read, but nobody can insert you. - **A special relationship with an AI company.** ChatGPT has shown labelled ads to UK users on its Free and Go plans since 11 August 2026, and [OpenAI says](https://openai.com/index/testing-ads-in-chatgpt/) the ads do not change the answer itself. Buying an advert beside the answer is not the same as being named in it. - **llms.txt or special markup sold as the fix.** Google says its search ignores llms.txt and that its [AI features need no special markup](https://developers.google.com/search/docs/appearance/ai-features). Both are cheap housekeeping, not a strategy. - **A page for every variation of a question.** Google's guide warns that mass-producing pages for query variations, mainly to influence AI answers, breaks its spam policy on scaled content. - **Content at volume with no expert input.** Dozens of AI-drafted articles a month give assistants nothing new to quote. - **Paid placements sold as PR.** Sponsored articles and pay-to-enter lists are advertising. - **The old SEO report with a new name.** Keyword rankings and traffic charts do not tell you what AI assistants say. - **Long lock-ins.** A twelve-month minimum with exit penalties puts all the risk on you. ## What does good reporting look like? The same buyer questions every month, what each assistant said, and what the provider did that might explain any change. The method we use for [ID-Pal](/work/id-pal) is the standard to ask for: "The same buyer questions run every month in ChatGPT, Gemini and Claude, in a fresh chat. A win only counts when it repeats, and losses are reported too." | A report worth paying for | A report to be wary of | |---|---| | The agreed questions, asked more than once in each assistant | A single "visibility score" with no working | | For each question: named or not, how described, which competitors instead | Screenshots of one good answer | | The sources the assistants cited | Share of voice with no questions behind it | | Work done that month, with links and reasons | Long activity lists linked to nothing | | Enquiries where the buyer mentioned an AI assistant | Impressions and reach as the headline | | A short plan for next month you can challenge | Every month green, nothing ever slipped | Our guide to [building an AI visibility dashboard and board report](/insights/build-ai-visibility-dashboard) shows the format in full, whether you run the reporting or a provider does. ## How do you compare proposals side by side? Put the proposals on one page and score them on the same things. Price matters, but only once you know what each price buys. | What to compare | What good looks like | Warning sign | |---|---|---| | Starting point | Real answers from several assistants, shown to you | A generic audit or a score with no working | | Buyer questions | Agreed with you, in your buyers' words | Chosen by the provider, or containing your name | | Website work | Named pages and specific changes | General promises with no detail | | Independent mentions | Named publications and linked past examples | Paid placements or bulk outreach only | | Team | Named people with relevant experience | A sales lead you never see again | | Reporting | Sample report shown before signing | "We will tailor it once you start" | | Guarantees | Effort, reporting and ownership | A guaranteed place in AI answers | | Terms | Short notice period, and you own everything | Long minimum term and exit penalties | Score each row from one to three and add them up. When two proposals score the same, the tiebreaker is usually the people, so meet whoever will run your account. ## How should the first three months be structured? Start with a defined first phase rather than a long commitment. Three months is enough to test whether a provider does what it says, though not to judge the full effect on your pipeline. By the end you should have a baseline, the main website fixes done and explained, a content plan tied to buyer questions with the first pieces live, the first independent coverage or a named list of targets, and monthly reports asking the same questions. If you want the diagnosis before choosing anyone, a fixed-scope piece of work such as our [Paid Discovery](/discovery) gives you every finding ranked and fixes written so your own team, or any provider, can ship them. ## What to do this month 1. Write down 20 questions your best clients would have asked before they found you. 2. Ask them in ChatGPT, Gemini, Copilot, Claude and Perplexity, more than once each, and save the answers as your own baseline. 3. Send the seven questions above to each provider you are considering, and ask each for a sample monthly report. 4. Ask each for two client references, including one who has left. 5. Score the proposals in the table and meet the person who would run your account. ## Frequently asked questions ### How much does a GEO agency cost in the UK? It depends on how much of the work is included: an audit, a content programme, earned coverage, tracking or all four. Compare what each proposal buys each month in named people's hours, pages changed and coverage pursued. Our guide to [what AI search optimisation costs in the UK](/insights/ai-search-optimisation-cost-uk) sets out published market ranges, and our own prices are on [what it costs](/what-it-costs). ### Can a GEO agency guarantee ChatGPT will mention us? No. Nobody controls what an AI assistant says, and answers vary from one run to the next. An agency can guarantee effort, reporting and your ownership of the work. Treat a guaranteed mention as a reason to ask how they will measure it, because it usually means they plan to measure something easy. ### What contract length is normal for an AI search agency? Terms vary, and there is no industry standard. In our view, a defined first phase of about three months, then monthly terms with a short notice period, suits this work better than a twelve-month minimum. Judge the work at three months and the results over six to twelve. ### Should we hire a GEO specialist or train our own marketer? If you have a capable marketer with time, a clear plan and a tracking method may be enough, with outside help for press relationships and technical fixes. If you have no marketing team, you will need someone to do the work. A middle route is a written plan your team builds from: for TCC Group, the result was "1 week to a plan TCC's own team built the new site from." ### Can our existing SEO agency do this work? Possibly. Much of the website work overlaps with good SEO. Ask them the same seven questions, especially how they will earn independent mentions and report what AI assistants say. If the answer is a renamed keyword report, you need a different provider or a specialist alongside them. ### Is a software tool enough on its own? A tool tells you what assistants say, which is valuable, but it does not change it. For some firms a tool plus a capable in-house marketer is enough. For most founder-led firms without a marketing team, the tool is the starting point and the work still needs doing. ### How do we check a provider's claims about past results? Ask for the client's name, the questions tracked and the answers before and after, then speak to that client yourself. Ask to speak to a client who has left, and find out why. Be wary of percentage improvements with no baseline, and of results you cannot trace to a named firm. Whoever you choose, start with your own baseline so you can judge every proposal against it. [Book a free Growth Review](/growth-review): 30 minutes with David Adams, and three fixes in writing that are yours to take into any conversation, including with other providers. For more on when outside help is worth paying for, see [whether you still need a marketing agency](/insights/ai-disrupting-marketing-agency-model). ## Sources - [Google Search Central, Optimizing your website for generative AI features on Google Search (July 2026)](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide) - [Google Trends, UK search interest in AI SEO, GEO and AEO terms (September 2026)](https://trends.google.com/trends/explore?date=today%2012-m&geo=GB&q=AI%20SEO,generative%20engine%20optimization,answer%20engine%20optimization,AI%20visibility,LLM%20SEO) - [Muck Rack, What is AI reading? Earned media drives 84% of AI citations (May 2026)](https://www.globenewswire.com/news-release/2026/05/07/3290268/0/en/generative-pulse-earned-media-consistently-drives-ai-citations-holding-at-84.html) - [OpenAI, Testing ads in ChatGPT (updated August 2026)](https://openai.com/index/testing-ads-in-chatgpt/) - [Google Search Central, AI features and your website (December 2025)](https://developers.google.com/search/docs/appearance/ai-features) --- Title: How do we find out whether ChatGPT is recommending our firm? URL: https://fortitudemedia.ai/insights/track-ai-recommending-your-business Author: Ashley Kenerson Published: 2026-03-31 Last reviewed: 2026-09-24 Topic: Costs, returns and the business case Short answer: Ask the assistants your buyers use the questions they ask, and record whether you are named, how you are described and which sources are cited. Use a fixed list of 20 to 50 questions, a dedicated UK account with memory off, and ask each question several times a month, counting a win only when it repeats. GA4 and Search Console now show AI visits and AI Overview impressions, but not how often you are named. Somewhere in your CRM there is probably a note that says "found us via ChatGPT". Maybe a new client mentioned it on the first call. Maybe your sales director heard it from a prospect who went elsewhere, because the assistant named three firms and you were not one of them. Either way, buyers are asking AI assistants who to hire, and you do not know how often you are in the answer. They are asking in numbers. In [G2's April 2026 survey of 1,076 software buyers](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html), 51% said they now start their research in an AI chatbot, up from 29% a year earlier. That is a survey of software buyers, but the wider picture is similar: [Forrester found in January 2026](https://www.forrester.com/blogs/b2b_buyers_make_zero_click_buying_number_one/) that 94% of business buyers use AI somewhere in their buying process. With Google you could see where you ranked. Nothing tells you how often ChatGPT, Gemini, Copilot or Perplexity recommended you last month, or who was named instead. So most founders ask ChatGPT once, then relax or panic, and a single answer justifies neither. Below is a free 15-minute check you can run today, then the repeatable method we use for tracking. ## Key takeaways - No report shows how often AI assistants name your firm, so you have to ask them yourself, systematically; GA4 and Search Console now show only part of the picture. - Track a fixed list of 20 to 50 questions your buyers actually ask, in their words, and leave your own name out of most of them. - Answers vary from run to run, so ask each question several times a month in a clean session and count a win only when it repeats. - Use a dedicated UK account with memory off, and record the date, the assistant and the model for every answer. - Record how you are described and which sources are cited, because those tell you what to fix next. - Tools such as Otterly, Peec AI, Profound, Semrush and Ahrefs Brand Radar automate the asking once a spreadsheet becomes a chore. ## Can we check what ChatGPT says about us in 15 minutes? Yes. Open ChatGPT in a Temporary Chat set to unpersonalised, which [does not use memory or custom instructions](https://help.openai.com/en/articles/8914046-temporary-chat-faq), and paste five questions a buyer might ask. Then run the same five in Google (looking for an AI Overview), Copilot and Gemini. For example, a managed IT firm in the Midlands would try these, and you would swap in your own service, client type and region: 1. **Shortlist:** "Which managed IT providers would you recommend for a 100-person manufacturer in the West Midlands?" 2. **Problem:** "How do we switch IT support provider without downtime?" 3. **Comparison:** "Outsourced IT support or an in-house IT manager for a 100-person firm?" 4. **Reputation:** your firm's name followed by "any good for manufacturers?" 5. **Head to head:** your closest competitor's name, "or", your name, then "for a 100-person manufacturer?" A good answer names you in the shortlist questions, describes what you do accurately and in current terms, gives a reason that matches the buyer's need, and cites a source you recognise, ideally independent coverage as well as your own site. Warning signs are hedges such as "according to its website", an old service or office, or a competitor named with a reason you could equally claim. Fifteen minutes will not give you a number you can trust, but it will tell you whether the full method is urgent. ## Why does a single check tell us so little? Because AI answers change from one run to the next. Ask ChatGPT the same question three times and you may get three different shortlists: each answer is written fresh, with an element of chance in the wording and sometimes in the firms named. Several other things move the answer: - **Who is asking.** Assistants may use saved memory, past conversations and account settings, so your own account can lean towards your firm. - **Where they are.** OpenAI says ChatGPT search ["may use an approximate location based on your IP address to provide relevant local results"](https://help.openai.com/en/articles/9237897-chatgpt-search). - **Which assistant.** ChatGPT, Google's AI features, Copilot, Gemini, Perplexity and Claude draw on different sources and often name different firms. - **Whether it searched.** For some questions the assistant searches the web live; for others it answers from what the model learned in training. - **Model updates and small changes in wording**, either of which can change the list. None of this makes tracking pointless. It means measuring the way a pollster does: the same questions, asked several times, in a controlled way. Patterns across many answers are reliable even when single answers are not. ## What should we measure? Measure how often you are named, how you are described and why. That comes down to six things for each question and assistant: - **Named or not.** Did the answer mention your firm at all? - **Prominence.** Main recommendation, one of a shortlist, or a passing mention? - **Description.** Accurate, current and specific, or vague and hedged? - **Competitors.** Which other firms were named, and in what order? - **Sources.** Which websites, articles and directories were cited or linked? - **Searched or not.** Did the assistant search the web for this answer? That tells you whether new coverage can change it quickly. The headline number is your share of answers: out of all the answers you collected, the proportion that named you. For example, ask 30 questions three times each in four assistants and you have 360 answers. If you are named in 36 of them, you appear in 10% of answers. Track that figure monthly, alongside the same figure for your main competitors. ## How do we set up tracking? Setting up takes an afternoon; running it takes a few hours a month. [ID-Pal](/work/id-pal) is measured this way: "The same buyer questions run every month in ChatGPT, Gemini and Claude, in a fresh chat. A win only counts when it repeats, and losses are reported too." ### Step 1: Write the questions your buyers actually ask Take them from sales calls and enquiry emails, in the buyer's words rather than your marketing language. Mix shortlist, problem, comparison and requirement questions ("which IT firms can help us get Cyber Essentials Plus?"), and add a few reputation questions that do include your name. Keep to a fixed list of 20 to 50: a short list that gets run beats a long one nobody runs. ### Step 2: Choose the assistants your buyers use Start with ChatGPT and Google's AI Overviews and AI Mode, then add Microsoft Copilot and Gemini, then Perplexity and Claude. Do not skip Copilot if your buyers work in Microsoft 365: Forrester found business buyers four times as likely as consumers to use it. Our guide to [which AI assistants buyers actually use](/insights/ai-search-landscape-chatgpt-perplexity-google-claude) explains how they differ. ### Step 3: Ask in a clean, consistent way Set up a dedicated account for tracking, used for nothing else, with memory and custom instructions switched off, and run it from a UK connection. In ChatGPT, use a Temporary Chat set to unpersonalised. We do not rely on logged-out sessions: what a logged-out visitor sees may differ from what a signed-in buyer sees, and you cannot always tell which model answered. Paste each question exactly as written, and note the date, the assistant and the model name shown in the app. For Google, search in a private window and note whether an AI Overview appeared at all. ### Step 4: Record every answer in one sheet Use one row per answer. These columns are enough, and you can copy them straight into a spreadsheet: | Column | What to enter | |---|---| | Date | The day you asked | | Assistant and model | For example ChatGPT and the model name shown | | Question | Exactly as asked, from your fixed list | | Run | 1, 2 or 3 | | Named | Yes or No | | Prominence | First, shortlist or passing | | Description | What it said about you, word for word | | Competitors named | In the order given | | Sources cited | Every URL shown | | Searched | Yes or No | Save the full answer text alongside it. Your share of answers is then `=COUNTIF(E:E,"Yes")/(COUNTA(E:E)-1)`, where column E is "Named" and the minus one skips the header row. Count a question as won only when you are named in at least two of its three runs. ### Step 5: Run a baseline Ask every question in every assistant, three times each, before you change anything. If you appear once in three runs, you are on the edge of the shortlist. Without a baseline you cannot show progress later. ### Step 6: Repeat on a fixed rhythm Run the list several times a month, for example once a week, and report monthly. When a major model update is announced, note it in the sheet so you can explain sudden shifts. ## Which tools track AI recommendations, and what do they cost? Several tools now ask your questions automatically, repeat them and record sources. They save hours once the manual method becomes a chore, but choose one that lets you enter your own buyer questions, shows the actual answers rather than just a score, and exports the data. | Tool | Your own questions | Assistants covered (as the vendor lists them) | Competitors and sources | |---|---|---|---| | [Otterly.AI](https://otterly.ai/) | Yes, a prompt library | ChatGPT, Google AI Overviews, Perplexity and Copilot on every plan; AI Mode, Gemini and Claude as add-ons | Yes, including a share of voice measure | | [Peec AI](https://peec.ai/) | Yes | ChatGPT, Perplexity and Gemini, with other engines listed | Yes, with source use by domain and URL | | [Profound](https://www.tryprofound.com/) | Not stated on its home page | ChatGPT, Perplexity, Claude, Gemini, Copilot, DeepSeek and AI Overviews | Yes; built for large enterprise teams | | [Semrush AI Visibility Toolkit](https://www.semrush.com/blog/how-to-track-your-chatgpt-visibility/) | Yes, through prompt tracking | ChatGPT covered in its guide | Yes, including domains that cite competitors but not you | | [Ahrefs Brand Radar](https://ahrefs.com/brand-radar) | Yes, as custom prompts | AI Overviews, AI Mode, ChatGPT, Perplexity, Gemini, Copilot and Claude | Yes, including AI share of voice | Prices change often, so check each vendor's current page. Most price by how many questions and assistants you track: Otterly's plans, for instance, differ mainly by the number of prompts included. Compare the cost per tracked question, and check a sample of the tool's answers by hand. Our guide to [building a simple AI visibility dashboard](/insights/build-ai-visibility-dashboard) shows how to report what the tool collects. ## What do GA4 and Search Console show now? Part of the picture, for free. Since 3 June 2026 Search Console has [generative AI performance reports](https://developers.google.com/search/blog/2026/06/gen-ai-performance-reports) showing impressions for AI Overviews and AI Mode, rolled out to all sites by 31 August 2026. Since 13 May 2026 GA4 has had a default [AI Assistant channel](https://www.nicelookingdata.com/blog/ga4-ai-traffic-chatgpt-referrals) for visits from ChatGPT, Gemini, DeepSeek, Copilot and Grok; Claude and Perplexity are not on its list, and clicks from AI Overviews and AI Mode still count as Organic Search. Neither report tells you how often you were named in an answer that did not link to you, or who was named instead. That is why the question list still matters. ## How do we read the results? Read for patterns, not single answers. Each pattern points to a different fix. **You are known by name but not recommended.** Assistants describe you correctly when asked directly, but leave you out of shortlists. ID-Pal was in exactly this position: "AI could describe ID-Pal. It never recommended it." Its case study adds: "Across 36 recorded AI answers to the questions ID-Pal's buyers actually ask, ID-Pal was named three times, and only for its brand or its Salesforce integration." After pages were rebuilt around buyer questions: "UK buyer questions where AI names ID-Pal: none in May, two in August, four in September 2026." The fix is usually pages that describe you in the buyer's terms, plus independent mentions that confirm them. **You are named only in a narrow niche.** You appear for "IT support for dental practices in Coventry" but not "IT support in Coventry". That is fine if the niche is your market; if not, you need content and coverage in the wider category. **You are named but described wrongly.** Out-of-date services, an old office, the wrong headcount. Our guide to [fixing wrong AI descriptions of your firm](/insights/ai-conflicting-information-online) explains how. **Competitors are cited from sources you are missing from.** If the same trade articles, directories or comparison pages keep appearing, those are the places you need to be. **Answers hedge about you.** "According to the company" suggests the assistant found your claims only on your own website, which points to a need for independent coverage. The same sheet gives you a competitor view, because you have already recorded every firm named. Our guide to [measuring AI share of voice against competitors](/insights/benchmarking-ai-visibility-competitors) turns it into a benchmark. ## How do we connect tracking to enquiries? Tracking tells you whether you are named, not whether it brings in work. For that, ask the people who get in touch. - Add "an AI assistant such as ChatGPT" to the "how did you hear about us?" field on your enquiry form, with a free-text box. - Ask the same question on every first call and record the answer in your CRM. - Watch the GA4 AI Assistant channel, and add Claude and Perplexity referrals yourself. Expect this to understate the effect: many buyers read an AI answer, then search your name or type your address, so the visit looks like ordinary search or direct traffic. When the channel works, it shows in the numbers: at [Helm Club](/work/helm-club), "1 in 5 new members this year came through an AI recommendation." Our guide to [tracking leads in the AI era](/insights/attribution-ai-era-track-leads) covers this in more depth. ## What to do this month 1. Run the 15-minute check above in ChatGPT, Google, Copilot and Gemini. 2. Ask your sales team for the questions prospects most often ask before they buy, and turn them into a fixed list of 20 to 50. 3. Set up a dedicated UK tracking account with memory off, and copy the sheet headers above. 4. Run your baseline, three times per question per assistant, recording the model each time. 5. Add an AI assistant option to your enquiry form and first-call notes, and check the GA4 AI Assistant channel. 6. Book the weekly runs and a monthly review, and share the baseline with your fellow directors. ## Frequently asked questions ### Is there a free tool to check if ChatGPT mentions my brand? ChatGPT itself, used in a clean session with a spreadsheet, is the free tool, and it is how we suggest every firm starts. Google Search Console and GA4 are free and show part of the picture for your own site. Some paid trackers offer trials: [Otterly](https://otterly.ai/) advertises a seven-day free trial with no card needed. ### Does ChatGPT give different recommendations in the UK and the US? It can. OpenAI says ChatGPT search may use an approximate location based on your IP address to give relevant local results, so the same question asked from London and New York can produce different firms. Track from a UK connection, and include the country or region in questions where your buyers would. ### How often do ChatGPT's recommendations change? Every answer can vary, which is why you ask several times. Assistants that search the web live can reflect a new page or article within weeks; answers drawn from a model's memory change only when the model is retrained, which takes months. ID-Pal's record shows the pace: "Claude names it for UK fund administrators, on a page rebuilt in August, and links NorthRow to ID-Pal in the same answer." ### What if we are never named at all? That is common, and it is a starting point rather than a verdict. Look at who is named and which sources are cited for them, check that your website says plainly what you do, for whom and where, and that AI crawlers can read it. Then earn independent mentions in the places the assistants already cite. ### Do we need to track Microsoft Copilot as well? Yes, if your buyers work in organisations that use Microsoft 365, because Copilot sits inside the tools they use every day. Start with ChatGPT and Google, add Copilot and Gemini, then Perplexity and Claude. If you want a first look before you build the sheet, [book a free Growth Review](/growth-review): 30 minutes with David Adams, with your site, two competitors and what Google and ChatGPT say about all three checked beforehand, and three fixes in writing afterwards. ## Sources - [G2, The Answer Economy: half of B2B software buyers now start research with AI chatbots (April 2026)](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html) - [OpenAI Help Center, ChatGPT search (checked September 2026)](https://help.openai.com/en/articles/9237897-chatgpt-search) - [OpenAI Help Center, Temporary Chat FAQ (September 2026)](https://help.openai.com/en/articles/8914046-temporary-chat-faq) - [Google Search Central Blog, Generative AI performance reports in Search Console (June 2026)](https://developers.google.com/search/blog/2026/06/gen-ai-performance-reports) - [Nice Looking Data, GA4's AI Assistant channel for ChatGPT referrals (checked September 2026)](https://www.nicelookingdata.com/blog/ga4-ai-traffic-chatgpt-referrals) - [Forrester, B2B buyers make zero-click buying number one (January 2026)](https://www.forrester.com/blogs/b2b_buyers_make_zero_click_buying_number_one/) - [Semrush, How to track your ChatGPT visibility (November 2025)](https://www.semrush.com/blog/how-to-track-your-chatgpt-visibility/) - [Ahrefs, Brand Radar (checked September 2026)](https://ahrefs.com/brand-radar) --- Title: How do we get trade journalists to call us for comment? URL: https://fortitudemedia.ai/insights/building-relationships-journalists-editors Author: Ross Williams Published: 2026-03-31 Last reviewed: 2026-09-24 Topic: Press, reviews and being trusted Short answer: Journalists quote sources who answer fast, speak clearly and know the subject. Start by answering their requests for comment, through ResponseSource, HARO, Qwoted or the #journorequest hashtag, before you pitch anything. Keep a list of fifteen to thirty journalists who cover your buyers' sector, prepare clear views on the topics that keep coming back, and stay useful. Each quote is also an independent mention AI assistants read: journalism made up 27% of links AI tools cited in Muck Rack's May 2026 study. Every month the same competitor seems to be quoted in the trade magazine your buyers read. When a new regulation lands, their founder explains what it means. When the sector's figures come out, their founder comments. Your team knows the subject at least as well. But nobody rings you. The difference is rarely expertise. A journalist on a deadline knows that founder's number, knows they will pick up and knows they will say something clear and usable. That is a relationship, and it can be built deliberately, without a large agency or a famous name. It matters more now, because trade coverage is part of what AI assistants read when buyers ask who to hire. In [Muck Rack's May 2026 analysis](https://www.globenewswire.com/news-release/2026/05/07/3290268/0/en/generative-pulse-earned-media-consistently-drives-ai-citations-holding-at-84.html) of more than 25 million links cited by ChatGPT, Claude and Gemini, earned media made up 84% of cited links and journalism alone 27%, while paid content was 0.3%. Each quote in a trusted trade title is an independent mention of your firm of exactly the kind those assistants cite. ## Key takeaways - Journalists quote sources they trust to answer quickly, speak clearly and know the subject, so reliability matters more than fame. - A list of fifteen to thirty journalists who cover your buyers' sector is enough for most founder-led B2B firms. - The best first contact is a fast, useful reply to a live request for comment, found through ResponseSource, HARO, Qwoted, Featured or #journorequest. - Prepared views on the five or six topics your sector keeps returning to let you reply within the hour. - A media page with each spokesperson's biography, subjects and direct number means journalists and search engines read the same facts about you. - Every quote in a trusted trade title is an independent mention of your firm that AI assistants can read and repeat. ## Why do some firms get quoted and others never do? Because journalists work under time pressure and go back to sources who have made their job easier before. Being quoted is less a reward for expertise than a reward for reliability. A regulator publishes something at ten in the morning. The story needs to be online by early afternoon, with a comment from someone who can explain what it means. The journalist has perhaps an hour. They go first to the people they know will answer the phone and say something clear. The ones who respond get quoted, and next time they are called first. That is why the same names recur, and the pattern is open to anyone willing to be useful, quickly and consistently. ## Which journalists matter to a firm like ours? The ones your buyers read, not the most famous ones. For most founder-led B2B firms, fifteen to thirty people is plenty: - reporters and editors on the two or three trade titles your buyers read for work - the regional business press, if your buyers are regional - writers of specialist newsletters and hosts of podcasts in your buyers' sector - one or two national journalists who cover the sector, as a longer-term aim Read the last few months of each title, note who writes about your topics and who quoted your competitors. Keep a simple spreadsheet: name, title, what they cover, recent pieces, how they like to be contacted and every interaction. It is what stops a relationship going cold because nobody remembered the last conversation. ## Where do UK journalists post requests for comment, and what are the HARO alternatives? In more places than they used to. Answering a request quickly and well is the best first contact there is, because you solve a problem the journalist has right now. - **#journorequest on X, Bluesky and LinkedIn.** Save a search for the hashtag on all three, not X alone. - **ResponseSource.** A UK service that sends journalist enquiries to your inbox. It is part of Pulsar Group, which also owns Vuelio, and is a paid subscription with a free trial ([ResponseSource](https://www.responsesource.com/)). - **The relaunched HARO.** Help A Reporter Out is now run by Featured, and sources can subscribe free to daily media queries ([HARO](https://www.helpareporter.com/)). - **Qwoted and Featured.** Platforms where journalists post requests and experts reply. [Qwoted](https://www.qwoted.com/) is free to join. Check one or two at set times each day and answer only the requests you can answer well. A weak reply outside your subject costs more trust than silence. ## How do we make first contact without pitching? Start by being useful and ask for nothing. A first email that pitches your firm reads like every other email in their inbox. ### Step 1: Read their recent work Read their last ten pieces: what they cover, the angles they take, the sources they quote. ### Step 2: Answer a live request A good reply to a request posted this morning does more than any introduction, because it arrives when they need it. ### Step 3: Send a short note offering help If there is no request to answer, send the no-pitch email below. ### Step 4: Follow their work and meet in person Share their good pieces with a comment that adds something, and say hello at trade events. A short conversation with no pitch beats ten emails. ## What do journalists want from a source? A clear, informed view, delivered quickly, from someone who understands how journalism works: - **Speed.** Reply within the hour when they are on deadline, even if only to say you cannot help. - **A clear opinion.** "It depends" is not a quote. Say what you think and why, in two or three sentences. - **Specifics.** An example or a practical consequence for their readers. - **Candour.** If you do not know, say so and suggest someone who does. - **No selling.** A comment that plugs your services will be cut, and remembered. - **On the record.** Give your name, role and firm, and assume anything you say may be quoted. Say a change is announced to how employers must calculate holiday pay. A journalist needs someone to explain what it means for a mid-sized employer by two o'clock. The source who replies at quarter past eleven with three clear sentences and a phone number gets quoted. The source who replies the next morning with a link to their services page does not. ## How do I write a good reply to a journalist's request for comment? Answer the question in the first line, in words their readers use, and keep it short enough to paste. Add your name, role and firm as you want them printed and a mobile you will answer; leave out attachments and links to your services. ### Template: a 60-word reply to a live request > Subject: Re: comment on holiday pay changes for mid-sized employers > > [Name], the biggest practical change for a mid-sized employer is [one consequence]. The businesses most exposed are [who], because [reason]. The first thing to do this week is [one action]. Happy to talk it through by phone before two. > > [Your name], [role], [firm], [mobile] ### Template: the no-pitch first email > Subject: Your piece on holiday pay > > Dear [Name], I read your piece on the holiday pay changes. Your point about [specific point] matches what we see with [type of client] every week. I run [firm], which advises [who] on [subject]. If you ever need someone to explain a point or check a detail against a deadline, I am happy to help, on the record and quickly. My mobile is [number]. > > [Your name] No attachment, no request for a call. The only aim is that your name is in their inbox when they next need someone. ## How do I become the expert journalists quote in my industry? By being prepared, reachable and consistent for months. Journalists come back to the source who answered well last time, so make every answer easy to give. - **Know the recurring topics.** Every sector has five or six subjects that keep coming back: a regulation, costs, skills, technology, a regular set of official figures. Write your view on each in plain sentences. - **Keep a comment bank.** Draft two or three quotable lines per topic that your founder will stand behind, so a request means editing, not starting from a blank page. - **Name one spokesperson and a deputy.** Journalists want a person, reachable by phone in the working day. ### Build a media page a journalist can use in two minutes A journalist who finds you through a search or an AI assistant checks your site before calling. Put these on one page: | Field | What to put | |---|---| | Name and role | Exactly as you want them printed | | One-line description of the firm | The same line every time, for example "[Firm] advises UK mid-sized employers on pay and HR compliance" | | Subjects | The five or six topics each spokesperson covers | | Short biography | 60 to 80 words of facts: years in the sector, qualifications, what the firm does | | Recent coverage | Links to the last five to ten pieces that quoted you | | Direct contact | A mobile and email that reach a person, plus the deputy's details | | Photo | A current headshot with its credit line | Then add [schema.org Person](https://schema.org/Person) markup for each spokesperson, using the same facts: name, jobTitle, worksFor, description, knowsAbout (their topics) and sameAs (their LinkedIn profile). Markup will not win coverage, and [Google says](https://developers.google.com/search/docs/appearance/ai-features) its AI features need no special structured data. It helps search engines read the same biography a journalist reads. Keep your founder's LinkedIn profile saying the same thing. ## Is it worth paying for Cision, Muck Rack or Vuelio as a small firm? Usually not at the start. For thirty journalists, reading the titles teaches you more than a database, and the request services cover most live opportunities. A database earns its place when you need hundreds of contacts across sectors or countries and someone will use it daily. None of the three publishes a price list. [Muck Rack says](https://muckrack.com/pricing) pricing depends on factors "including the number of users"; [Cision](https://www.cision.co.uk/pricing/) and [Vuelio](https://www.vuelio.com/uk/pricing/) ask you to request a demo or quote. | Option | What you get | Cost | Best for | |---|---|---|---| | Reading titles and LinkedIn | Names, subjects, how each journalist writes | Your time | Every firm | | HARO (run by Featured) | Daily email of journalist queries | Free to subscribe | Spotting live requests | | Qwoted, Featured | Platforms for requests and replies | Qwoted free to join | Spotting live requests | | ResponseSource | UK journalist enquiries by email | Paid, free trial | UK trade and national requests | | Cision, Muck Rack, Vuelio | Media database, pitching, monitoring | Quote on request | PR at scale or across markets | ## How much founder time does this take? Less than most founders fear, if it is a routine. In our experience a short daily check and a small weekly slot beat a burst of effort every few months. For example, a routine might look like this (illustrative times): | Task | How often | Example time | |---|---|---| | Check request services and saved searches | Daily | 10 minutes | | Reply to a relevant request | When one appears | 20 minutes | | Update the comment bank | Weekly | 30 minutes | | Send one no-pitch note or follow-up | Weekly | 15 minutes | | Review the list and log contacts | Monthly | 1 hour | ## How do I prepare my founder for a press interview? Agree three points, prepare for the awkward question and settle the terms before anyone speaks. - Ask the angle, the deadline and whether it is a quick quote, a longer interview or recorded. - Write three points, each with one example from your own work, and practise saying each in two or three sentences. - Prepare answers to the two questions you least want asked. - Agree terms before anything sensitive is said (see "off the record" below). - Afterwards, send any figure you promised in writing. Do not ask to approve the copy. ## How do we pitch a journalist we already know? Briefly, with a story for their readers, and with the grace to accept a no. Three short paragraphs: the story in one sentence, why it matters to their readers now, and what you can offer, such as data, a client willing to speak or an exclusive. If they pass, do not re-pitch the same idea in new words. The strongest pitches rest on something only you have, such as your own data, covered in [how to turn our own data into coverage in the trade press](/insights/leverage-data-research-press-coverage). Longer bylined pieces are covered in [whether guest articles in trade magazines are still worth your time](/insights/guest-articles-authority-building-ai). ## How do we keep relationships going without being a nuisance? Stay in touch every few months with something useful, and never make them feel chased. Send a report they might have missed or a contact for a story. Thank them when a piece is published, without asking for changes. Never complain that a quote was cut. Point out factual errors politely and privately. Respect their rules on gifts and hospitality. When they move title, stay in touch: the relationship moves with them. ## Do journalists use ChatGPT to find sources? Some do, and it is sensible to assume more will. Muck Rack's [State of Journalism 2026](https://muckrack.com/research/state-of-journalism), a survey of nearly 1,100 journalists, looks at how journalists now use AI in their work. A journalist who asks an assistant "who are the UK experts on holiday pay?" gets much the same names a buyer would. So the two routes feed each other: quotes give assistants evidence to name you, and being named puts you in front of journalists looking for a source. ## Why does this matter for AI answers? Each quote is an independent mention of your firm on a trusted site, next to the subject you want to be known for. A founder quoted many times in the same terms becomes a name ChatGPT, Gemini, Copilot or Claude links with that subject. Timing works two ways. Assistants that search the web live, such as ChatGPT search, Perplexity, Copilot and Google's AI features, can reflect a new article within weeks; answers from a model's own memory change only when it is retrained, which takes months. Recency counts too: in the [Muck Rack study](https://www.globenewswire.com/news-release/2026/05/07/3290268/0/en/generative-pulse-earned-media-consistently-drives-ai-citations-holding-at-84.html), more than half of the journalism AI tools cited was published in the previous 12 months, with citations dropping sharply after six months. A steady stream of quotes beats one big story. Use the same one-line description with every journalist, and list your coverage on your own site. For the wider picture, see [how to get ChatGPT to recommend your firm](/insights/what-is-llm-optimisation) and [whether ChatGPT trusts press coverage more than your website](/insights/earned-owned-paid-authority-signals). ## What to do this month 1. List the fifteen to thirty journalists who cover your buyers' sector, with what they write about and how to reach them. 2. Write your view on the five or six recurring topics in your sector, with two quotable lines for each. 3. Save #journorequest searches on X, Bluesky and LinkedIn, and subscribe to one request service. 4. Build a media page for each spokesperson and add Person markup that matches it. 5. Send three short notes, with no pitch, to journalists whose recent work you have read. ## Frequently asked questions ### How long before a journalist starts calling us? Expect months rather than weeks. Answering a live request well can lead to a quote the same day, but becoming someone they ring without prompting usually takes several useful exchanges. The firms that get there keep going through the quiet periods. ### Can we ask to see the article before it is published? Generally no. Most journalists will not give sources approval over their copy, and asking can damage the relationship. You can offer to check technical facts or figures, which many welcome. ### What does off the record mean, and should we use it? The terms are used inconsistently, so never assume. Many journalists treat "off the record" as information they cannot use at all, while "on background" usually means they can use it without naming you. Agree terms before you say anything sensitive, never afterwards. The simplest rule is to say only what you would be happy to see in print with your name beside it. ### Do we need a PR agency to do this? Not necessarily. Journalists want the expert, so the relationships sit with your founder. An agency or adviser can build the list, prepare positions, spot requests and draft replies, but it cannot be the expert on the phone. ### What if a journalist writes something negative about us? Respond calmly and privately. If it is wrong, send the correct information and ask for a correction. If it is fair, deal with the issue rather than argue in public. If you want help building the list, preparing your founder's positions and spotting the requests worth answering, that is part of [Reputation That Wins](/reputation-that-wins). The relationships stay yours. If you are not sure where to start, the free [Growth Review](/growth-review) gives you three fixes in writing. ## Sources - [Muck Rack, What is AI reading? Earned media drives 84% of AI citations (May 2026)](https://www.globenewswire.com/news-release/2026/05/07/3290268/0/en/generative-pulse-earned-media-consistently-drives-ai-citations-holding-at-84.html) - [Muck Rack, The State of Journalism 2026 (2026)](https://muckrack.com/research/state-of-journalism) - [Muck Rack, Pricing (September 2026)](https://muckrack.com/pricing) - [HARO, Help A Reporter Out, run by Featured (September 2026)](https://www.helpareporter.com/) - [ResponseSource, journalist enquiry service (September 2026)](https://www.responsesource.com/) - [Schema.org, Person (September 2026)](https://schema.org/Person) - [Google Search Central, AI features and your website (December 2025)](https://developers.google.com/search/docs/appearance/ai-features) --- Title: How do we measure the return we get from our website? URL: https://fortitudemedia.ai/insights/measure-website-roi-b2b Author: Ashley Kenerson Published: 2026-03-31 Last reviewed: 2026-09-24 Topic: Websites that bring work in Short answer: Measure two figures. Direct return is the fees from clients who first enquired through the website, divided by the site's full yearly cost, including senior time. Influenced value is the fees from clients who came in another way but used the site before buying, reported alongside and never added. Ask every new client how they found you, value each client the same way every year, and track qualified enquiries and AI visits monthly. Sooner or later your finance director asks the question outright: what do we actually get back from the website? When you try to put a number on it, you find yourself holding a traffic report that says nothing about money, or a supplier's slide that promises returns nobody in the room believes. A B2B website rarely closes a deal on its own. It is where a buyer checks you out after a referral, where a finance director confirms you have worked with firms like theirs, and increasingly the source AI assistants read before they recommend anyone. [Forrester reported in January 2026](https://www.forrester.com/blogs/b2b_buyers_make_zero_click_buying_number_one/) that 94% of business buyers now use AI in their buying process, and that buyers rated generative AI as a more meaningful source than vendor websites. Much of your website's value now happens without a form being filled in, and some of it without a visit at all. This guide sets out a method you can run with your own figures, that a sceptical finance director will accept, and that separates what you can prove from what you believe. ## Key takeaways - Measure website return in two layers: enquiries the site brings in directly, and deals it helps win that start elsewhere. - Ask every new client how they found you and what they looked at, because analytics alone misses much of the website's role. - Put the full cost of the website into the sum, including senior time and running costs, not just the build invoice. - Visits from AI assistants are undercounted, because many arrive with no referrer and are recorded as Direct. - Compare the website's cost per client with events, paid search and business development hires using the same formula for each. - Report a conservative direct figure and a separate influenced figure, and never add speculative "brand value" to either. ## Why is website return so hard to measure? Because the website does several jobs at once, over a long buying process, for several people at the buyer's firm, and most of those jobs leave no trace in your CRM (the system where you track prospects and clients). Consider a typical deal. A managing director hears your name from a peer, reads two case studies and the team page that evening, and forwards the link to her operations director. Three weeks later he calls your office. Your CRM records the enquiry as "referral" or "phone". The website did real work, but it appears nowhere. Two more things make it harder now. Buyers do most of their research before they speak to anyone: [6sense's 2025 Buyer Experience Report](https://6sense.com/newsroom/the-timeline-for-influencing-b2b-buyers-is-shrinking-insights-from-6senses-2025-buyer-experience-report/) found buying groups had ranked their preferred vendors before first contact in 94% of cases. And a buyer can ask an AI assistant who to hire and get a shortlist without visiting anyone's site. If the assistant named you because your website stated clearly what you do, the site earned that recommendation, but no visit was recorded. None of this means the return cannot be measured. It means you need to ask people as well as read reports. ## What should you count as the website's return? Count two things, keep them separate, and leave out a third. ### Direct return Fees from clients whose first contact with you came through the website: a contact form, a meeting booking, a call to the number on the site, or an email to an address listed there. This is the figure you can defend line by line. ### Influenced return Fees from clients who came in another way, such as a referral, an event or LinkedIn, but who told you they used the website before or during the buying decision. This is real value, but it is shared with whatever brought them in, so report it separately and do not claim all of it. ### What to leave out Leave out "brand value", awareness multiplied by market size, assumed pricing premiums and industry averages you cannot check. They inflate quickly and make the rest of your case easier to dismiss. If the website helps in ways you cannot measure, say so in words. ## How to measure the return, step by step ### Step 1: Add up what the website really costs Spread the build cost over the years you expect the site to last. Add hosting, maintenance, licences, content, photography, any agency retainer, and a fair share of your own people's time. Senior time is the cost most firms forget, and in a founder-led firm it is often the largest. Our guide to [what a B2B website costs in the UK](/insights/real-cost-of-a-website) lists the costs that tend to be missed. ### Step 2: Record the source of every enquiry From today, ask every new enquiry how they found you, and record the answer in your CRM. Give clear options: Google search, ChatGPT or another AI assistant, referral, LinkedIn, event, website (already knew you), other. Then ask whether they looked at your website first, and what they read. This takes thirty seconds on a first call and catches what tracking code cannot. Visit data alone would struggle to show a result like [Helm Club](/work/helm-club)'s: "1 in 5 new members this year came through an AI recommendation." A buyer told about a firm by an assistant often searches for its name or types the address, so the visit looks like brand search or Direct. We cover combining the answers with analytics in [how to tell which of your new leads came from ChatGPT](/insights/attribution-ai-era-track-leads). ### Step 3: Trace last year's clients back through the website You will not have perfect records for the last twelve months, so reconstruct them. List every new client, note the first point of contact from your CRM or inbox, and ask whoever managed the relationship whether the client mentioned the website. Where the relationship allows, ask the client directly. In our experience most firms find a handful of direct website clients and a larger group of influenced ones. ### Step 4: Put a consistent value on each client Decide how you will value a client and use the same rule every time. First-year fees are the simplest and most conservative. Lifetime value is fairer for firms with long retainers but needs a clear basis. Write the rule down, so the calculation can be repeated next year. ### Step 5: Calculate direct and influenced return separately Direct return is the value of direct website clients divided by the full yearly cost of the website. Influenced value is reported alongside it, not added to it. Example: a specialist engineering consultancy spends £30,000 a year on its website, counting the build spread over four years, hosting, updates, content and a share of senior time. In the last twelve months, three clients came in directly through the website, each worth £25,000 in first-year fees. That is £75,000 of direct return, or two and a half times the cost. Another five clients came through referrals or events but said they read the case studies and team pages before agreeing to a meeting, worth £120,000 between them. The firm reports that as influenced value and does not claim it in full. A finance director can check every number in that paragraph. That is the point. ### Step 6: Track the early signals monthly Clients take months to arrive, so watch the measures that move first: - **Qualified enquiries** through the website each month, not total form fills. - **Visits from AI assistants,** treated as a minimum, for the reasons below. - **How often you are named** when you ask ChatGPT, Google's AI Mode, Copilot and Gemini your buyers' questions. - **Visits to service pages and case studies,** which show buyers doing real research. If these rise, the return usually follows. For a wider health check, see [how to tell whether your website is working](/insights/evaluate-whether-website-working). ## Why does analytics undercount visits from AI assistants? Because many arrive without a label. Links opened from AI apps often lose their referrer on the way to the phone's browser and are recorded as Direct ([Nice Looking Data, July 2026](https://www.nicelookingdata.com/blog/ga4-ai-traffic-chatgpt-referrals)). GA4's default "AI Assistant" channel, added in May 2026, covers ChatGPT, Gemini, DeepSeek, Copilot and Grok, but [Google's definitions](https://support.google.com/analytics/answer/9756891?hl=en) leave out Claude and Perplexity, and count AI Overviews and AI Mode as Organic Search. To segment what you can, add a custom channel in GA4 under Admin, Data display, Channel groups, where session source matches `chatgpt\.com|claude\.ai|perplexity\.ai|gemini\.google\.com|copilot\.microsoft\.com`, placed above Referral. [OpenAI says](https://help.openai.com/en/articles/12627856-publishers-and-developers-faq) ChatGPT adds utm_source=chatgpt.com to its referral links, which helps. Then watch Direct sessions that land on service pages and case studies, which few people type from memory, and report that trend beside the channel. ## How do we compare the website's cost per client with other channels? Use the same formula for every channel: its full yearly cost divided by the new clients whose first contact came through it, over the same twelve months. This is customer acquisition cost by channel, and for a services firm it is only fair if every line includes people's time. | Channel | Full yearly cost to include | Clients to count | Cost per client | |---|---|---|---| | Website | Build spread over its life, hosting, maintenance, content, agency fees, senior time | First contact through the site | Full yearly cost ÷ direct website clients | | Events | Tickets or stands, travel, materials, days out of the office | First met at an event | Full yearly event cost ÷ event clients | | Paid search | Ad spend, management fees, landing pages, review time | First contact from an ad click | Spend plus fees ÷ paid search clients | | Business development hire | Salary, employer's National Insurance, pension, bonus, tools, travel, management time | Clients the hire sourced | Full yearly cost of the role ÷ clients sourced | In the example above, the website line is £30,000 ÷ 3, or £10,000 per direct client. Keep influenced clients out of every line, or each channel will claim the same client, and judge a new hire over a full year after they settle in. ## What about buyers who never visit the website? They still count, but you measure them differently. [Pew Research Center found in July 2025](https://www.pewresearch.org/short-reads/2025/07/22/google-users-are-less-likely-to-click-on-links-when-an-ai-summary-appears-in-the-results/) that people clicked a traditional search result on 8% of visits when Google showed an AI summary, against 15% when it did not. Your website may supply the answer without receiving the visit, as we explain in [why website traffic falls when rankings have not](/insights/zero-click-search-your-business). For these buyers, the site's job is to be a clear, accurate source that AI assistants trust. Measure that by asking the assistants your buyers' questions, and by watching for enquiries that mention AI assistants in your Step 2 records. If enquiries and wins hold while visits fall, the website may simply be doing more of its work before the click. ## How do you present the figure to a finance director? Keep it conservative, show the method, and give two numbers rather than one. A single table works well. Here it is filled in with the illustrative consultancy figures from Step 5, ready to copy into a board pack. | Line | Figure | How it was worked out | |---|---|---| | Full yearly cost | £30,000 | Build spread over four years, hosting, updates, content, senior time | | Direct clients | 3 | First contact through a form, booking, call or email from the site | | Direct value | £75,000 | 3 clients × £25,000 first-year fees | | Direct return | 2.5 times cost | £75,000 ÷ £30,000 | | Cost per direct client | £10,000 | £30,000 ÷ 3, on the same basis as other channels | | Influenced clients and value | 5 clients, £120,000 | Came via referral or events, read case studies first; shared, not added | | Early signals | Six-month trend | Qualified enquiries, AI visits, times named, service page visits | | Next year | Planned changes | Each change listed with the signal it should move | What finance directors distrust is a single large multiple with no working shown. What they accept is a modest, checkable number and a plain description of what it leaves out. For the wider budget case, see [getting your board to approve budget for AI search](/insights/business-case-ai-optimisation-board). ## What is a good ROI for a B2B website? One that clears its own break-even. There is no reliable benchmark for B2B firms, because client values, sales cycles and markets differ too much, so ask what the website needs to achieve to pay for itself. For example, if your typical client is worth £40,000 in the first year and your website costs £20,000 a year in full, one direct client every two years covers the cost. Anything beyond that is return. For founder-led firms selling high-value services the bar is usually lower than people expect. If your site is not clearing it, the answer is rarely to spend less. It is usually to fix what the site says, how easy it is to act on, and whether AI assistants can read it. We explain what standing still costs in [what it costs to leave your website unchanged for years](/insights/hidden-cost-static-website). ## What to do this month 1. **Work out the full yearly cost.** Include the spread build cost, running costs and senior time. 2. **Add two questions to every first call.** How did you find us, and did you look at our website first? 3. **List last year's new clients.** Mark each as direct, influenced or neither, and value them with one rule. 4. **Calculate the two figures and the cost per client.** Direct return against cost, influenced value alongside. 5. **Set up the monthly signals.** Add the GA4 AI channel, and track qualified enquiries, how often you are named and service page visits. ## Frequently asked questions ### Should we use first-year fees or lifetime value? Use first-year fees for the most conservative figure and the least argument. Use lifetime value if most clients stay for several years and you have records to support it. Either works, as long as you state which you used and apply it the same way every year. ### Our sales team says all our work comes from referrals. Is the website irrelevant? Probably not. Referrals often lead straight to your website, where the buyer checks you are credible before calling. Ask your last ten referred clients whether they looked at the site first. If most did, the website supports your referral engine, and that value belongs in the influenced figure. ### How do we set up a "how did you hear about us" field that sales will fill in? Make it a required dropdown in the CRM when a deal or contact is created, with no more than eight fixed options, including "ChatGPT or another AI assistant", plus a short free-text box. Ask the question on the first call rather than relying on the web form, because people skip optional fields. Review the answers monthly with the team, so they see the data being used. ### How do we track phone calls that come from our website? Show a phone number on the website that appears nowhere else, so every call to it started on the site. Call-tracking services go further, showing a different number to each visitor and matching calls to the pages they read. On mobile, record taps on the number as an event in GA4, and ask every caller how they found you. ### How do we attribute a deal to the website in HubSpot? HubSpot sets an Original Traffic Source on each contact automatically, with values including Organic search, Referrals, Direct traffic and AI Referrals, and [its documentation](https://knowledge.hubspot.com/properties/understand-source-properties) says associated companies and deals mirror those properties. Report deals by that property for the direct figure. Add your "how did you hear about us" field to the deal too, because the original source records only the first visit HubSpot saw, not the referral or call that followed. ### How long before website improvements show up in the return? Early signals such as better enquiries and more accurate AI descriptions can move within a few months: assistants that search the web live can reflect a changed page within weeks, while answers from a model's training change only when it is retrained. New clients take longer, because B2B decisions do, so judge the return over a full year. ### Should we count recruitment and other benefits? Mention them, but keep them out of the figure. A good website helps you hire and reassures existing clients, but those benefits are hard to value. List them in words below the numbers, so your finance director sees them without having to trust a guess. The plain answer to "what does the website give us back" is usually more than a traffic report suggests and less than a supplier's slide claims. If you want a site built to bring in work you can trace, see how we approach it in [Website That Sells](/website-that-sells), or start with a free [Growth Review](/growth-review). ## Sources - [Forrester, B2B buyers make zero-click buying number one (January 2026)](https://www.forrester.com/blogs/b2b_buyers_make_zero_click_buying_number_one/) - [6sense, The timeline for influencing B2B buyers is shrinking: 2025 Buyer Experience Report (November 2025)](https://6sense.com/newsroom/the-timeline-for-influencing-b2b-buyers-is-shrinking-insights-from-6senses-2025-buyer-experience-report/) - [Pew Research Center, Google users are less likely to click on links when an AI summary appears (July 2025)](https://www.pewresearch.org/short-reads/2025/07/22/google-users-are-less-likely-to-click-on-links-when-an-ai-summary-appears-in-the-results/) - [Nice Looking Data, AI traffic in GA4: how ChatGPT, Perplexity, Gemini and Google AI Mode referrals show up (July 2026)](https://www.nicelookingdata.com/blog/ga4-ai-traffic-chatgpt-referrals) - [Google Analytics Help, Default channel group (September 2026)](https://support.google.com/analytics/answer/9756891?hl=en) - [OpenAI Help Center, Publishers and developers FAQ (August 2026)](https://help.openai.com/en/articles/12627856-publishers-and-developers-faq) - [HubSpot Knowledge Base, Understand source properties (August 2026)](https://knowledge.hubspot.com/properties/understand-source-properties) --- Title: How do we turn our own data into coverage in the trade press? URL: https://fortitudemedia.ai/insights/leverage-data-research-press-coverage Author: Ashley Kenerson Published: 2026-03-31 Last reviewed: 2026-09-24 Topic: Press, reviews and being trusted Short answer: Journalists cover findings about a market, not news about your firm. Answer one question your buyers argue about with your own records or a short survey of the right people, and publish the method, sample size and dates with the findings on a web page. Offer it first to the trade title your buyers read under a clear embargo, pitch others with their own angle, and repeat the study every year so it becomes the reference point. You have sent out press releases about a new hire, a new office and a tenth anniversary. None was picked up. Meanwhile the trade magazine your buyers read quotes your competitors, usually citing a number from something they published. The gap is usually not expertise. Journalists write about what is new, and an announcement about your firm is only new to you. A finding about the market, with numbers behind it, is new to everyone. A small, well-run piece of research gives a journalist a story they can stand behind, and puts your firm's name in the article as the source. That coverage now works twice. It reaches buyers directly, and it is the kind of source AI assistants cite: in [Muck Rack's May 2026 analysis](https://www.globenewswire.com/news-release/2026/05/07/3290268/0/en/generative-pulse-earned-media-consistently-drives-ai-citations-holding-at-84.html) of more than 25 million links cited by ChatGPT, Claude and Gemini, earned media made up 84% and journalism 27%. Why your own research builds that trust is covered in [whether publishing our own research helps ChatGPT recommend us](/insights/original-research-data-ai-trust). This guide is about the press side: choosing the question, running the study, packaging it and pitching it. ## Key takeaways - Journalists write about findings that tell them something new about a market, and ignore announcements that only tell them something about your firm. - Most founder-led firms already hold useful data in their own records, which can be anonymised and summarised into a story. - A finding is newsworthy when it is specific, surprising and useful to decision-makers, and does not simply argue for what you sell. - Publish the method with the findings, including sample size, dates and who was asked, because journalists will not cite numbers they cannot explain to their editor. - Put the findings on a web page, one citable sentence per finding with its base, rather than only in a PDF. - Offer the research to one title first under a clear embargo, then pitch others with an angle for their readers, and repeat the study every year. ## Why do journalists write about research and ignore announcements? Because they need stories their readers do not already know, and evidence to stand them up. Research does both jobs. An announcement does neither. A trade editor's inbox is full of promotions, contract wins and launches that interest nobody outside the firms involved. What readers want is something about the sector: costs rising faster than expected, a regulation nobody is ready for, a skills gap in a region. Bring a finding like that and the piece becomes about the market, with your firm as the source of the evidence and your founder explaining what it means. Research also lasts. A story about your office move is forgotten within a week. A figure about the sector gets quoted again by other journalists, trade bodies and people's presentations, each time with your name attached. ## What data does a firm like ours already have? More than most founders think. Before commissioning a survey, look at your own records. - **Your operational data.** A firm that has done a few hundred pieces of work holds patterns: how long projects take, how costs have moved, which problems clients bring most. Summarised and anonymised, that is a story nobody else can tell. Check client contracts and data protection obligations first. - **Public data nobody has joined up.** Companies House filings, Office for National Statistics releases, job adverts, planning applications and tender notices are free to analyse. Two public sources side by side, explained for your sector, is often a story. - **A survey of your market.** A short online survey through a panel provider or your own contacts. The most flexible route, and the one that needs most care. - **Interviews.** Twenty or thirty structured conversations can produce findings and quotes a survey cannot, where the group is small, such as heads of legal at a particular kind of company. ## What makes a finding newsworthy? A finding is newsworthy when it tells decision-makers in your sector something specific they did not know and would act on. It helps if it surprises them, and it must not read as an advertisement for what you sell. - **It is about the market.** "How firms like ours get chosen" is a story. "Why clients choose us" is a brochure. - **It surprises, or settles an argument.** A gap between what people say and what they do is the classic headline. - **It is specific.** "Many firms struggle with hiring" is not a finding. A named role, region and time period is. - **It is useful.** If a reader can take the number into a board meeting, an editor will want it. - **It does not simply argue for what you sell.** A cyber security firm concluding that everyone should spend more on cyber security reads as marketing. Showing which controls prevented incidents reads as evidence. **Example:** here is how one study might fit together. The figures are invented to show the shape, not real results. | Part | Hypothetical example | |---|---| | Question | Have finance directors changed how they chase late payment since interest rates rose? | | Method | Online panel survey of 300 UK finance directors at firms with 50 to 500 staff, plus ten interviews | | Headline finding | Six in ten call cash collection a top priority, but only one in five has changed their credit control process in three years | | Base line | "Of 300 UK finance directors surveyed online in [month and year]" | | Pitch line to a finance trade title | "Most FDs say cash is king. Few have changed how they collect it." | ## How much does a study cost, and how long does it take? It depends mostly on who you need to hear from. Your own records cost only your team's time; a panel survey of senior or niche B2B decision-makers costs the most per response, in our experience, because they are harder to reach. Ask two or three providers for a quote on the same brief before you decide. | Route | What you pay for | Time, in our experience | |---|---|---| | Your own records | Staff time to extract, anonymise and analyse | Days to a few weeks once the data is pulled | | Public data | Staff time to join and explain sources | A few weeks | | Interviews | Staff time, or a researcher to run and write them up | Several weeks, depending on diaries | | Panel survey of 250 to 500 UK decision-makers (providers such as Censuswide or YouGov) | Recruitment and fieldwork per completed response, plus your time on questions and analysis | Fieldwork often takes days; the whole study, several weeks | | Full agency study | Design, fieldwork, analysis, report and often the press work | Longer, and priced per project | Most of the time goes on the question, the questionnaire, the analysis and the packaging, not the fieldwork. ## How do we run a small study properly? You do not need a research department, but you do need discipline. ### Step 1: Pick one question your buyers argue about Choose a question that comes up in sales meetings, where there is disagreement or no good UK data. Write it as one sentence. ### Step 2: Choose the method that fits Use your own records if they answer it, a survey if you need to know what a group thinks or does, and interviews if the group is small. A survey plus a few interviews gives you numbers and quotes. ### Step 3: Write fair questions Avoid leading questions such as "How concerned are you about rising costs?", which assumes costs are rising and the respondent is worried. Keep a survey to about ten minutes. Ask role, sector, size and region so you can split the results. Test it on five people first. ### Step 4: Collect enough of the right responses Who answers matters more than how many. A few hundred finance directors of UK firms beats several thousand members of the public for a B2B story. Use a provider bound by the [Market Research Society Code of Conduct](https://www.mrs.org.uk/standards/code-of-conduct), and screen out anyone outside your target group. ### Step 5: Write the method note before anything else Pick the three to five clearest findings. Then write a short method note: who you asked, how many, when, how they were recruited and whether results were weighted. If you cannot write that paragraph plainly, the findings are not ready. ## What sample size does a survey need to get press coverage? There is no single threshold. Journalists judge the sample against the claim: a few hundred respondents from the right group supports a claim about that group, and a few dozen can support a snapshot if described as one. For example, the textbook margin of error for a result near 50%, at 95% confidence, shrinks slowly as the sample grows: | Completed responses | Margin of error | |---|---| | 100 | about 10 points either way | | 250 | about 6 points | | 500 | about 4 points | | 1,000 | about 3 points | Online panels are not random samples, so treat these as a best case. Splits matter most: 500 responses across five regions is about 100 per region, so plan around the smallest group you want to quote. ## Which survey companies do UK journalists trust? Journalists care more about whether the method holds up than about the brand. When members of the British Polling Council publish a poll, they must disclose the client, fieldwork dates, method, the population represented, sample size and coverage, and put full tables online within two working days ([BPC rules](https://www.britishpollingcouncil.org/objects-and-rules/)). Firms in the Market Research Society's Company Partner scheme are bound by its Code of Conduct, whose current edition took effect on 15 May 2023. Whoever you use, ask for the full data tables, publish the method note and name the provider. A study you can defend line by line is one journalists will quote. ## How do we package the findings so a journalist can use them in an afternoon? Put the findings on a proper web page first, with a PDF as a download at most. A PDF is hard to link to, quote from or update. A good findings page has: - **One citable sentence per finding**, with the figure and its base, for example "40% of the 300 finance directors surveyed said..." - **A chart for each finding**, with the sentence as text beside it, and the full method note and data tables on the page. - **A reuse licence.** A licence such as [Creative Commons CC BY 4.0](https://creativecommons.org/licenses/by/4.0/) lets anyone share and adapt the findings, even commercially, provided they give credit and link back. That is what you want. - **Dataset markup.** Google supports [Dataset structured data](https://developers.google.com/search/docs/appearance/structured-data/dataset), which needs a name and description and recommends fields such as creator, licence, temporal coverage and the variables measured. It helps search engines read the study correctly; it is not a ranking trick. - **A stable address for each edition**, so last year's coverage still links to last year's figures. Alongside it: a one-page summary, a spokesperson available that day, and a personal note to each journalist leading with the finding that matters to their readers. ## How do I write a press release for survey results? Lead with the finding, give the base in the first paragraph, and put the method in the notes to editors. A release about the market gets read; a release about your firm does not. 1. **Embargo line**, if there is one: "EMBARGOED UNTIL 00:01 [day, date]". 2. **Headline**: the finding, in plain words, with no firm name. 3. **First paragraph**: the headline figure, its base ("of 300 UK finance directors surveyed"), and who ran the study. 4. **Second paragraph**: one or two supporting findings, including any surprise. 5. **Quote**: your founder on what it means for decision-makers, in two or three sentences. 6. **Notes to editors**: the method note, a link to the findings page and tables, and the spokesperson's direct contact details. ## What is a press embargo, and how long should it be? An embargo is an agreement that a journalist receives the research early but publishes nothing before a set time. It gives them time to write a better piece, and it rests on trust, so agree it with each journalist before you send anything. In our experience a few working days is enough for most trade titles, with more for monthly print titles whose deadlines fall well before publication; ask the journalist what they need. Put the embargo line at the top of every document and in the email subject. If a title breaks it, tell the other journalists who had it that the story is out and they may publish, then deal with the title that broke it privately, and think twice before trusting it with the next one. ## How do we pitch research without being ignored? Pitch it personally, with an angle for each reader, and give one title a head start. - **Offer an exclusive first.** Give the title your buyers respect most the research a few days early, under embargo, on the understanding it publishes first. If it declines, offer it to the next. - **Then pitch the rest with their own angle.** The regional split for a regional paper, the sector split for a sector title. Five pitches with five angles beat one release sent to five hundred addresses. - **Time it sensibly.** Avoid weeks dominated by big news, such as a Budget, and tie the release to a date in your sector's calendar. - **Follow up once, then move on.** More than that damages the relationships described in [how to get trade journalists to call us for comment](/insights/building-relationships-journalists-editors). ## How does one study keep working for a year? Write a short article on your site for each main finding, offer a trade title a guest piece, present the findings at a client event, and give your sales team the one-page summary. Then repeat it with the same questions. The second edition is a better story, because now there is a trend, and journalists who quoted you last year look for the update. Keep every edition live at its own address. When we rebuilt arum's website, "215 articles, research pieces and news items moved across, going back to 2014", and "Every old address redirected, so years of rankings carried over." ([arum case study](/work/arum)). That steady presence is what [what twelve months of consistent PR looks like](/insights/compounding-effect-consistent-pr-12-months) describes. ## Which mistakes get research ignored or distrusted? Leading questions, small samples presented as "what UK businesses say", a hidden method, the best number buried on page seven, and only publishing findings that suit you. Research that sometimes surprises its sponsor is more credible, and more quotable, than research that always agrees with it. ## What to do this month 1. List three questions your buyers argue about in sales meetings where nobody has good UK data. 2. Check whether your own records could answer one of them, and what you would need to anonymise first. 3. If they cannot, draft a ten-minute survey and ask two or three providers to quote on the same brief. 4. Choose the trade title you would most like to break the findings, and find the journalist who covers that subject. 5. Put a date in the diary for a second edition twelve months after the first, so the study is a series from the start. ## Frequently asked questions ### Should we use a research firm or do it ourselves? For a survey, a panel provider is usually worth paying for, because finding the right respondents is the hard part. For your own records or client interviews, doing it yourself is often better. Many firms write the questions themselves and pay for recruitment and an independent check of the analysis. ### What if the results are not what we expected? Publish them anyway, framed plainly. Unexpected results are often the better story, because they challenge an assumption. Firms that only publish findings in their own favour soon find journalists stop taking their research seriously. ### Can we use client data in published research? Often yes, if it is aggregated and cannot identify any client, but check your contracts and data protection obligations first and take advice where there is doubt. Some firms also ask a few clients to be named alongside the research, which gives journalists real examples to follow up. ### How long does a piece of research stay useful? In our experience a study is quoted until something newer exists on the topic, then journalists want current numbers. That is why a repeated annual study is worth more than a one-off: each edition refreshes the figures and adds a trend, and older editions become the baseline. ### What if a competitor has already published research on the same topic? That shows the topic matters to journalists. Ask a different question, survey a different group, or cover a region or sector they left out. Repeating their study with a smaller sample does not work. If you want help choosing the question, running the study and getting it in front of the right editors, that is part of [Reputation That Wins](/reputation-that-wins). We usually start with the questions your buyers already ask in sales meetings. ## Sources - [Muck Rack, What is AI reading? Earned media drives 84% of AI citations (May 2026)](https://www.globenewswire.com/news-release/2026/05/07/3290268/0/en/generative-pulse-earned-media-consistently-drives-ai-citations-holding-at-84.html) - [British Polling Council, Objects and rules (September 2026)](https://www.britishpollingcouncil.org/objects-and-rules/) - [Market Research Society, Code of Conduct (May 2023)](https://www.mrs.org.uk/standards/code-of-conduct) - [Google Search Central, Dataset structured data (September 2026)](https://developers.google.com/search/docs/appearance/structured-data/dataset) - [Creative Commons, Attribution 4.0 International licence (September 2026)](https://creativecommons.org/licenses/by/4.0/) --- Title: How does a commercial property firm get named when businesses ask ChatGPT? URL: https://fortitudemedia.ai/insights/ai-optimisation-property-real-estate Author: Ross Williams Published: 2026-03-31 Last reviewed: 2026-09-24 Topic: Sector guides Short answer: ChatGPT names commercial property firms it can match to a place, a property type and a service, and then verify. State your patch, sectors and size bands plainly, publish a deal record and dated market reports, and explain how you charge for lease work. Keep your website, Google, Companies House and RICS Find a Surveyor listing consistent, and get credited on deals in the trade press. Proptech suppliers are judged mainly on review sites and fair comparison pages. For example, an operations director needs warehouse space near the M6 and wants an agent who knows the West Midlands industrial market. A retailer has a lease renewal coming up on a shop in Manchester and wants someone to argue the rent down. A landlord wants to replace the spreadsheets that track its leases. Not long ago, each would have rung a contact or asked their solicitor. Now many start by asking ChatGPT, Microsoft Copilot or Google, and the answer comes back with two or three names. This is now normal business behaviour. [Forrester's January 2026 research](https://www.forrester.com/blogs/b2b_buyers_make_zero_click_buying_number_one/) found that 94% of business buyers use AI somewhere in their buying process, and that business buyers are four times as likely as consumers to use Copilot, the assistant many of them already have at work. If your firm is not among the names the assistant gives, the instruction, the lease advisory brief or the software trial goes elsewhere before your phone rings. The good news for regional agencies, lease advisers, property consultancies and proptech suppliers is that these questions are specific: a use class, a size, a place, a type of lease. A firm that knows its patch and publishes its record can be named ahead of a bigger brand whose local pages are thin. This guide is for the owner or partner who wants that name to be theirs, with a short section for residential agents towards the end. ## Key takeaways - AI assistants name commercial property firms they can match to a place, a property type and a service, so a vague "commercial property services" page rarely gets you named. - Your deal record is your strongest evidence: publish the size, type and location of transactions you have handled, with client permission, and make sure you are credited in the trade press. - Your details must match across your website, Google, Companies House, LinkedIn and the RICS Find a Surveyor directory, or assistants hedge and name a competitor. - Quarterly market reports for a named sector and region make you the source assistants quote when occupiers and landlords ask what is happening. - Buyers ask what lease renewals and rent reviews cost, so explain how you charge on your own site rather than hiding behind "contact us". - Proptech suppliers are judged like any software firm: review sites, comparison pages and named customers carry the most weight. ## What do occupiers, landlords and investors ask ChatGPT? They describe a requirement, a problem or a worry, and they name a place. Typical questions look like this: - "Which commercial property agents in Birmingham are best for industrial units?" - "Who can advise on a lease renewal for a shop in Manchester?" - "How much do commercial agents charge for a lease renewal or rent review?" - "Should we use a lease adviser for our rent review or negotiate it ourselves?" - "Do we need to improve our office's EPC before we re-let it?" - "Which lease management software do UK commercial landlords use?" - "Is [your firm] any good?" Choosing questions ask who to use. Deciding questions ask what something costs and what the rules require. Checking questions ask whether a named firm can be trusted, usually after a pitch. Each is answered from different evidence, and most firms only prepare for the first. ## What does an AI assistant check before it names a commercial property firm? Broadly, five things, and whether they all agree. | Evidence | Where assistants and buyers find it | What to do | |---|---|---| | Patch and specialism | Your service and office pages, Google Business Profile | Name the towns, sectors and size bands you cover, per service line | | Regulation and people | RICS Find a Surveyor, your team pages | Keep your RICS firm listing current and name your chartered surveyors and RICS Registered Valuers | | Deal record | Your deal pages, EG, regional business press, CoStar, your clients' own announcements | Publish deals with permission and make sure you are credited under one name | | Market knowledge | Your market reports, trade press comment | Publish dated reports for a named sector and region | | Client view | Google reviews, named testimonials, case studies | Ask for reviews that mention the job and the outcome | ### Your patch and specialism "Commercial property services across the Midlands" is hard to match to anything. "Industrial and logistics agency in Birmingham, the Black Country and along the M6 and M42, from trade counters to large distribution sheds" is easy. Say the same for each service line: office acquisitions, retail lease advisory, rating, valuation, investment. ### Your regulation and your people The [RICS Find a Surveyor directory](https://www.ricsfirms.com/) lets users search regulated firms by location and by commercial service, including industrial, offices and retail. Check your listing shows the services you want to be known for and matches your website exactly. Name the surveyors who lead each service line, with their RICS status and the deals they have led. Buyers choose people as much as firms. ### Your deal record, and where it appears CoStar and EG are where the market records deals and who acted on them. Much of that sits behind a subscription, so what an assistant can usually read is the public side: deal news in EG and the regional business press, your clients' announcements and your own deal pages. Buyers and journalists read the subscription side too. Make sure every mention credits you under the same firm name, and keep a deal list on your site with property type, size band, location and your role. A deal record an assistant can cross-check is worth more than any claim about being "market leading". ### Whether everything agrees Agencies merge, rebrand and close offices, and directories keep the old details for years. If Google shows an office you closed and your RICS listing uses a former name, an assistant has several versions of the truth and often resolves it by naming someone else. We cover how to find and fix these conflicts in [why ChatGPT describes your firm wrongly](/insights/ai-conflicting-information-online). ## How does an assistant answer "best industrial agents in Birmingham"? It runs several searches and names the firms whose evidence best fits the question. ChatGPT search, Copilot and Google's AI features all start from a search index, and Google now breaks one question into several related searches, which it calls query fan-out. That is why ranking helps but does not guarantee a mention: [Ahrefs found in March 2026](https://ahrefs.com/blog/ai-overview-citations-top-10) that only 38% of AI Overview citations came from pages ranking in the top ten, down from 76% in July 2025. The follow-up searches are predictable: industrial availability in the West Midlands, who let a particular scheme, rents for mid-box units. A firm that answers those on its own site, with a named industrial team, recent lettings and a dated quarterly report, gives the assistant several reasons to name it. A firm with one "Agency" page and a property search feed gives it none. ## Market reports: the content that makes you the local source Market commentary is what assistants reach for when an occupier asks what rents are doing or a landlord asks how long a letting will take. Data providers give the numbers, not what they mean for this sector, in this place, this quarter. A firm that publishes that regularly becomes the source. Here is a format that works for people and for machines. ### Step 1: Define your patch and sector Report on a named sector and area, such as "industrial and logistics, West Midlands". One focused report beats a national overview that repeats what the large firms publish. ### Step 2: Combine published figures with your own Start with figures you can cite and are licensed to quote; check your data provider's terms first. Then add what only you can see: enquiry levels, how incentives are moving, how asking rents compare with agreed rents. Keep client details out of it. ### Step 3: Say what it means This earns the recommendation. A sentence on what is happening to a named size band in a named place, and what occupiers should expect, is the kind an assistant can repeat. A table without a view is not. ### Step 4: Give advice by type of client End with a short paragraph each for occupiers, landlords and investors. Say what you would do in their position this quarter. ### Step 5: Date it and keep the archive Put the quarter in the title and keep every past report. A dated run shows you have tracked the market consistently, and assistants lean towards recent sources. Our article on [how old is too old for content](/insights/content-freshness-recency-ai) explains why dates matter. ## What do lease renewals and rent reviews cost, and should you say? Yes, say how you charge. Occupiers ask assistants what lease advisers charge and whether they need one, and an assistant can only quote what it can find. If your site says "contact us for a bespoke quote", it quotes someone else. - Explain your fee basis for lease renewals, rent reviews and lease re-gears: a fixed fee, a share of the saving against the landlord's opening figure, or a mix, and what is included. - Give typical ranges on your own site for the instructions you handle most, with the date they apply to. - Explain the process and the timetable, including the heads of terms stage and RICS's professional standard, the Code for leasing business premises, which covers business lettings in England and Wales. - Say when a tenant could reasonably negotiate alone, such as a small unit with a simple review clause, and when advice pays for itself. That last point feels risky, yet a fair answer to "do I need an adviser?" is more quotable and more believable than a claim that advice is always essential. ## Energy standards: a question every landlord now asks Landlords ask assistants whether a building can be let and what upgrades it needs, and a clear, dated explainer makes your firm the one quoted. The current rule in England and Wales is that privately rented non-domestic property needs at least an EPC E, and since 1 April 2023 that has applied to all such lettings, not only new ones ([GOV.UK landlord guidance](https://www.gov.uk/guidance/non-domestic-private-rented-property-minimum-energy-efficiency-standard-landlord-guidance)). In its [June 2026 interim response](https://www.gov.uk/government/consultations/non-domestic-private-rented-sector-minimum-energy-efficiency-standards-epc-b-implementation/outcome/minimum-energy-efficiency-standards-mees-in-the-non-domestic-private-rented-sector-interim-response), the government said it proposes EPC B from 2031 for rented buildings over 1,000 square metres, and dropped the interim EPC C milestone planned for 2027. Those changes need secondary legislation first, so describe them as proposals and date your page. ## Proptech suppliers: how assistants answer software questions Asked which lease management software UK landlords use, an assistant builds its answer from review sites, comparison articles and vendors' pages. Buyers trust the independent part most: in [G2's April 2026 survey of 1,076 software buyers](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html), 51% said they now start research in an AI chatbot, and 45% said citations of review sites were the most confidence-inspiring signal in an AI answer. - Name your category in the words landlords use, such as lease management or service charge budgeting, rather than a brand phrase. - Ask customers for reviews on the software review sites your buyers read, and name customers with their permission. - Publish fair comparison pages for the alternatives buyers weigh you against, including when a competitor suits a buyer better. - List integrations plainly, because "does it work with our accounting system?" is a common follow-up. Comparison pages work when they answer the buyer's real question. For [Helm Club](/work/helm-club), "We built the structured data AI engines read, then a comparison page for every question a founder asks when weighing up peer groups: Helm against YPO, Vistage and the rest." And: "Every comparison page we have published ranks first for its question on Google and the AI engines." Our guide for [B2B technology companies](/insights/ai-optimisation-technology-companies) covers the rest of the software playbook. ## Residential agents: what changed in 2026 Landlords now ask assistants what 2026's changes mean for them. Under the government's [implementation roadmap](https://www.gov.uk/government/publications/renters-rights-act-2025-implementation-roadmap), the Renters' Rights Act 2025 took effect in England on 1 May 2026: section 21 "no fault" evictions were abolished, tenancies moved to assured periodic tenancies, rental bidding was banned and rent in advance was capped at one month. A private rented sector database rolls out from late 2026, and membership of the new landlord ombudsman is expected to become mandatory in 2028. - Publish a plain explainer on what the Act means for your landlords' fees, notice and services, and date it. - Keep your client money protection certificate on show: GOV.UK says agents must display it in any office where they deal with the public and on their website ([GOV.UK](https://www.gov.uk/client-money-protection-scheme-property-agents)). - Keep showing your redress scheme and Propertymark membership on every branch page. - Never buy or write reviews. Fake reviews became a banned practice under the Digital Markets, Competition and Consumers Act 2024 in April 2025, and the CMA can fine up to 10% of global turnover ([CMA, March 2026](https://www.gov.uk/government/news/fake-and-misleading-reviews-5-businesses-under-cma-investigation)). - Show how you handle material information in listings. National Trading Standards has guided agents on this since its Estate and Letting Agency Team published Part A in 2022, and in June 2026 the government announced plans for homebuying sales packs ([National Trading Standards](https://www.nationaltradingstandards.uk/work-areas/estate-agency-team/)). ## Why your website may be hiding you from AI Most property websites are built around a listings feed and an enquiry form, which leaves assistants little to read. Listings disappear when a unit lets, office pages are identical apart from the address, and reports sit in PDFs. Add pages that stay: service lines, deal lists, dated reports, fee pages and team pages. The main AI crawlers fetch pages but do not run JavaScript, so key text must be in the page itself, not loaded by a search widget. Ask your web provider to confirm AI crawlers are not blocked, since Cloudflare has blocked them by default on new domains since July 2025. Our guide to [how AI crawlers read your website](/insights/ai-crawlers-vs-google-spiders) explains the settings. ## How do you tell whether it is working? Keep a fixed list of 20 to 50 client questions, by sector, service and town, and run each several times a month in ChatGPT, Copilot, Gemini and Google in a clean session. Answers vary from run to run, so count a win only when it repeats. Add "AI assistant" to the options when you log how a new instruction found you. Our guide to [tracking whether ChatGPT recommends your firm](/insights/track-ai-recommending-your-business) sets out the method. Assistants that search the web live can reflect a new page or fixed listing within weeks. Answers drawn from a model's memory change only when it is retrained, which takes months. For [ID-Pal](/work/id-pal), an identity verification firm we work with, the dated result was this: "UK buyer questions where AI names ID-Pal: none in May, two in August, four in September 2026." ## What to do this month 1. Write your service and office pages so each names its places, sectors and size bands in plain words, with the surveyors who lead it. 2. Check your RICS Find a Surveyor listing, Google Business Profile, Companies House record and LinkedIn page, and correct anything that does not match your website. 3. Publish a deal list for your strongest service line, with property type, size band, location and your role, cleared with clients. 4. Publish your first quarterly market report for one sector and region, using the five steps above. 5. Write one page explaining how you charge for lease renewals and rent reviews, and when a tenant might not need you. ## Frequently asked questions ### Should each office and service line have its own page? Yes. Assistants answer place by place and service by service, so each office needs a page with its address, team, areas covered and recent deals, and each service line needs its own page. Copies with only the town name changed do little. A page with local deals and commentary, written by the people who work there, gives an assistant a reason to name that team. ### Is it risky to publish our fee basis when competitors can see it? Competitors already learn your fees by pitching against you. The people who cannot are occupiers comparing advisers, who increasingly ask assistants to do it for them. Publishing how you charge puts you in that comparison on your own terms, and you still quote each instruction individually. ### Can a regional independent beat the national firms? Often, for local and sector questions. Assistants look for the best-evidenced answer for a particular place and property type, and a regional firm with a clear specialism, a readable deal record, current market reports and consistent listings can be named ahead of a national firm whose local page is thin. ### What does the Renters' Rights Act mean for a letting agent's fees and services? From 1 May 2026, landlords in England can no longer use section 21, tenancies are periodic, and agents cannot run rental bidding or take more than a month's rent in advance. The Tenant Fees Act still applies. An agent that explains its services and fees against the new rules, plainly and with a date, is the one an assistant can quote. ### Can we pay to appear in ChatGPT's answer? Not in the answer itself. ChatGPT has shown labelled ads to UK users on its Free and Go plans since 11 August 2026, and OpenAI says ads do not change the answer. Being named in the answer is still earned through the evidence described above. If your website is a listings feed with a contact form, our [Website That Sells](/website-that-sells) service turns it into something buyers and assistants can read and quote. If you would like a view on where you stand first, [book a free Growth Review](/growth-review). ## Sources - [Forrester, B2B buyers make zero-click buying number one (January 2026)](https://www.forrester.com/blogs/b2b_buyers_make_zero_click_buying_number_one/) - [G2 via PR Newswire, Half of B2B software buyers now start their research with AI chatbots (April 2026)](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html) - [Ahrefs, AI Overview citations from top-ten pages (March 2026)](https://ahrefs.com/blog/ai-overview-citations-top-10) - [GOV.UK, Non-domestic private rented sector MEES: interim response (June 2026)](https://www.gov.uk/government/consultations/non-domestic-private-rented-sector-minimum-energy-efficiency-standards-epc-b-implementation/outcome/minimum-energy-efficiency-standards-mees-in-the-non-domestic-private-rented-sector-interim-response) - [GOV.UK, Non-domestic private rented property: minimum energy efficiency standard, landlord guidance (May 2026)](https://www.gov.uk/guidance/non-domestic-private-rented-property-minimum-energy-efficiency-standard-landlord-guidance) - [MHCLG, Implementing the Renters' Rights Act 2025: our roadmap (November 2025)](https://www.gov.uk/government/publications/renters-rights-act-2025-implementation-roadmap) - [CMA, Fake and misleading reviews: 5 businesses under CMA investigation (March 2026)](https://www.gov.uk/government/news/fake-and-misleading-reviews-5-businesses-under-cma-investigation) - [GOV.UK, Protecting clients' money if you're a property agent (accessed September 2026)](https://www.gov.uk/client-money-protection-scheme-property-agents) --- Title: How does a construction firm get named when clients ask ChatGPT? URL: https://fortitudemedia.ai/insights/ai-optimisation-construction-trades Author: Ross Williams Published: 2026-03-31 Last reviewed: 2026-09-24 Topic: Sector guides Short answer: ChatGPT names construction firms it can place clearly: a stated specialism, a defined region, recent written-up projects and accreditations it can check. Most contractors' websites show a gallery and a phone number. Write project pages that say what, where, for whom and how, make your Constructionline, SSIP, Common Assessment Standard and self-certification listings match, and keep dated pages on the rules clients worry about, such as Part L and the Building Safety Regulator. The estates manager for a group of independent schools needs the heating replaced in two boarding houses over the summer holidays, with no room for slippage. Before she calls the firms on last year's list, she asks ChatGPT: "which M&E contractors in Surrey have experience working in occupied schools over the summer". She gets three names. Your firm did exactly that job for two schools last August, on programme. You are not mentioned. Construction still runs on repeat clients, supply chains and frameworks. What has changed is where the first shortlist comes from when the client is new or a procurement team wants names to invite. In a 2025 survey of more than 4,000 B2B buyers, 6sense found that buying groups had already ranked their preferred suppliers before first contact in 94% of cases ([6sense, November 2025](https://6sense.com/newsroom/the-timeline-for-influencing-b2b-buyers-is-shrinking-insights-from-6senses-2025-buyer-experience-report/)). More of that early research now happens in ChatGPT, Google's AI Overviews and AI Mode, Microsoft Copilot, Gemini, Perplexity and Claude. Most contractors' websites show a gallery captioned "Commercial project, Guildford", a list of services and a contact form. That tells an AI assistant almost nothing. This guide is for the owner of a specialist contractor, M&E firm, fit-out company or regional builder who wants to be on that shortlist. ## Key takeaways - Clients and main contractors now ask AI assistants who to shortlist, and name contractors they can place clearly: a specialism, a region, recent projects and checkable accreditations. - A written project page that says what, where, for whom and how is worth far more than a gallery of uncaptioned photos. - Your Constructionline, SSIP, Common Assessment Standard and self-certification scheme details should match your website and Companies House exactly. - Dated compliance pages matter because the rules are moving: the Building Safety Regulator left the HSE in January 2026, and new Part L standards apply in England from March 2027. - Start with ten project write-ups and a clean set of listings, then track a fixed list of client questions. ## What do clients ask before they pick a contractor? They ask about their project, their constraints and their risks. The people asking are estates managers, developers, housing associations, project managers and main contractors filling a package. Typical questions: - "Which fit-out contractors in Manchester have done Cat B office fit-outs for professional firms?" - "Who are the best M&E contractors in the South West for schools and colleges?" - "Which fire door installers near Birmingham are third-party certified?" - "Do I need Constructionline Gold or is CHAS enough to get on a main contractor's supply chain?" - "Who approves building work on higher-risk buildings now?" - "Who can replace a flat roof on a warehouse without stopping operations?" Each names a type of work, and most name a place, a type of building or a constraint such as "occupied". An assistant answers by looking for firms whose websites and listings use the same words about real projects; "delivering quality construction services across the UK" matches none of them. Clients also research the rules before they research contractors, so a firm that explains the compliance question well is the natural one to name. ## How does ChatGPT decide which contractors to name? It names the firms it can place with confidence, using four kinds of evidence: - **A specialism in plain words.** "Mechanical and electrical services for schools, colleges and care homes" can be matched. "Building services solutions" cannot. - **A region.** Where you are based and the counties or cities you cover. - **Recent, written-up projects.** The type of building, the client type, the location, the scope and the outcome. - **Checks that agree.** Accreditations, trade bodies, Companies House, your Google Business Profile and reviews that say the same as your website. Assistants look for agreement. If Companies House lists one name, your website a shortened trading name and Constructionline an address you left three years ago, the assistant has three versions of you to reconcile. Size is not the deciding factor: a twenty-person specialist with thirty well-written project pages can be named ahead of a national contractor. ## Why do project pages matter more than your homepage? Project pages are the evidence: where an assistant learns what you build, for whom and where, and what a project manager reads before inviting you to tender. Most contractors have the raw material in their photo library and tender submissions. A good project page covers the building and client type ("a 1960s secondary school, occupied during term, for a multi-academy trust"), the location, the scope and a value band if the client agrees, what made it hard, what you did about it and the outcome, and the certifications that applied, with a short client quote given with permission. When we rebuilt [arum's website](/work/arum), "Each of the 12 case studies leads with a number, and each of the nine sector pages names the regulation in play." The same discipline works for a contractor: lead with the programme, the size or the saving, and name the rules the job had to meet. Assistants read text, and get little from an uncaptioned photo or a brochure saved as an image. The main AI crawlers fetch pages but do not run JavaScript ([Vercel, December 2024](https://vercel.com/blog/the-rise-of-the-ai-crawler)), so put the words in the page itself and link every project page normally. Our note on [how AI crawlers read your website](/insights/ai-crawlers-vs-google-spiders) explains more. ## Which accreditations and schemes do clients check? The ones your clients' procurement teams check. Each listing should use the same company name, address and description as your website. | Type | Examples | What it answers | |---|---|---| | Health and safety prequalification | SSIP member schemes such as CHAS, SafeContractor and SMAS Worksafe | "Is this firm safe to have on site?" | | Supply chain prequalification | Build UK's Common Assessment Standard, through Constructionline and other recognised bodies | "Will a main contractor accept them?" | | Self-certification schemes | NICEIC or NAPIT for electrical work, Gas Safe Register, MCS for heat pumps and solar | "Who is certified to do this work?" | | Third-party certification | FIRAS or BM TRADA for fire doors and fire stopping | "Is the installer certified?" | | Trade and quality bodies | FMB, BESA, ECA, TrustMark, Considerate Constructors Scheme | "Is this an established firm?" | | Management standards | ISO 9001, 14001 and 45001, with the certifying body | "Do they run a controlled business?" | The schemes that let installers sign off their own work under the Building Regulations are now called self-certification schemes, and the government says scheme operators must keep a public register of every certified business ([GOV.UK, July 2026](https://www.gov.uk/government/publications/self-certification-schemes-mandatory-technical-competence-requirements)). Check your entry is current, because it is the record an assistant can find. Main contractors increasingly ask for Build UK's Common Assessment Standard as well as SSIP; the questions below cover which certificate to hold. ## How should you write about cost, compliance and programme? Explain what drives cost, how you handle the rules and how you keep a job on programme, without publishing anything you would price in a tender. **Cost.** You do not need a price list. A client asking what a Cat A office fit-out costs per square foot in Manchester wants to know what makes it cost more or less: the state of existing services, working hours, access and lead times on plant. A page that explains those factors is quotable. If you give ranges from your own projects, date them and state the assumptions; tender pricing itself is best kept off the site. **Compliance (checked 24 September 2026).** Clients are nervous about the rules, and three facts have moved: - The Building Safety Regulator is still the building control authority for higher-risk buildings, but on 27 January 2026 it left the HSE to become a standalone body under the Ministry of Housing, Communities and Local Government ([GOV.UK, January 2026](https://www.gov.uk/government/news/bsr-becomes-standalone-body-in-landmark-step-towards-single-construction-regulator)). Applications for higher-risk building work still go to the regulator, not to local authority or private building control. - In England, the current Part L guidance is [Approved Document L, 2021 edition incorporating 2023 amendments](https://www.gov.uk/government/publications/conservation-of-fuel-and-power-approved-document-l). The Future Homes and Buildings Standards were published on 24 March 2026, and the amending regulations come into force on 24 March 2027, or 24 September 2027 for higher-risk building work, with transitional arrangements for work already notified ([GOV.UK, March 2026](https://www.gov.uk/government/publications/the-future-homes-and-buildings-standards-building-circular-012026)). - CDM 2015 still requires a principal designer to plan, manage and coordinate health and safety before construction on any project with more than one contractor ([HSE](https://www.hse.gov.uk/construction/cdm/2015/principal-designers.htm)). The Building Regulations have a separate principal designer role, which our [guide for architecture practices](/insights/ai-optimisation-architecture-design) explains. Date these pages and review them, because a stale compliance page does more harm than none. **Programme and disruption.** Schools, surgeries and warehouses ask whether you can work while they keep running. Pages on how you phase work in occupied buildings show you have done it before. ## How narrow should you go on specialism and region? Narrower than feels comfortable. A firm that is clearly the specialist in one or two types of work in one region is named far more often than one that claims everything everywhere, so state your area and back it with projects there. Avoid dozens of near-identical town pages with the name swapped, which search engines and assistants treat as thin, as our article on [why thin pages get ignored](/insights/ai-penalises-thin-content) explains. ## How do you tell whether it is working? Pages and listings can show within weeks in assistants that search the web live, such as ChatGPT search, Perplexity, Copilot and Google's AI features. Answers drawn from a model's own memory change only when it is retrained, which takes months. Track a fixed list of 20 to 50 questions your best clients would ask, each run several times a month in a clean session, and count a win only when it repeats. Then ask "how did you hear about us" on every enquiry and count invitations to tender from new clients. ## What to do this month 1. **Fix your listings.** Match your company name, address and description across Companies House, Constructionline, your SSIP scheme, your self-certification scheme register, Google Business Profile and your website. 2. **Write an accreditations page.** List every scheme, standard and trade body in full, with the level you hold and a link to each listing. 3. **Write up ten projects.** Pick recent jobs in the work you want more of and publish a short page for each. 4. **Date your compliance pages.** Update anything on the Building Safety Regulator or Part L and add a "last checked" date. 5. **Decide what you lead with.** Agree one or two specialisms and a region, and rewrite your homepage opening to say so. ## Frequently asked questions ### Our work comes from repeat clients and main contractors. Does this matter for us? Yes, though less urgently than for a firm chasing new clients. Main contractors and project managers use assistants to check firms they have been given, and a firm that looks thin online can lose out at that stage even when the introduction came through a relationship. ### Can we name clients and show project values? Only with permission, which many clients give at handover. Otherwise, describe the client type and building instead, such as "a regional housing association" or "a Grade II listed office in York", and give a value band rather than an exact figure. The assistants need specifics about the work, not necessarily the client's name. ### Will a national contractor always be named ahead of us? No. National firms tend to be named for broad questions, but most client questions are narrow: a type of work, a building, a place and a constraint. A regional specialist with clear evidence for that combination can be the closer match. ### What is the Common Assessment Standard, and do subcontractors need it? It is Build UK's single prequalification standard for construction supply chains, delivered by seven recognised assessment bodies, including Achilles, Constructionline, SMAS Worksafe and Veriforce CHAS, at a desktop or a site-based level, with a building safety section ([Build UK, July 2026](https://builduk.org/priorities/common-assessment-standard/)). Build UK says a growing list of contractors and clients accept it. If the main contractors you want to work for use it, you need it; if they ask only for SSIP, that may be enough for now. ### Do I need Constructionline Gold, or is CHAS enough? It depends on what the main contractor asks for, so ask them. Every Constructionline level includes SSIP health and safety, Gold is the level that meets the Common Assessment Standard, and Platinum adds an on-site audit ([Constructionline, September 2026](https://www.constructionline.co.uk/products-services/memberships/sub-contractor-memberships/)). CHAS can also certify to the standard. Find out which certificate and level each main contractor accepts, then say plainly on your accreditations page which one you hold. The contractors who get named are rarely the biggest. They are the ones whose work is written down where clients and assistants can read it. If your website is still a gallery and a contact form, our [Website That Sells](/website-that-sells) service turns your projects and accreditations into pages that bring enquiries in. ## Sources - [6sense, 2025 Buyer Experience Report (November 2025)](https://6sense.com/newsroom/the-timeline-for-influencing-b2b-buyers-is-shrinking-insights-from-6senses-2025-buyer-experience-report/) - [Vercel, The rise of the AI crawler (December 2024)](https://vercel.com/blog/the-rise-of-the-ai-crawler) - [GOV.UK, Self-certification schemes: mandatory technical competence requirements (July 2026)](https://www.gov.uk/government/publications/self-certification-schemes-mandatory-technical-competence-requirements) - [Build UK, Common Assessment Standard (July 2026)](https://builduk.org/priorities/common-assessment-standard/) - [Constructionline, Sub-contractor memberships (September 2026)](https://www.constructionline.co.uk/products-services/memberships/sub-contractor-memberships/) - [GOV.UK, BSR becomes standalone body (January 2026)](https://www.gov.uk/government/news/bsr-becomes-standalone-body-in-landmark-step-towards-single-construction-regulator) - [GOV.UK, The Future Homes and Buildings Standards: Building Circular 01/2026 (March 2026)](https://www.gov.uk/government/publications/the-future-homes-and-buildings-standards-building-circular-012026) - [HSE, CDM 2015: principal designers (September 2026)](https://www.hse.gov.uk/construction/cdm/2015/principal-designers.htm) --- Title: How does a health tech or healthcare supplier get recommended by ChatGPT? URL: https://fortitudemedia.ai/insights/ai-optimisation-healthcare-medical Author: Ashley Kenerson Published: 2026-03-31 Last reviewed: 2026-09-24 Topic: Sector guides Short answer: ChatGPT and the other assistants name healthcare suppliers they can match to a buyer's question and check against public records. For digital health, that means stating your DTAC position, Data Security and Protection Toolkit status, Cyber Essentials certificate, clinical safety officer and any MHRA registration in plain text on a dated page. For occupational health, it means stated coverage, SEQOHS accreditation and named clinicians. Then publish pages that answer the questions NHS, private sector and HR buyers type, and keep every listing consistent. A digital transformation lead at an NHS trust has six weeks to recommend a remote monitoring pilot. Before she emails a supplier, she asks Copilot on her work laptop which virtual ward platforms other trusts use and what she must check before a pilot. An HR director with forty depots asks ChatGPT which occupational health providers can cover every site. Each assistant replies with three or four names. If yours is not one of them, you are not on the list that goes to procurement, and nobody tells you the list existed. This is now how business buyers work. Forrester found that 94% of business buyers use AI in their buying process, and that they are four times as likely as consumers to use Microsoft Copilot ([Forrester, January 2026](https://www.forrester.com/blogs/b2b_buyers_make_zero_click_buying_number_one/)). In G2's survey of software buyers, 51% said they now start their research in an AI chatbot, up from 29% a year earlier ([G2, April 2026](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html)). Neither survey is specific to healthcare, but health tech is software, and its buyers are busy people with a shortlist to produce. Health is also where the assistants are most careful, leaning on sources they can check: NHS England, NICE, the regulators and public registers. That helps a smaller supplier, because being easy to verify matters more than budget. This guide is for founders of health tech and medtech firms, occupational health providers and other suppliers selling to the NHS, private providers and employers. Patient-facing clinics get one section near the end. ## Key takeaways - AI assistants name healthcare suppliers they can check against public records, so assurance evidence counts for more than any claim about transforming care. - Digital health suppliers should state their DTAC position, DSP Toolkit status, Cyber Essentials certificate, clinical safety officer and any MHRA registration on one dated page, in plain text. - NHS England introduced a shorter DTAC form on 24 February 2026, and the old form should not be used from 6 April 2026, so older assurance packs need updating. - Occupational health providers get named for multi-site questions when their coverage, turnaround, SEQOHS accreditation and named clinicians are written down. - Independent evidence carries the most weight: framework listings, trade press, independently judged awards and case studies that name the customer. - Patient-facing clinics still need verifiable clinicians, a CQC rating shown with its date and details that match every directory. ## What do NHS, private provider and employer buyers ask ChatGPT? They describe a problem, a care setting and a constraint, rarely in the words on your homepage. These are typical of the questions that decide who gets named. | Buyer | Typical question | What the answer leans on | |---|---|---| | NHS digital or transformation lead | "Which virtual ward platforms do NHS trusts use, and what should I check before a pilot?" | Case studies naming trusts, assurance pages, trade press | | NHS procurement | "Which suppliers are on NHS Supply Chain frameworks for patient monitoring?" | Framework listings, MHRA registration, contract notices | | Private hospital group | "Which pre-assessment software works with our patient administration system?" | Integration pages, named customers, G2 and Capterra reviews | | HR director, multi-site employer | "Which occupational health providers cover multi-site employers across the UK?" | Coverage details, SEQOHS listing, named clinicians | | Anyone checking a name | "Is [supplier] any good? Does it meet the DSP Toolkit?" | Reviews, the public DSPT search, news, your assurance page | Most health tech websites answer only "who can do this", with a product tour. The questions about what to check, and whether a named supplier is sound, are where a cautious buyer forms a view. ## Why are AI assistants so careful with health suppliers? Because a wrong answer about health can hurt someone, the assistants are trained to be cautious on medical topics, and that caution carries into supplier questions. They prefer suppliers whose claims match a public record and avoid naming anyone when the evidence is thin. "Clinically proven" and "trusted by the NHS" carry little weight on their own; a named trust, a dated DSPT status and a named clinical safety officer do. We explain the thinking in [how AI assistants judge experience and expertise](/insights/eeat-why-ai-cares-more). A common result is a supplier the assistants can describe but will not recommend. ID-Pal, which sells identity verification into another regulated market, was there: "AI could describe ID-Pal. It never recommended it." Its industry pages were rebuilt so each "opens with the question a compliance lead types, names the regulator, and answers inside the first ninety words". As the [ID-Pal case study](/work/id-pal) records: "UK buyer questions where AI names ID-Pal: none in May, two in August, four in September 2026." The approach transfers directly to NHS England or the MHRA. ## What do NHS trusts check before buying a digital health product? They start with the Digital Technology Assessment Criteria (DTAC), NHS England's framework for assuring digital health technology, then look at data security, cyber security, device regulation and evidence of benefit. NHS England says manufacturers must "keep an up-to-date DTAC form and required supporting documents for each product version" ([NHS England, DTAC](https://transform.england.nhs.uk/key-tools-and-info/digital-technology-assessment-criteria-dtac/)). DTAC covers clinical safety, data protection, technical security, interoperability, and usability and accessibility. NHS England introduced an updated form on 24 February 2026 with 25% fewer questions, and says the previous version should not be used from 6 April 2026. | What buyers check | What it is | Where it can be checked | What to publish | |---|---|---|---| | DTAC | NHS England's assessment criteria, updated February 2026 | Completed by the buyer; no public register | That you hold a current form on the 2026 version, and its date | | Data Security and Protection Toolkit | Self-assessment against ten data security standards, which [all organisations with access to NHS patient data and systems must use](https://digital.nhs.uk/data-and-information/looking-after-information/data-security-and-information-governance/data-security-and-protection-toolkit) | The public [DSPT organisation search](https://www.dsptoolkit.nhs.uk/OrganisationSearch), showing year and status, such as "Standards met" | Your status, the year and a link to the search | | Clinical safety (DCB0129) | The clinical risk management standard for manufacturers of health IT | Your safety case, on request | Your clinical safety officer's name and registration | | Cyber Essentials | The government-backed minimum cyber standard; the NCSC says "a growing number of organisations require suppliers to be certified to bid for work" ([NCSC, September 2026](https://www.ncsc.gov.uk/cyberessentials/overview)) | IASME's certificate search | Level and renewal date | | MHRA registration | Required before a medical device is placed on the Great Britain market | The MHRA's [public registration database](https://www.gov.uk/guidance/register-medical-devices-to-place-on-the-market) | Device class and registration | | Evidence of benefit | NICE's [evidence standards framework for digital health technologies](https://www.nice.org.uk/corporate/ecd7) | Your published evaluations | Setting, sample and period for each | If you sell through a framework, such as one run by [NHS Supply Chain](https://www.supplychain.nhs.uk/) or a Crown Commercial Service framework such as G-Cloud, name the framework and lot and link to the listing, so an assistant can make the connection. ## How do we show buyers, and assistants, that we meet DTAC and the DSP Toolkit? Put the evidence on one public assurance page, in plain text, dated and linked to the public record. Most suppliers keep it in a PDF sent after a call, where no assistant can see it. ### Step 1: List every standard on one page Give your DTAC form version and date, DSPT status and year, Cyber Essentials level, clinical safety officer, any MHRA device class, where data is hosted and which interoperability standards you support, as text rather than badges or downloads. ### Step 2: Link each claim to its public record Link your DSPT status to the organisation search, your certificate to IASME and your device to the MHRA database. A claim a buyer can check in one click gives an assistant a second source that agrees with you. ### Step 3: Date every line The DSPT is assessed each year and certificates expire. Put the date each item was confirmed beside it; an out-of-date line does more harm than a missing one. ### Step 4: Say what you share on request, and link the page everywhere Say that the DTAC form, a data protection impact assessment template and the clinical safety case exist and how to ask for them. Then link the page from every product page, case study and tender answer, using the same wording, so your website, framework listing and tenders never disagree. ## What should an occupational health provider publish to be named? State your coverage, accreditation and clinicians in plain text, because an assistant needs all three to answer "which providers cover multi-site employers across the UK". A provider whose site says only "nationwide service" gives it nothing to repeat. Accreditation sets a provider apart more than many assume. In a Department for Work and Pensions survey of 200 NHS, in-house and private providers, "only around half of the OH providers surveyed were SEQOHS accredited or working towards SEQOHS accreditation" ([DWP, August 2025](https://www.gov.uk/government/publications/understanding-occupational-health-provision-2023-24/summary-understanding-occupational-health-provision-2023-24)). SEQOHS is hosted by the Faculty of Occupational Medicine and has a [public search of accredited services](https://www.seqohs.org/). If you hold it, say so on every service page and keep your entry's name and locations identical to your website. For a multi-site employer, publish: - **Coverage.** Clinic addresses, mobile units, remote assessments and which regions your own clinicians serve. - **Turnaround.** Typical time from referral to appointment and to report, and how you measure it. - **Clinicians.** Named occupational physicians and nurses, with GMC or NMC registration and Faculty membership where held. - **Services by need.** Management referrals, health surveillance, fitness for work medicals and drug and alcohol testing, with the sectors you serve. Comment from your clinicians in Personnel Today, People Management or HR Magazine puts your name where assistants look for specialists. ## Which outside sources do assistants trust for healthcare suppliers? Independent sources carry more weight than anything you say about yourself. | Source | Examples | What to do | |---|---|---| | Registers | DSPT search, IASME, MHRA database, SEQOHS, CQC | Keep each entry current and in the same name as your website | | Procurement | NHS Supply Chain and Crown Commercial Service frameworks, contract notices | Link to your listings and name the lot | | Trade press | HSJ, Digital Health, Nursing Times, Pulse, Personnel Today | Comment on NHS policy and publish anonymised data | | Awards | HSJ Awards, HSJ Partnership Awards, Digital Health Awards | Enter the independently judged ones | | Customers | Case studies naming the trust or employer, joint conference talks | Get written permission and name the customer's lead | | Reviews | G2 and Capterra for software, Google reviews for services | Ask customers for specific reviews | Reviews matter more than most health tech firms assume: in G2's survey, 45% of software buyers said citations of review sites were the most confidence-inspiring signal in an AI answer ([G2, April 2026](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html)). Our guide to [whether awards and accreditations help you get recommended](/insights/awards-accreditations-ai-trust) explains which kinds count, and firms that also sell outside healthcare should read our guide for [B2B technology companies](/insights/ai-optimisation-technology-companies). ## What about patient-facing clinics? Clinics get named when an assistant can verify their clinicians, registration and details, and when their pages answer what patients ask before booking. - **Clinicians.** Give each clinician a page with their registration number (GMC, GDC, NMC, HCPC or GPhC), qualifications and any NHS post, in plain text. - **Rating and its date.** Show your CQC rating with its date and link to the report. Ratings can be old: the government's review of the CQC found "the average age of current ratings across all locations is 3.9 years" as of July 2024 ([DHSC, October 2024](https://www.gov.uk/government/publications/review-into-the-operational-effectiveness-of-the-care-quality-commission-full-report/review-into-the-operational-effectiveness-of-the-care-quality-commission-full-report)). When a patient asks which private clinics in Leeds have an Outstanding CQC rating, a live assistant leans on the CQC's own listings, so your location names must match your registration exactly. - **Cosmetic services.** In August 2025 the government said the highest-risk cosmetic procedures will become CQC regulated activities, with local authority licensing for lower-risk ones ([DHSC, August 2025](https://www.gov.uk/government/consultations/licensing-of-non-surgical-cosmetic-procedures)). Say which register your practitioners are on now, and watch for the licensing dates. - **Reviews.** Fake reviews and concealed incentivised reviews have been banned under the Digital Markets, Competition and Consumers Act 2024 since 6 April 2025 ([DMCC Act, Schedule 20](https://www.legislation.gov.uk/ukpga/2024/13/schedule/20)). The CMA's [fake reviews guidance](https://www.gov.uk/government/publications/fake-reviews-cma208) sets out what businesses must do, and the CMA can now decide for itself whether consumer law has been broken and sanction firms ([CMA, April 2025](https://www.gov.uk/government/publications/the-cmas-approach-to-direct-consumer-protection)). - **Advertising.** The ASA's remit has covered "marketing claims on companies own websites" since 2011 ([ASA, September 2026](https://www.asa.org.uk/media_channel/Online_catch_up_TV_and_radio_in_app_and_in_game.html)), and prescription-only medicines [may not be advertised to the public](https://www.legislation.gov.uk/uksi/2012/1916/regulation/284). Publish treatment pages reviewed by the treating clinician, with a dated reviewed-by line, and fees in plain text. None of this is legal advice. ## How do we tell whether it is working? Keep a fixed list of 20 to 50 buyer questions, run each several times a month in a clean session in ChatGPT, Google AI Overviews and AI Mode, Copilot, Gemini, Perplexity and Claude, and count a win only when it repeats, because answers vary between runs. Add "an AI assistant such as ChatGPT or Copilot" to the "how did you hear about us" options on your forms. Our guide to [checking whether ChatGPT recommends your firm](/insights/track-ai-recommending-your-business) has the full method. First, check the site can be read. The main AI crawlers fetch pages but do not run JavaScript ([Vercel and MERJ, December 2024](https://vercel.com/blog/the-rise-of-the-ai-crawler)), so product detail that appears only in a script, a PDF or behind a form is invisible to them. Cloudflare has also blocked AI crawlers by default on new domains since July 2025. Our [technical checklist for AI search](/insights/technical-checklist-ai-ready-website) lists what to ask your web team. ## What to do this month 1. Build a public assurance page listing your DTAC form version, DSPT status, Cyber Essentials level, clinical safety officer and any MHRA registration, each dated and linked to its public record. 2. Check that your DTAC evidence follows the form NHS England introduced on 24 February 2026. 3. Write down 20 to 50 questions your NHS, private sector and employer buyers ask, and run them in ChatGPT, Copilot and Google's AI Mode. 4. Rebuild your most important product or service page around one buyer question, naming the care setting and the standard in the first paragraph. 5. Make your name and status match across the DSPT search, IASME, SEQOHS or the MHRA database, framework listings and Google. 6. Ask two customers for written permission to name them in a case study. ## Frequently asked questions ### Is there such a thing as being "DTAC approved"? No. DTAC is the framework NHS commissioners and providers use to assess a digital health product, and NHS England expects manufacturers to keep an up-to-date form and supporting documents for each product version. There is no central approval or public register. Say that you hold a current form on the February 2026 version, give its date and explain how buyers can request it. ### Can we name NHS trusts as customers? Yes, with written permission, and it is one of the strongest signals you can publish. Many trusts have their own rules on supplier publicity, so agree the wording with the trust's communications team. A case study that names the trust, the problem, the result and the period it covers is far more useful than a logo wall. ### Which AI assistants matter most to healthcare buyers? ChatGPT and Google's AI Overviews and AI Mode first, then Microsoft Copilot and Gemini, then Perplexity and Claude. Copilot deserves extra attention because it is what many business buyers have at work, and Forrester found business buyers four times as likely as consumers to use it. ### If we rank well on Google, will ChatGPT name us? Ranking helps but does not guarantee a mention. Google's AI features, Copilot and ChatGPT search all start from a search index, but Ahrefs found only 38% of AI Overview citations came from top-ten pages in March 2026, down from 76% in July 2025, because Google now runs several related searches for one question ([Ahrefs, March 2026](https://ahrefs.com/blog/ai-overview-citations-top-10)). Covering the follow-up questions matters as much as ranking. ### How long does it take to be recommended? Assistants that search the web live, such as ChatGPT search, Perplexity, Copilot and Google's AI features, can reflect a new page or trade press article within weeks. Answers drawn from a model's own memory change only when the model is retrained, which takes months. Being named for broad questions usually takes several months of steady publishing and coverage. If your product is sound but the assistants cannot see the evidence, [Reputation That Wins](/reputation-that-wins) builds the outside record and [Website That Sells](/website-that-sells) puts the assurance on the page. A free [Growth Review](/growth-review) is 30 minutes with David Adams, with three fixes in writing afterwards. ## Sources - [NHS England, Digital Technology Assessment Criteria (DTAC) (February 2026)](https://transform.england.nhs.uk/key-tools-and-info/digital-technology-assessment-criteria-dtac/) - [Forrester, B2B buyers make zero-click buying number one (January 2026)](https://www.forrester.com/blogs/b2b_buyers_make_zero_click_buying_number_one/) - [G2, The Answer Economy: half of B2B software buyers now start their research with AI chatbots (April 2026)](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html) - [Department for Work and Pensions, Understanding Occupational Health Provision 2023-24 (August 2025)](https://www.gov.uk/government/publications/understanding-occupational-health-provision-2023-24/summary-understanding-occupational-health-provision-2023-24) - [Department of Health and Social Care, Review into the operational effectiveness of the Care Quality Commission (October 2024)](https://www.gov.uk/government/publications/review-into-the-operational-effectiveness-of-the-care-quality-commission-full-report/review-into-the-operational-effectiveness-of-the-care-quality-commission-full-report) - [Department of Health and Social Care, Licensing of non-surgical cosmetic procedures: government response (August 2025)](https://www.gov.uk/government/consultations/licensing-of-non-surgical-cosmetic-procedures) - [Competition and Markets Authority, Fake reviews guidance CMA208 (April 2025)](https://www.gov.uk/government/publications/fake-reviews-cma208) - [Ahrefs, AI Overview citations from top-ten pages (March 2026)](https://ahrefs.com/blog/ai-overview-citations-top-10) --- Title: How does an architecture practice get recommended by ChatGPT? URL: https://fortitudemedia.ai/insights/ai-optimisation-architecture-design Author: Ross Williams Published: 2026-03-31 Last reviewed: 2026-09-24 Topic: Sector guides Short answer: ChatGPT recommends practices it can read about, not look at. Name your sectors, write project pages that explain the brief, constraints, decisions and outcome, and link each architect's ARB registration number to their entry on the Architects Register. Keep the practice name and description identical across your site, RIBA's directory, the AJ100, Dezeen, Archello and planning records, because images alone tell an assistant almost nothing. A development director at a housing association asks Claude which architects in the North West have designed later living schemes. The estates lead for a multi-academy trust asks ChatGPT for practices that have delivered classroom blocks on occupied school sites. The chief operating officer of a professional firm moving offices asks Copilot for workplace architects in Leeds. Each assistant replies with a handful of names, and each of those names now has a head start on the shortlist. This is where business clients now start. Forrester found in January 2026 that 94% of business buyers use AI somewhere in their buying process, and that they are four times as likely as consumers to use Microsoft Copilot ([Forrester, January 2026](https://www.forrester.com/blogs/b2b_buyers_make_zero_click_buying_number_one/)). If your practice is not named, it is rarely because your work is weaker. It is usually because an assistant cannot see your work at all. Your best projects live in photographs, and assistants read words. The encouraging part is that architects already write a great deal. Design and access statements, heritage statements, planning reports and competition submissions are full of the reasoning an assistant needs, and almost none of it reaches the practice website. This guide is for the founder, director or partner of an architecture or interior design practice that works for developers, housing associations, school trusts, public bodies and commercial occupiers. ## Key takeaways - AI assistants cannot judge a building from photographs, so practices are recommended on what their written pages say. - Project pages that explain the brief, the constraints, the key decisions and the outcome are the strongest evidence a practice can publish. - Each architect's ARB registration number, linked to their entry on the online Architects Register, is a public record an assistant can check. - RIBA's Chartered Practice directory, the AJ100, Dezeen, Archello, planning records and awards should all describe the practice the same way. - Practices that lead with one or two sectors, such as later living or education, are named far more often than practices that say they design everything. - Clear pages on process questions, such as the two principal designer roles, bring your name into answers long before a client appoints anyone. ## What do business clients ask when they need an architect? The questions combine a building type, a client type, a place and a constraint. Typical questions look like this: - "Which architects in Manchester specialise in later living schemes for housing associations?" - "Which practices have delivered school extensions on occupied sites?" - "Workplace architects in Leeds for a Cat B office fit-out" - "Architects with experience of converting offices to residential in London" - "Is the principal designer under the Building Regulations the same as under CDM?" - "What percentage of build cost do architects charge for a commercial project?" - "Is [your practice] good to work with?" Some are finding questions: who should we appoint. Some are deciding questions: what will it cost, what does the process involve, what are our legal duties. The last is a checking question, which usually comes after a client has met you and wants a second opinion. A practice that only thinks about the first kind misses most of the moments when its name could come up. ## Why don't beautiful images get a practice recommended? Assistants recommend practices based on text, because text is what they can read and compare. A portfolio page with a project name, a location and twenty photographs tells ChatGPT almost nothing about what you did, why it was difficult or who it was for. Image descriptions, where they exist, usually say "exterior view". For example, picture a twelve-person practice in Bristol with a RIBA regional award for a timber school building, whose project page carries a title, a year and a gallery. A competitor two streets away has less striking buildings, but each project page explains the brief, the tight site, the planning history, the timber frame decision, the programme and what the head teacher said a year after handover. When a trust's estates lead asks for architects experienced in education work, the competitor has given the assistant something to work with. None of this means dropping the photography. It means putting words beside it. Captions should say what a picture shows and why it matters, such as "the clerestory windows bring daylight deep into the plan without overheating south-facing classrooms", and every project needs a written account that stands on its own. ## What does an AI assistant check before it names a practice? Four things: what you design and where, whether your credentials can be checked, what independent sources say about you, and whether everything agrees. ### Sectors, building types and region Say plainly which sectors you work in and where. "Architecture, interiors and masterplanning" cannot be matched to a housing association's question. "Later living and affordable housing for housing associations across the North West" can. ### Credentials that can be checked "Architect" is a protected title. Under section 20 of the Architects Act 1997, nobody may practise or carry on business under a name, style or title containing the word unless they are registered ([legislation.gov.uk](https://www.legislation.gov.uk/ukpga/1997/22/section/20)). ARB, the regulator, keeps the official list of the UK's 42,000+ architects and points out that anyone can provide architectural services; only the title is protected ([ARB](https://arb.org.uk/public-information/before-hiring-an-architect/who-can-use-the-title-architect/)). So do more than write "ARB registered". On each architect's team page, give their registration number and link it to their entry, in the form ARB itself suggests: "I am registered and regulated by ARB: architects-register.org.uk/architect/000000X" ([ARB](https://arb.org.uk/architect-information/title-protection-toolkit/6-simple-ways-to-help-protect-your-title/link-to-your-register-entry/)). Keep the register entry public and current, because it is the record an assistant can match to your name. Then add the memberships relevant to your work: RIBA Chartered Practice status, CIAT for architectural technologists, BIID for interior designers, the Landscape Institute, certified Passivhaus designers, and RIBA's specialist registers, such as its conservation register and its register of chartered members for the principal designer role. ### Independent records of your work These are the places an assistant finds your name beside a judgement of your work. List each one on your site and keep the practice name identical everywhere. | Source | What it shows | What to check | |---|---|---| | RIBA Find an Architect | Chartered Practices from a membership of over 4,000, listed by location with projects ([RIBA](https://www.riba.org/explore/find-an-architect/)) | Sectors, location and projects match your website | | The AJ100 | The Architects' Journal's annual ranking of UK practices | Your practice name matches other records | | Dezeen and Archello | Project pages with text, credits and images | The project name, client type and credits match your own project page | | Planning portals | Applications that usually name the agent | The agent name matches your trading name | | Awards and frameworks | RIBA awards, Civic Trust Awards, sector awards and public sector frameworks | Each award appears on the relevant project page, not only in the footer | ### Consistency Practices change names when partners retire, merge with others or trade under a name that differs from the registered company. If your website, Companies House, RIBA's directory, LinkedIn and old award listings describe you differently, an assistant struggles to connect the evidence. The same thing happened after a merger in another sector: at ID-Pal, "The redirect from northrow.com was technically correct, but nothing on either site told Google or the AI engines that NorthRow and ID-Pal were the same company." The [ID-Pal case study](/work/id-pal) shows how it was fixed. Pick one name and one description, say on your site what the practice used to be called, and update every listing you can find. ## How do you write a project page an assistant can use? A useful project page reads like a short, clear account of the job. Around 500 to 900 words for your most important projects is plenty. ### Step 1: State the brief in the client's terms Who the client was, what they needed and why: more classrooms without losing the playground, forty later living flats on a constrained town centre site, a headquarters that brings three offices under one roof. Name the client if they agree, and the sector and scale if they do not. ### Step 2: Describe the site and the constraints Planning policy, conservation area or listed status, neighbours, ground conditions, budget, programme, an occupied building. Constraints are where your expertise shows, and they are what a client with a similar problem is searching for. ### Step 3: Explain the key decisions Three or four decisions that shaped the building, and why you made them: the structural approach, the layout, the materials, how you handled a planning objection. Link each decision to the brief or a constraint. ### Step 4: Record how it was delivered Which RIBA Plan of Work stages you were appointed for, the procurement route, the contractor, the consultant team and any statutory roles you held, such as principal designer under the Building Regulations or under CDM 2015. ### Step 5: Say what happened next Planning outcome, completion, awards, what the client or users said after moving in, and measured performance if you have it. If you do not have numbers, describe the result plainly rather than inventing them. ## Is the principal designer under the Building Regulations the same as under CDM? No. They are two appointments under two sets of rules, and a client may need both. Since 1 October 2023, the Building Regulations in England have required dutyholders, including a principal designer, to plan, manage and monitor design work so that it complies with all relevant building regulations ([GOV.UK](https://www.gov.uk/government/publications/building-control-changes-for-higher-risk-buildings-and-wider-changes-to-building-regulations/circular-letter-changes-to-the-building-control-process)). The CDM 2015 principal designer plans, manages and coordinates health and safety before construction on any project with more than one contractor ([HSE](https://www.hse.gov.uk/construction/cdm/2015/principal-designers.htm)). | | Building Regulations principal designer | CDM 2015 principal designer | |---|---|---| | Concerned with | Compliance of the design with building regulations | Health and safety risks in the pre-construction phase | | In force since | 1 October 2023, for all building work in England | 6 April 2015 ([legislation.gov.uk](https://www.legislation.gov.uk/uksi/2015/51/regulation/1/made)) | | Set out in | Building Regulations 2010, regulations 11A to 11O, added in 2023 | Construction (Design and Management) Regulations 2015 | A short, dated page explaining both roles, and which ones your practice takes on, answers a question clients put to assistants before they appoint anyone. Contractors ask about the same rules, which our [guide for construction firms](/insights/ai-optimisation-construction-trades) covers from their side. ## How do you put planning and technical writing to work? Turn what you already write into plain-English guides for clients. Clients ask deciding questions long before they appoint anyone, and the practice that answers them clearly becomes a source the assistant quotes. The public parts of design and access statements show how you think about context; the notes you give clients at the start of a project explain process and fees; CPD talks explain technical subjects. A few examples that work well for business clients: - The RIBA Plan of Work for first-time commercial clients, stage by stage, with what they must decide at each. - What the Building Safety Act means for developers and housing associations commissioning residential buildings. - How long planning takes in the local authorities where you work most, and what slows it down. - How your fees are structured for the project types you do most. Our guide on [answering the questions your customers actually ask](/insights/content-around-customer-questions) explains how to choose these topics. ## Why does "we design everything" lose to a clear specialism? Assistants match specific questions to specific evidence, so a practice with a clear specialism is far easier to recommend. Most practices work across several sectors, and that is fine; lead with the one or two where you have the deepest record. A practice that is plainly "the later living architects in the North West" will be named for that question, and often for adjacent ones. Developers and their agents ask their own questions of assistants, which our guide for [commercial property firms](/insights/ai-optimisation-property-real-estate) covers. ## Why might your website be invisible to AI? Many practice websites are designed to be looked at, and common choices hide the content from machines: - Project pages that are galleries with a title and a few lines of text. - Brochures and bid documents available only as PDFs, or text set inside images. - Portfolio filters and project pages loaded by scripts. The main AI crawlers fetch pages but do not run JavaScript ([Vercel, December 2024](https://vercel.com/blog/the-rise-of-the-ai-crawler)). - Team pages with names and photographs but no roles, registrations or experience. - AI crawlers blocked in the site's settings without anyone knowing; [Cloudflare changed its default](https://blog.cloudflare.com/content-independence-day-no-ai-crawl-without-compensation/) to block AI crawlers in July 2025. Ask your web designer to confirm that crawlers can read every project and team page and that images carry descriptive alternative text. Structured data, the labels in the page code that state your practice's name, address and people, helps search engines read those facts correctly, but Google says its AI features need no special markup ([Google Search Central](https://developers.google.com/search/docs/appearance/ai-features)). Keep it accurate and treat it as housekeeping rather than a trick. ## How long before this changes what assistants say? Rewritten project pages and fixed listings can show within weeks in assistants that search the web live, such as ChatGPT search, Perplexity, Copilot and Google's AI features. Answers drawn from a model's own memory change only when it is retrained, which takes months. Track a fixed list of 20 to 50 questions your clients ask, each run several times a month in a clean session, and count a win only when it repeats. Most practices notice it first in better-briefed enquiries. ## What to do this month 1. Write one sentence that says which sectors you lead with and where, and use it on your homepage, LinkedIn and RIBA directory listing. 2. Add each architect's ARB registration number, linked to their register entry, plus roles and experience, to every team page. 3. Rewrite your three most important project pages using the five steps above, with descriptive captions. 4. Check your practice name and credits on RIBA's directory, Dezeen, Archello, award listings and planning records, and correct any that differ. 5. Publish one dated page on the two principal designer roles and which your practice takes on. ## Frequently asked questions ### Which architects in Manchester specialise in later living schemes for housing associations? An assistant answers this from practices whose pages, listings and awards say "later living", "housing association" and "Manchester" about completed schemes. A practice that wants to be in that answer needs a later living sector page, project pages for each scheme with the client type and location, and the same wording on its RIBA directory entry and in any trade press coverage. Awards and framework places for housing clients strengthen the match. ### What percentage of build cost do architects charge for a commercial project? We do not quote a market figure here, because fees depend on the building type, the scale, the RIBA stages in the appointment and the procurement route. Practices charge in three common ways: a percentage of construction cost, a fixed fee per stage or time charges. Clients ask assistants this before they call anyone, so explain how you charge and what changes the fee on your own site; an assistant can only quote your approach if it can find it. ### Is RIBA Chartered Practice status worth having? For a practice that wants business clients, usually yes, because it is a public, checkable standard. RIBA requires at least one full-time director or partner to be a RIBA Chartered Member, at least one in eight staff to be Chartered Members, professional indemnity insurance and policies covering employment, equality, health and safety, environmental management and quality management ([RIBA](https://www.riba.org/work/join-us/riba-chartered-practice-uk-membership/)). Chartered Practices appear in RIBA's directory, which gives an assistant another independent record of your sectors and location. ### How do we publish projects when clients want confidentiality? Agree what can be shared at appointment, not at completion. Many clients are happy with a written account once the building is complete, especially when it presents their project well. Where they are not, describe the sector, scale, location and constraints without naming the client or showing sensitive areas; a headquarters fit-out can be described without the tenant's name. ### Do RIBA awards still matter if clients ask ChatGPT? Yes, because they create independent public records that name your practice beside a project and a judgement of its quality. The citation, the shortlist and any press coverage are all text an assistant can read. Put every award on the relevant project page with a short explanation of what the judges recognised. ### Can a small practice compete with large ones? Often more easily than in a formal bid. A six-person practice that is clearly the specialist in listed commercial buildings in one city can be named ahead of a large practice whose relevant work is buried among hundreds of projects. Clear specialism, well-written project pages and checkable credentials matter more than size. If you want to see how your practice compares today, [book a free Growth Review](/growth-review): 30 minutes with David Adams, and three fixes in writing afterwards. If the project pages themselves need rebuilding, [Website That Sells](/website-that-sells) is the service for that. ## Sources - [Forrester, B2B buyers make zero-click buying number one (January 2026)](https://www.forrester.com/blogs/b2b_buyers_make_zero_click_buying_number_one/) - [ARB, Who can use the title architect (September 2026)](https://arb.org.uk/public-information/before-hiring-an-architect/who-can-use-the-title-architect/) - [RIBA, Find an architect (September 2026)](https://www.riba.org/explore/find-an-architect/) - [RIBA, Chartered Practice UK membership (September 2026)](https://www.riba.org/work/join-us/riba-chartered-practice-uk-membership/) - [GOV.UK, Circular letter: changes to the building control process (March 2024)](https://www.gov.uk/government/publications/building-control-changes-for-higher-risk-buildings-and-wider-changes-to-building-regulations/circular-letter-changes-to-the-building-control-process) - [HSE, CDM 2015: principal designers (September 2026)](https://www.hse.gov.uk/construction/cdm/2015/principal-designers.htm) - [Vercel, The rise of the AI crawler (December 2024)](https://vercel.com/blog/the-rise-of-the-ai-crawler) - [Google Search Central, AI features and your website (December 2025)](https://developers.google.com/search/docs/appearance/ai-features) --- Title: How does our agency get named when clients ask ChatGPT for one? URL: https://fortitudemedia.ai/insights/ai-optimisation-marketing-creative-agencies Author: Ross Williams Published: 2026-03-31 Last reviewed: 2026-09-24 Topic: Sector guides Short answer: ChatGPT names agencies it can match to a specific question, so lead with one clear specialism by sector, discipline or problem. Publish case studies in plain text that show the client's problem and approved results, and keep Clutch, the IPA Agency Finder, PRCA or CIPR listings and your trade press name consistent. PR agencies are judged most on earned coverage. Say plainly who you are for and how you price, because buyers ask assistants both. Most agency founders built their business on referrals, pitches and the odd award. That route is narrowing. More of the people who hire agencies now start by asking ChatGPT, Copilot, Gemini or Perplexity for a shortlist: "Which B2B branding agencies in Manchester work with tech companies?" The assistant names three or four. If you are not among them, you are not on the list, and you rarely learn that you were ever considered. Forrester found in January 2026 that 94% of business buyers use AI somewhere in their buying process, and that business buyers are four times as likely as consumers to use Microsoft Copilot ([Forrester, January 2026](https://www.forrester.com/blogs/b2b_buyers_make_zero_click_buying_number_one/)). The marketing director choosing an agency is one of those buyers, and often asks a second question in the same conversation: whether they need an agency at all. This is uncomfortable territory for an industry that sells visibility. Plenty of excellent agencies are close to invisible in these answers, because the habits that win awards and pitches, such as a portfolio built from images and a homepage line like "we make brave work", give a machine almost nothing to read. This guide is for the founder of a marketing, PR, creative or digital agency who wants to fix that without turning the agency into a content mill. ## Key takeaways - AI assistants name agencies with a clear specialism they can match to the question, so "full-service" positioning is rarely recommended. - Clutch ranks agencies partly on how specialised they are in the service being mapped, so specialism helps in the directories as well as in the answers. - Directories and bodies such as Clutch, the IPA Agency Finder, the PRCA, the CIPR and the PRWeek UK Top 150 give assistants independent proof that you exist and what you do. - For PR agencies, earned coverage is the strongest signal: Muck Rack found earned media made up 84% of the links AI tools cited in 2026. - Case studies written in plain text around a client's business problem and approved results are the strongest evidence an agency can publish. - Publishing how you price, who you are for and what you measure answers the questions buyers put to assistants before they call. ## What do marketing directors ask when they need an agency? They combine a discipline, a sector, a location and a worry. Typical questions look like this: - "Which B2B marketing agencies in Manchester specialise in SaaS?" - "Which PR agencies are good at fintech launches?" - "Best branding agency in London for a professional services firm" - "What does a B2B marketing agency cost per month in the UK?" - "Should a founder-led B2B firm hire a specialist agency or a full-service one?" - "Which UK agencies specialise in getting brands recommended by ChatGPT?" - "Is [your agency name] good to work with?" Google's UK search suggestions in September 2026 show the same worries in shorter form: people type "how much do marketing agencies charge uk", "how to choose a b2b digital marketing agency" and "marketing agency red flags". Almost none of these questions are generic. A buyer with a budget describes their problem: their sector, their channel, their stage. And the last question is about you by name. Prospects who meet you at a pitch increasingly check you with an assistant before the second meeting, and what it tells them depends on what it can find. ## Why do full-service agencies rarely get named? An assistant recommends an agency by matching a specific question to specific evidence, and "full-service" gives it nothing to match. If your site says you do brand, web, social, content, paid and PR for "ambitious businesses", there is no reason for ChatGPT to put you forward for fintech PR over an agency that says it does exactly that. The directories reward the same thing. Clutch's Leaders Matrix plots agencies on two axes, and one of them, Focus, measures "how specialized the company is in the service being mapped" ([Clutch methodology](https://clutch.co/methodology)). Specialism can run along several lines, and you only need one to be clear: a sector (brand and web for engineering firms), a discipline (employer branding, paid social), a problem (rebrands after a merger) or a client type (founder-led firms, private equity portfolio companies). For example, picture a 25-person agency in Leeds whose site lists eleven services and forty clients across every sector, while its best work and most of its revenue come from brand and website projects for engineering and manufacturing businesses. If it builds its site, case studies and profiles around that, it becomes the obvious answer to "branding agency for an engineering company in the North". It can still take other work. It stops hiding the thing it is best at. ## Which directories and awards feed "best agency for" answers? The ones that describe agencies in structured, checkable detail. When a buyer asks for the best agency for something, assistants that search the web often draw on listings and review sites as well as agency websites. G2's April 2026 survey of 1,076 software buyers found 45% said citations of review sites were the most confidence-inspiring signal in an AI answer ([G2, April 2026](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html)). That survey covers software buyers; in our view Clutch plays the same role for services. | Source | What it shows | What to keep current | |---|---|---| | Clutch | Verified client reviews, the Leaders Matrix and award lists: Clutch Global, Clutch Champions, the Clutch 100 and the Clutch 1000 | Service focus, location and recent reviews | | The Manifest, Sortlist and DesignRush | Agency listings that buyers use to build shortlists | One name, one description and the same specialism everywhere | | IPA Agency Finder | Member agencies of the Institute of Practitioners in Advertising ([IPA](https://ipa.co.uk/agency-finder/)) | Membership and location | | Partner directories | Google, Meta, HubSpot and LinkedIn partner listings | Partner tier and services | | Trade press | Campaign, The Drum, Marketing Week and PRWeek | Your agency name spelt the same way in every byline and mention | Which of these an assistant cites varies by question and by run, so check for yourself. Perplexity, ChatGPT search and Google's AI Overviews show their sources; ask your own buyer questions and note which directories appear. Then fix those listings first. Clutch describes how it checks reviews: the reviewer's identity is authenticated through LinkedIn, Google or a company email, its software checks the reviewer's digital footprint, and an editor checks every submission ([Clutch methodology](https://clutch.co/methodology)). An abandoned profile with an old name and two reviews from years ago works against you. On awards, our view is that independently judged schemes with published shortlists, such as D&AD or the IPA Effectiveness Awards, create records worth having, while awards you pay to enter with little visible judging add much less. ## How do PR agencies get named? Through the same things they sell: earned coverage, named spokespeople and the industry's own rankings. Muck Rack's May 2026 study of more than 25 million links found earned media made up 84% of the links AI tools cited, journalism 27% and paid content 0.3%, and that more than half of cited journalism had been published in the past twelve months ([Muck Rack, May 2026](https://www.globenewswire.com/news-release/2026/05/07/3290268/0/en/generative-pulse-earned-media-consistently-drives-ai-citations-holding-at-84.html)). A PR agency's own coverage record matters as much as its clients'. For a prompt such as "PR agency for fintech", the sources that confirm a PR firm are: - **Membership of the PRCA or the CIPR**, shown on your site and matching your listing with them. - **The PRWeek UK Top 150**, the annual ranking of UK PR consultancies. If you appear in it, make sure the name there matches your website. - **Coverage you won for clients**, turned into case studies that name the titles, with the client's approval. - **Your founders' own bylines and comment** in the fintech, legal or technology trade press your clients read, not only in PR titles. Our guide to [earned, owned and paid signals](/insights/earned-owned-paid-authority-signals) explains why coverage outweighs anything you pay for. ## What else do assistants check before naming an agency? Your website, what clients say and whether it all agrees. A crawler reads words, so a page that says "brand strategy and identity for engineering and manufacturing firms, from positioning workshop to website launch" does far more than a full-screen video with the word "Brand" across it. A client's marketing director writing about "our rebrand with [agency]" on their company blog is a strong independent signal, and one most agencies never ask for. Agencies rebrand, merge, add service lines and lose founders more often than most businesses, which leaves old names and retired services on directories for years. If Clutch says you are a social media agency, your site says brand consultancy and LinkedIn still shows the name you dropped three years ago, an assistant struggles to describe you, so it describes someone else. ## How do you write case studies that assistants can use? Explain a business problem, what you did and what changed, in text, with enough detail to be believed. Most agency case studies are a hero image, a three-word headline and a gallery. ### Step 1: State the client's problem in their words Open with the situation and why it mattered to their business: a new market, a launch that stalled, a brand that no longer matched the company. Name the sector and size even if you cannot name the client. ### Step 2: Explain what you found Say what your research, workshops or audit showed. This is where your thinking becomes visible, and it is the part competitors cannot copy. ### Step 3: Describe what you did Strategy first, then execution: the positioning, the creative idea, the channels, the timeline. Keep the visuals, but make sure the text stands on its own. ### Step 4: Show what changed Use results the client has approved for publication, with a date. If numbers are confidential, describe the change plainly and say how it was judged. Never round up to make a better story. ### Step 5: Let the client speak A named quote from the client, with their job title, is worth more than any adjective you use about yourself. Five case studies like this in your specialism will do more for you than thirty galleries spread across every sector. ## How should an agency answer "do we still need an agency?" Answer it plainly on your own site, specific to your discipline. Say which tasks you now do faster with AI tools, which you do not trust them with, and how that shows up in your pricing and timelines. In our experience AI tools have made routine production cheaper and faster, while strategy, judgement and accountability for results still need people who have done it before. We set out the wider argument in [should you still pay a marketing agency now that AI can do the work](/insights/ai-disrupting-marketing-agency-model). ## What does this look like for a real agency? We apply the same rules to ourselves, so here is Fortitude as the worked example. We state one specialism in one line: we are the growth team for founder-led B2B firms with £1m to £20m+ revenue, where the founder still decides. Each of our three services has its own page with a plain name: [Website That Sells](/website-that-sells), [Reputation That Wins](/reputation-that-wins) and [Pipeline That Fills](/pipeline-that-fills). Our prices sit on one public page, [what it costs](/what-it-costs), so an assistant asked what we charge has an answer to find. Our case studies use figures the client has approved. Comparison pages are part of the same approach. For Helm Club, a founder peer group, "We built the structured data AI engines read, then a comparison page for every question a founder asks when weighing up peer groups: Helm against YPO, Vistage and the rest." The result: "1 in 5 new members this year came through an AI recommendation." The [Helm Club case study](/work/helm-club) has the detail. An agency can build the same pages for the comparisons its buyers make: specialist against full-service, agency against in-house, retainer against project. And we measure it the way we report it to clients: "The same buyer questions run every month in ChatGPT, Gemini and Claude, in a fresh chat. A win only counts when it repeats, and losses are reported too." The [ID-Pal case study](/work/id-pal) shows that record month by month. ## Why might your agency website be working against you? Many agency websites are built to impress other creatives rather than to be read. The usual problems: portfolios of images and video with a single line of text each; headlines such as "Brave ideas, beautifully made" that give a machine nothing to match; single-page sites where content loads only as someone scrolls; services described in the agency's internal language; and no sector pages, team pages or dates on thinking pieces. Keep the craft in the design and the substance in the text. A simple test: read your homepage with the images switched off and ask whether a stranger could say what you specialise in. ## How long does it take, and how do you tell it is working? Fixes to positioning, profiles and case studies can show within weeks in assistants that search the web live, such as ChatGPT search, Perplexity, Copilot and Google's AI features. Answers drawn from a model's own memory change only when it is retrained, which takes months. Track a fixed list of 20 to 50 buyer questions, each run several times a month in a clean session, and count a win only when it repeats. Most agencies notice it first in the quality of inbound enquiries. ## What to do this month 1. Write one sentence that says who you are best for and what you do for them, and use it on your homepage, LinkedIn and every directory profile. 2. Update or claim your listings on Clutch, the IPA Agency Finder or your PR body, and the other directories that appear when you check your buyer questions, removing old names and retired services. 3. Rewrite your three strongest case studies using the five steps above, with client-approved, dated results. 4. Publish how you price: your model, what changes the price and, if you can, your starting points. 5. Publish one plain piece on how AI tools have changed your work, your timelines and your pricing. ## Frequently asked questions ### What does a B2B marketing agency cost per month in the UK? We do not quote market ranges in this guide, because monthly fees depend on scope, seniority and whether the work is a retainer or a project. Ask any agency you shortlist to show its pricing in writing before the first call, and compare what each fee includes. Our own prices are on our [what it costs](/what-it-costs) page, and our guide on [what AI search optimisation costs in the UK](/insights/ai-search-optimisation-cost-uk) covers that service specifically. ### Which UK agencies specialise in getting brands recommended by ChatGPT? Several SEO, PR and specialist firms, including ours, now offer it under names such as AI search or AI SEO. Judge them on method rather than claims: a fixed list of buyer questions, a dated record of results and reports that show losses as well as wins. Nobody can guarantee what an AI assistant will say, so be wary of anyone who promises a place in ChatGPT's answers; a provider can guarantee effort, reporting and outcomes it controls. Our guide on [choosing a firm to help you get recommended by ChatGPT](/insights/how-to-compare-ai-optimisation-providers) sets out the questions to ask. ### How do I check an agency's results before hiring them? Ask for case studies with dated figures the client has approved, then ask to speak to that client. Check whether the client mentions the agency on its own website or LinkedIn, and read the agency's verified reviews on Clutch. For AI search work specifically, ask to see the question list and the month-by-month record behind any claim. ### Is Clutch a reliable way to choose a marketing agency? It is a useful filter rather than a verdict. Clutch checks reviewer identity and has an editor check every review, and its rankings weigh verified evidence and how focused an agency is. But agencies choose which clients to invite, so reviews are a sample of the happiest ones. Use Clutch to build a shortlist, then check case studies and speak to clients. ### Should a founder-led B2B firm hire a specialist agency or a full-service one? In our view, usually a specialist for the problem that matters most, because the people doing the work will know your buyers and your channel. A full-service agency can suit a firm that needs many things done at a basic level and has no one to coordinate several suppliers. Whichever you choose, one person on your side should own the result. ### Should an agency publish its prices? It helps more than most agencies expect. Buyers ask assistants what agencies charge, and an assistant can only give your figures if it can find them. Typical project ranges, a starting point for monthly retainers and an explanation of what changes the price are enough, and prospects arrive with realistic budgets. If your own presence keeps slipping behind client work, [Reputation That Wins](/reputation-that-wins) runs it alongside your team, and our [page for agencies and consultancies](/for/agencies-and-consultancies) shows how we work with firms like yours. ## Sources - [Forrester, B2B buyers make zero-click buying number one (January 2026)](https://www.forrester.com/blogs/b2b_buyers_make_zero_click_buying_number_one/) - [Clutch, Methodology (September 2026)](https://clutch.co/methodology) - [G2, The Answer Economy: half of B2B software buyers now start with AI chatbots (April 2026)](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html) - [Muck Rack, Generative Pulse: earned media consistently drives AI citations (May 2026)](https://www.globenewswire.com/news-release/2026/05/07/3290268/0/en/generative-pulse-earned-media-consistently-drives-ai-citations-holding-at-84.html) - [IPA, Agency Finder (September 2026)](https://ipa.co.uk/agency-finder/) --- Title: How does our training company get recommended when HR asks ChatGPT? URL: https://fortitudemedia.ai/insights/ai-optimisation-education-training Author: Ashley Kenerson Published: 2026-03-31 Last reviewed: 2026-09-24 Topic: Sector guides Short answer: ChatGPT recommends training providers it can match to a specific need and check. Make each programme page clear on topic, audience, format, price and accreditation, such as ILM or CMI centre status, LPI accreditation or a named CPD accreditor. Add named trainers, a published method for measuring results and client case studies in plain text. Keep listings consistent, and publish a dated explainer on what the apprenticeship levy now pays for, because HR buyers ask that too. An operations director at a 300-person logistics firm needs to develop twenty new team leaders before peak season. She asks ChatGPT: "Which ILM-accredited leadership programmes can be delivered on-site in the Midlands?" A sales director at a software company asks Copilot for B2B sales training providers who can work with a forty-person team. An L&D manager asks Claude how to prove the return on a leadership programme to the board. Each assistant names three or four providers, sometimes with a line on format or price. If your company is not one of them, you are not in the quote round, and you may never know the need existed. Budgets are tighter than they were: in the government's Employer Skills Survey, 59% of UK employers had funded or arranged training in the previous 12 months, and total training spend fell 10.2% in real terms between 2022 and 2024 ([Department for Education, July 2025](https://explore-education-statistics.service.gov.uk/find-statistics/employer-skills-survey/2024)). Buyers with less to spend shortlist harder, and many now shortlist with an assistant: Forrester found that 94% of business buyers use AI in their buying process ([Forrester, January 2026](https://www.forrester.com/blogs/b2b_buyers_make_zero_click_buying_number_one/)). The providers that get named are the ones whose programmes, approvals and results an assistant can read and check, whatever their size. This guide is for the founder or managing director of a corporate training company selling to HR and L&D, whether in leadership, management, sales or professional skills, who wants to be on that shortlist. ## Key takeaways - AI assistants name training providers whose programme pages state the topic, audience, level, format, location, price and accreditation in plain text. - ILM and CMI centre status, LPI accreditation and a named CPD accreditor give assistants independent records to check, but CPD accreditation is not statutory and several private bodies offer it. - The apprenticeship levy now pays for apprenticeship units, short courses of 1 to 16 weeks, while Level 7 apprenticeships are funded only for apprentices under 22 from 1 January 2026, and HR buyers ask assistants what the levy covers. - Ofsted has graded further education and skills providers with report cards since 10 November 2025, so apprenticeship providers should quote the new grades exactly. - Large catalogues of near-identical course pages, with prices and dates hidden in booking widgets, are one of the main reasons providers go unnamed. - A published method for measuring what changed at work sets a provider apart, because proving return is one of the questions L&D buyers ask most. ## What do HR and L&D managers ask ChatGPT? The questions combine a topic, an audience, a format and a constraint, usually time, budget or proof. These are typical of the ones that decide who gets named. | Kind of question | Example | What the answer leans on | |---|---|---| | Finding | "Which leadership development companies in the UK are ILM or CMI approved centres?" | Centre finders, programme pages, reviews | | Finding | "Sales training provider for a 40-person B2B software team, delivered in-house" | Programme pages, named trainers, case studies | | Deciding | "How much does in-house sales training cost per day in the UK?" | Providers that publish prices and what is included | | Deciding | "What can the apprenticeship levy be spent on in 2026?" | GOV.UK funding rules, dated provider explainers | | Deciding | "How do I prove the return on a leadership programme to my board?" | Evaluation guides written by practitioners | | Checking | "Is [your company] any good?" | Reviews, trade press, client names, Ofsted report cards | Most providers write only for the finding questions, yet the deciding questions are where an L&D buyer forms a view of who knows the subject. ## What does an AI assistant check before it names a training provider? Broadly, five things: what your programmes are, who approves them, who delivers them, what clients say, and whether all of it agrees. ### Programme pages it can read Each page should state, in text, the topic, level, who it is for, what people will be able to do afterwards, duration, format (on-site, virtual, blended), locations, group size, whether it can be tailored and the price. If a buyer has to phone to find out whether a programme is virtual, or whether it costs a few hundred pounds or a few thousand, an assistant cannot tell either. ### Accreditation and approvals Say which awarding organisation or professional body stands behind each programme and link to its public record. ILM qualifications are part of City & Guilds, and ILM runs a public [centre finder](https://www.i-l-m.com/); the Chartered Management Institute has its own [Find a Centre](https://www.managers.org.uk/) search. When a buyer asks which leadership companies are ILM or CMI approved centres, those lists are what an assistant can check, so make sure your entry uses the same name as your website. The Learning and Performance Institute offers [accreditation for learning providers](https://www.thelpi.org/). Where programmes are CPD-accredited, name the body: CPD accreditation is not statutory, and it is offered by several private organisations, such as the [CPD Certification Service](https://www.cpduk.co.uk/), which describes itself as independent. Say whether a qualification is regulated by Ofqual. ### Trainers Name the people who will be in the room, with their background and specialisms. A facilitator who spent fifteen years leading software sales teams is a reason to choose a sales programme; anonymous "expert facilitators" give an assistant nothing. ### Reviews and directories Reviews on Google and Trustpilot, and listings on course directories such as findcourses.co.uk and Reed Courses, feed answers about whether a provider is any good. Ask the HR or L&D lead who commissioned the programme for a review as well as the delegates, because the buyer asking the assistant is usually the commissioner. ### Consistency Directories keep retired courses and old prices for years, and an assistant may repeat them or hedge. We explain how to find and fix these conflicts in [why ChatGPT describes your firm wrongly](/insights/ai-conflicting-information-online). ## What can the apprenticeship levy be spent on in 2026? Apprenticeships, foundation apprenticeships and, since 28 April 2026, apprenticeship units, with tighter rules on Level 7. The levy is paid by employers whose annual pay bill is over £3 million, and the government is turning it into a growth and skills levy that it says will deliver "greater flexibility to employers, more opportunities for young people and support the industrial strategy" ([Department for Education, 2026 to 2027 funding rules](https://www.gov.uk/government/publications/apprenticeship-funding-rules-and-assessment-plan-guidance-2026-to-2027)). | Route | What it is | What applies now | |---|---|---| | Apprenticeship units | "A targeted flexible training programme designed to upskill employees in a critical skill", lasting 1 to 16 weeks and 30 to 140 delivery hours | Funded for units starting on or after 28 April 2026, delivered only by providers approved for each unit ([Department for Education, March 2026](https://www.gov.uk/government/publications/apprenticeship-unit-funding-rules-2026-to-2027/apprenticeship-unit-funding-rules-april-2026-to-31-july-2026)) | | Level 7 apprenticeships | Master's-level programmes, including senior leader apprenticeships | From 1 January 2026, funded only for apprentices under 22, or under 25 with an education, health and care plan or experience of local authority care ([Department for Education, January 2026](https://assets.publishing.service.gov.uk/media/69384d3b7a605b2d61cd8fc4/SOC_2526__Version_3.pdf)) | | Foundation apprenticeships | A new entry route | Added to the funding rules in 2025 to 2026, with rules of their own | | Other apprenticeships | Levels 2 to 6 | The practical period must last at least 8 months | For a corporate training provider, apprenticeship units are the change to write about: short, focused and paid for from the levy, but deliverable only by providers on the Apprenticeship Provider and Assessment Register approved for that unit. If you are approved, say which units and link to the Skills England listing. If you sold senior leader programmes to experienced managers through the levy, say plainly what has changed and what you offer instead. A dated explainer that links to the GOV.UK rules is the page assistants quote. ## How should apprenticeship providers quote their Ofsted grade? Exactly as the report card shows it, with the date and a link. Ofsted's report cards took effect from 10 November 2025 for further education and skills inspections, replacing single-word judgements with a five-point scale: exceptional, strong standard, expected standard, needs attention and urgent improvement ([Ofsted, September 2025](https://www.gov.uk/government/news/ofsted-confirms-changes-to-education-inspection-and-unveils-new-look-report-cards)). Do not turn "strong standard" into "rated excellent". If your latest report predates November 2025, it uses the old judgements, so give its date and say which framework it was graded under. Buyers and assistants both check the report, and a mismatch reads as spin. ## Why do big course catalogues fail with AI? Because 150 near-identical pages give an assistant little to tell one course from the next. The usual pattern is one template with the course name swapped, generic text, and dates and prices inside a booking widget that crawlers (the programs AI companies use to read web pages) often cannot see into. For example, a provider with 150 courses may earn most of its revenue from eight of them: leadership, first-line management, sales and a handful of professional skills programmes. If those eight pages are thorough and the rest are merged or retired over time, the provider becomes much easier to match to a buyer's question. The same problem appears outside training. At TCC Group, a regulatory compliance group for UK financial services, part of what failed was "pages that half-answered the same question so neither was cited", and the plan rested on "624 pages reviewed, each given a decision: migrate, merge or retire" ([TCC Group case study](/work/tcc-group)). ## How do we build a programme page an assistant can use? Answer every question a buyer would otherwise ask on the phone. ### Step 1: Say who it is for Role, experience and sector, such as first-time managers in professional services. Say who it is not for as well. ### Step 2: Say what they will be able to do Three to six outcomes written as things people can do afterwards, such as "run a return-to-work conversation" or "qualify a deal against agreed criteria", not topics covered. ### Step 3: Explain how it is delivered Format, duration, locations, group size and whether it can be run in-house or tailored. ### Step 4: Name the accreditation and assessment The awarding organisation or accrediting body, the qualification level where relevant, how learners are assessed and what certificate they receive. ### Step 5: Publish the price Per-delegate prices for open courses, a typical range for in-house delivery for a stated group size, and what is included: materials, registration fees, certification, travel. ### Step 6: Show the evidence A named trainer, a short client case study, recent reviews and any completion or pass figures you can stand behind, with how they were measured. ## How do I prove the return on a leadership programme to my board? Agree the measures with the sponsor before the programme starts, take a baseline, and follow up at work months later, not only on the last day. L&D buyers ask assistants this question often, and a provider that publishes a clear, practical method becomes the source the assistant draws on. - **Before.** Agree two or three business measures the programme should move, such as retention in the managers' teams or time to productivity for new starters, and record where they stand. - **During.** Capture reaction and learning, the first two levels of the Kirkpatrick model, with specific questions rather than a single score. - **After.** Return at three and six months to check behaviour and results, the model's third and fourth levels, with the sponsor and the participants' own managers. - **Report.** Publish anonymised results with the period, the sample and how they were measured. Only publish figures you can support. If you gather feedback across hundreds of programmes a year, the patterns can become research in their own right, which we explore in [whether publishing your own research helps ChatGPT recommend you](/insights/original-research-data-ai-trust). ## Which outside sources do assistants trust for training providers? Independent sources carry more weight than your own claims. For a corporate learning firm, these matter most. | Source | Examples | What to do | |---|---|---| | Trade press | CIPD's People Management, Training Journal, HR Magazine, Personnel Today | Offer comment and practical articles by named trainers | | Events | The Learning Technologies show, CIPD events | Speak with a client, not alone | | Accreditation and approvals | LPI accreditation, ILM and CMI centre finders, your CPD accreditor | Keep each entry current and in the same name | | Awards | The LPI's Learning Awards, Training Journal's awards | Enter the independently judged ones | | Reviews | Google, Trustpilot, course directories | Ask commissioning clients as well as delegates | Reviews count in AI answers: in G2's April 2026 survey of software buyers, 45% said citations of review sites were the most confidence-inspiring signal in an AI answer ([G2, April 2026](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html)). Training firms sit close to consultancies, and our guide for [management consultancies](/insights/ai-optimisation-management-consultancies) covers the same shortlisting problem from their side. Our guide to [awards and accreditations](/insights/awards-accreditations-ai-trust) explains which kinds count. ## How do we tell whether it is working? Keep a fixed list of 20 to 50 questions covering your main programmes, sectors and regions, run each several times a month in a clean session in ChatGPT, Google's AI Overviews and AI Mode, Copilot, Gemini, Perplexity and Claude, and count a win only when it repeats, because answers vary between runs. Record whether you are named, for which programmes, how you are described and which sources are cited. Add "AI assistant such as ChatGPT or Copilot" to the options on your enquiry form. Our guide to [checking whether ChatGPT recommends your firm](/insights/track-ai-recommending-your-business) has the full method. ## What to do this month 1. Pick the five programmes that matter most to your revenue and rebuild each page using the six steps above. 2. Put prices, formats and dates for those programmes in plain text, not only inside a booking widget. 3. Check your ILM, CMI, LPI and CPD listings match your website, and remove retired courses and old prices from course directories. 4. Publish a dated explainer on what the levy now pays for, including apprenticeship units and the Level 7 change, linked to the GOV.UK funding rules. 5. Write up your method for measuring what a leadership programme changed at work, with one anonymised example. 6. Add named trainer profiles and link each one to the programmes they deliver. ## Frequently asked questions ### Should we publish prices for in-house training? Yes, as ranges. Buyers ask assistants what in-house training costs per day, and a provider that publishes a typical day rate for a stated group size, with what is included, can be quoted, while one that says "contact us" cannot. We do not quote market rates here, because they vary with tailoring, trainer seniority, travel and certification fees. Publish your own and explain what changes the price. ### Does CPD accreditation make a difference to AI assistants? It helps as one signal among several. Buyers often include it in their questions, such as "CPD-accredited" or "ILM-accredited", and it gives an assistant an outside record. But CPD accreditation is not statutory and several private bodies offer it, so always name the body. It will not make up for a thin programme page. ### We win most work through tenders and frameworks. Does this matter to us? Yes. Procurement teams and the managers who write specifications research providers before and during a tender, and assistants are part of that research. Clear programme pages, approvals and case studies support your bid when an evaluator checks your claims, and help you win the direct work that sits outside frameworks. ### If we rank well on Google, will ChatGPT name us? Ranking helps but does not guarantee a mention. Ahrefs found that only 38% of AI Overview citations came from pages in the top ten in March 2026, down from 76% in July 2025, because Google now runs several related searches for one question ([Ahrefs, March 2026](https://ahrefs.com/blog/ai-overview-citations-top-10)). Pages that answer the follow-up questions, such as cost and proof of return, matter as much as ranking. ### How long does it take to be recommended? Assistants that search the web live, such as ChatGPT search, Perplexity, Copilot and Google's AI features, can reflect a rebuilt page or a corrected listing within weeks. Answers drawn from a model's own memory change only when the model is retrained, which takes months. Being named for broad questions usually takes several months of steady publishing, reviews and outside coverage. If you would rather spend more time talking to the HR and L&D leaders who buy your programmes, [Pipeline That Fills](/pipeline-that-fills) arranges those conversations through researched outbound, with every message written and approved by a person, and it works alongside everything in this guide. A free [Growth Review](/growth-review) is a good place to start. ## Sources - [Department for Education, Employer Skills Survey 2024 (July 2025)](https://explore-education-statistics.service.gov.uk/find-statistics/employer-skills-survey/2024) - [Forrester, B2B buyers make zero-click buying number one (January 2026)](https://www.forrester.com/blogs/b2b_buyers_make_zero_click_buying_number_one/) - [Department for Education, Apprenticeship funding rules 2025 to 2026: summary of changes (January 2026)](https://assets.publishing.service.gov.uk/media/69384d3b7a605b2d61cd8fc4/SOC_2526__Version_3.pdf) - [Department for Education, Apprenticeship funding rules: 2026 to 2027 (August 2026)](https://www.gov.uk/government/publications/apprenticeship-funding-rules-and-assessment-plan-guidance-2026-to-2027) - [Department for Education, Apprenticeship unit funding rules: April 2026 to July 2026 (March 2026)](https://www.gov.uk/government/publications/apprenticeship-unit-funding-rules-2026-to-2027/apprenticeship-unit-funding-rules-april-2026-to-31-july-2026) - [Ofsted, Ofsted confirms changes to education inspection and unveils new-look report cards (September 2025)](https://www.gov.uk/government/news/ofsted-confirms-changes-to-education-inspection-and-unveils-new-look-report-cards) - [G2, The Answer Economy: half of B2B software buyers now start their research with AI chatbots (April 2026)](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html) - [Ahrefs, AI Overview citations from top-ten pages (March 2026)](https://ahrefs.com/blog/ai-overview-citations-top-10) --- Title: How long does PR take to work, and what should a year look like? URL: https://fortitudemedia.ai/insights/compounding-effect-consistent-pr-12-months Author: Ross Williams Published: 2026-03-31 Last reviewed: 2026-09-24 Topic: Press, reviews and being trusted Short answer: In our experience, plan on a year for the full effect. The first quarter builds relationships and material, with a few quotes but little change in enquiries. Journalists start coming back in months four to six, and enquiries and AI answers improve in the second half. Assistants that search the web live can reflect new coverage within weeks, but answers from a model's memory change only when it is retrained. Keep the effort steady: in Muck Rack's May 2026 study, AI tools cited recent journalism far more than older pieces. Three months ago you agreed to take PR seriously. Since then your founder has been quoted twice in the trade press, one article has been accepted and there have been a lot of emails. No new enquiries mention any of it. At the next board meeting someone asks, reasonably, whether it is working. This is the point at which most firms stop, and it is usually just before the work starts to pay. PR does not add up in a straight line. Each quote makes the next one easier, each article gives journalists and buyers a reason to look you up, and each mention adds to the evidence AI assistants such as ChatGPT, Gemini, Copilot and Claude use when deciding who to name. Steadiness matters because coverage fades. In [Muck Rack's May 2026 study](https://www.globenewswire.com/news-release/2026/05/07/3290268/0/en/generative-pulse-earned-media-consistently-drives-ai-citations-holding-at-84.html) of more than 25 million links cited by AI tools, earned media made up 84% of citations, and more than half of the journalism cited had been published in the previous 12 months, with citations dropping sharply after six months. A firm that stops in month three has little left in that window by month nine. This guide sets out what a year of steady, well-aimed PR looks like for a founder-led B2B firm, quarter by quarter: what to count, what the founder must put in, how retainers usually work and how to track AI answers fairly. ## Key takeaways - PR compounds because each quote, article and relationship makes the next one easier to win and more valuable once it lands. - The first quarter builds relationships and material; expect a few quotes and articles but little change in enquiries. - Assistants that search the web live can reflect new coverage within weeks, while answers from a model's memory change only when it is retrained, which takes months. - AI answers vary from run to run, so track the share of a fixed set of buyer questions where you are named, repeated each month, not single screenshots. - A first-quarter report should show activity and relationships, not enquiries, and a six-month term is the least that gives PR a fair test. - Stopping and starting, changing your message and slow founder responses break the compounding and send you back towards the beginning. ## Why does PR compound rather than add up? Because the results of each month feed the next. Four things build on themselves. - **Relationships.** The first time you answer a journalist's request, you are a stranger. The third time, you are a known source. After a few more, they may call you before posting a request at all. - **Material.** An editor considering your pitch will look you up, and a recent article in a respected title makes a yes easier. Coverage is the best credential for more coverage. - **Evidence for AI assistants.** One article is a data point. A year of articles, quotes and listings that describe you consistently is a pattern, and a pattern lets an assistant name you plainly. We explain why agreement across sources matters in [why ChatGPT sounds unsure about your firm](/insights/ai-cross-references-business-sources). - **Links and search standing.** Coverage in respected titles often links back to you, and links still count. An [SE Ranking study reported in November 2025](https://www.searchenginejournal.com/new-data-top-factors-influencing-chatgpt-citations/561954/) found the number of referring domains was the strongest predictor of whether ChatGPT cited a page. Google rankings help too, though they do not guarantee a mention in AI answers. None of this happens in a month. All of it builds over a year, provided the effort is steady. ## What should a year of PR look like, quarter by quarter? Groundwork first, then pull, then a reputation that brings work to you. Judge each quarter against its own job. | Quarter | Main job | What to count | A good sign | |---|---|---|---| | Months 1 to 3 | Build the list, topics and founder views | Requests answered, pitches sent, quotes and articles won | Journalists start replying | | Months 4 to 6 | Turn contact into pull | Quotes in target titles, requests that arrive unprompted | A journalist comes back directly | | Months 7 to 9 | Become a regular source | Invitations to write or speak, enquiries mentioning coverage | Prospects arrive knowing your founder's views | | Months 10 to 12 | Choose well and plan year two | Share of buyer questions where you are named, enquiry sources | Coverage comes to you more than you chase it | Set your own targets for each count from your month-one baseline. The right number of quotes depends on how many titles your buyers read, which varies by sector. ### Months one to three: groundwork that feels slow You build the list of journalists, titles and events that matter to your buyers, agree the three to five topics you want to be known for and prepare your founder's views on each. You start answering requests for comment and sending short notes with no pitch attached. If you are running research, the questions are being written. Expect a handful of quotes, perhaps one accepted article, and a clearer picture of which titles are open to you. Enquiries are unlikely to change. Assistants that search the web live may already find the newest pieces, but answers drawn from a model's memory will not have moved. ### Months four to six: the first signs of pull Journalists who quoted you once come back, sometimes directly. Articles pitched in the first quarter are published. If you ran research, the findings are out, often producing the busiest month so far. Your sales team starts using coverage in follow-up emails. When you put buyer questions to assistants that search the web, some of your coverage may appear among their sources and descriptions of your firm may become more specific. This is the stage that firms which paused in month three never reach. We cover how research fits the rhythm in [how to turn our own data into coverage in the trade press](/insights/leverage-data-research-press-coverage). ### Months seven to nine: becoming the name that comes up Your founder is a regular source for one or two trade titles and is invited to write, not just allowed to pitch. Coverage starts to refer back to earlier coverage. Enquiries mention an article or a talk, and early sales conversations get shorter because credibility is already there. For questions close to your topics, you are more likely to be named and described in the terms your coverage uses. For broad category questions you may still be missing, because established competitors have years of coverage behind them. Be selective: say no to opportunities outside your topics, however flattering. ### Months ten to twelve: a reputation that works without constant pushing Journalists contact you when your topics are in the news. Your research, if you repeat it, is anticipated. Your founder's name is linked with specific subjects in your sector, and your sales team uses coverage as a matter of routine. Across ChatGPT, Google's AI features, Copilot, Gemini and Claude, you should see more consistent descriptions and more appearances for the questions your buyers ask. How far you have come depends on your starting point and how crowded your category is. Plan year two: repeat what worked, drop what did not and add one bigger step. ## How long does it take for PR to show up in ChatGPT answers? It depends on which kind of answer. Assistants that search the web live, such as ChatGPT search, Perplexity, Copilot and Google's AI features, can reflect a new article or page within weeks. Answers drawn from a model's own memory change only when the model is retrained, which takes months. The clearest dated example we can give comes from website work, not PR, but it shows how fast live answers can move once the evidence exists. For ID-Pal: "UK buyer questions where AI names ID-Pal: none in May, two in August, four in September 2026." And: "Where a page has been rebuilt, ID-Pal appears. Where it has not yet, it does not, which tells us exactly what to build next." ([ID-Pal case study](/work/id-pal)). Coverage works on the same live-search route, and it has to keep coming: as the Muck Rack figures above show, journalism more than six months old is cited far less. ## How do we track progress when AI answers change every time? Treat AI answers as a sample, not a snapshot. Assistants give different answers to the same question from one run to the next, so one screenshot proves little in either direction. ### Step 1: Record your starting point Before the first pitch, write a fixed list of 20 to 50 questions your buyers ask. Put each to ChatGPT, Google's AI features, Copilot, Gemini and Claude in a clean session, and save the answers. List your coverage in target titles over the past year and note how enquiries currently hear about you. ### Step 2: Keep a monthly log Record each quote, article and talk with title and date, any journalist who contacted you first, and any enquiry that mentioned coverage. Run the same questions several times each month and record the share of runs in which you are named. Count a win only when it repeats. This is how we measured it for ID-Pal: "The same buyer questions run every month in ChatGPT, Gemini and Claude, in a fresh chat. A win only counts when it repeats, and losses are reported too." ### Step 3: Review each quarter against its own job In the first quarter, judge activity and relationships. In the second, look for pull. In the third and fourth, look for enquiries and changes in AI answers. Judging month three against month-twelve expectations is how good programmes get cancelled. ### Step 4: Compare the year end with the baseline Put the month-one and month-twelve answers side by side with the coverage lists and enquiry sources. For the mechanics, see [how to find out whether ChatGPT is recommending our firm](/insights/track-ai-recommending-your-business). ## How much founder time does each quarter need? The most in the first quarter, when views are written and relationships started, and then a steady weekly rhythm. What cannot be delegated is a fast, clear response when a journalist calls. For example, a founder might plan on something like this (illustrative hours, adjust to your sector): | Quarter | What the founder does | Example hours a week | |---|---|---| | Months 1 to 3 | Agree topics, write views, brief the team, first replies | 3 | | Months 4 to 6 | Answer requests, approve articles, first interviews | 2 | | Months 7 to 9 | Interviews, writing and speaking invitations | 2 | | Months 10 to 12 | Interviews, year-two planning | 1 to 2 | ## How are UK PR retainers usually structured? Many are monthly retainers with a minimum initial term, because the first months are groundwork. [Orange PR](https://www.orangepr.co.uk/how-much-does-a-pr-agency-cost), for example, works on a minimum initial commitment of six months, and [MirrorMe PR](https://mirrormepr.co.uk/how-much-does-pr-cost-in-the-uk/) says most campaigns take three to six months to build momentum. For prices, ask for a written quote. Before you sign, check the minimum term, the notice period after it, who does the work day to day, and what you own if you leave: the media list, the drafts and the research. A first-quarter report should contain: - the list of titles and journalists, and why each matters to your buyers - every request answered, pitch sent and reply received, with dates - coverage won, with links - your baseline of AI answers and enquiry sources - the plan for the next quarter, and what the agency needs from your founder A six-month term is usually long enough to see pull, but not the full effect on enquiries or AI answers. Judge it on the quarter-two signs above. ## What breaks the compounding? Most failures come from broken rhythm rather than bad strategy. - **Stopping and starting.** Three busy months followed by three quiet ones let relationships go cold. - **Changing the message.** If you describe your firm differently each quarter, coverage does too, and assistants cannot form a clear picture. - **Saying yes to everything.** Opportunities outside your topics dilute what you are known for. - **Going quiet after a big win.** A major feature is a platform for the next stage. - **Slow responses.** A journalist who waits a day for your comment will not wait next time. - **Switching approach every few months.** Changing agency or method before anything has had time to work keeps you in the groundwork phase. Which kinds of PR to put into the rhythm is covered in [which kinds of PR win work for a B2B firm](/insights/pr-for-b2b-what-works). The cost of waiting to start is covered in [whether we can wait a year before doing anything about AI search](/insights/first-mover-advantage-ai). ## What to do this month 1. Write your fixed list of buyer questions and save today's answers from the main assistants as your baseline. 2. Agree the three to five topics you want to be known for, and keep them fixed for the year. 3. Block the founder time for the first quarter in the diary now. 4. Start a simple log of coverage, journalists who contact you first and enquiries that mention coverage. 5. Put quarterly reviews in the diary, with each quarter's job written into the agenda. ## Frequently asked questions ### Can we speed this up by spending more? Somewhat. More resource lets you pitch more titles, run research sooner and prepare your founder better, which can shorten the groundwork phase. What money cannot buy is the time it takes for journalists to trust you and for coverage to accumulate. A budget spread over twelve months usually does more than the same budget spent in three. ### How many pieces of coverage a month is good for a B2B firm? There is no universal number. A firm whose buyers read three trade titles needs regular appearances in those three, and a dozen mentions in titles its buyers never read are worth less. Count quotes and articles in your target titles, and track the trend against your own baseline. ### What if nothing seems to be happening after six months? Check the basics before changing strategy. Are you aiming at the titles your buyers read? Are pitches specific to each editor's readers? Is your founder responding quickly with clear views? Most stalled programmes have a fixable problem in one of those. ### When should I give up on my PR agency? When the activity is missing, not when enquiries are slow in month three. If after a quarter they cannot show the requests answered, pitches sent and replies received, or reports list mentions in titles your buyers never read, raise it. If nothing changes by the end of the minimum term, it is fair to leave. ### Does one big feature in a national title shortcut the process? It helps, but it does not replace the pattern. A national feature is a strong credential and may bring enquiries for a few weeks, but buyers and assistants look for consistent evidence over time. Use it as a platform for the next quarter's pitches. ### What happens after the first year? The work gets easier but does not stop. More opportunities arrive unprompted and your role shifts towards choosing well. Firms that stop in year two find the effect fades, because older coverage is cited less. If you want a partner to run that rhythm with you and show you the evidence each month, that is how [Reputation That Wins](/reputation-that-wins) is set up. You can see the kind of report we mean in our [example monthly client report](/samples/monthly-report). ## Sources - [Muck Rack, What is AI reading? Earned media drives 84% of AI citations (May 2026)](https://www.globenewswire.com/news-release/2026/05/07/3290268/0/en/generative-pulse-earned-media-consistently-drives-ai-citations-holding-at-84.html) - [Search Engine Journal, SE Ranking data on the top factors influencing ChatGPT citations (November 2025)](https://www.searchenginejournal.com/new-data-top-factors-influencing-chatgpt-citations/561954/) - [Orange PR, How much does a PR agency cost (April 2025)](https://www.orangepr.co.uk/how-much-does-a-pr-agency-cost) - [MirrorMe PR, How much does PR cost in the UK (April 2026)](https://mirrormepr.co.uk/how-much-does-pr-cost-in-the-uk/) --- Title: How much does AI search optimisation (GEO) cost in the UK? URL: https://fortitudemedia.ai/insights/ai-search-optimisation-cost-uk Author: David Adams Published: 2026-03-31 Last reviewed: 2026-09-24 Topic: Costs, returns and the business case Short answer: Published UK prices for AI search optimisation, which some agencies call GEO (generative engine optimisation) or AEO (answer engine optimisation), run from about £800 a month for entry retainers to £2,000 to £10,000 for programmes and £10,000 or more for enterprise work (MarGen, June 2026). Tracking tools start at about $29 a month, and audits from about £500. Quotes vary because scope varies. Set your own budget from a new client's lifetime gross profit and plan for twelve months. It is budget season. Your finance director wants next year's numbers, and somewhere on the list is a line nobody quite knows how to fill: whatever we are doing about AI search. You know buyers are asking ChatGPT and Gemini who to hire. You have proposals from three agencies at three very different prices, each describing a slightly different thing. Some of those agencies will call the work GEO (generative engine optimisation) or AEO (answer engine optimisation). Both mean the same practical job: getting your firm named and cited when buyers ask ChatGPT, Google's AI Overviews and AI Mode, Copilot, Gemini, Perplexity or Claude. Others call it AI SEO. [Google's own guide](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide) names GEO and AEO and says that, for Google's AI features, optimising for them is still SEO. The label matters less than what the fee buys. UK agencies now publish their price ranges, so there is a market rate to compare against. This guide sets out what those ranges are, what each level usually buys, what tools and audits cost, and then the method we would use to choose your own number: start from what a new client is worth to you. ## Key takeaways - Published UK ranges run from about £800 a month for entry retainers to £10,000 to £20,000 or more for enterprise programmes, with most serious programmes between £2,000 and £10,000 a month (MarGen, June 2026). - Doing it yourself with a tracking tool costs from about $29 a month (Otterly, September 2026) plus your team's time; audits run from about £500 for a short action plan to £5,000 or more. - Quotes vary because scope varies: monitoring, content, earned coverage and website work are different jobs, and cheap packages usually cover only the first two. - Set your budget from a new client's lifetime gross profit, then work out how many extra clients would repay it, on a lifetime and a same-year basis. - Count your own team's time as a cost and plan for twelve months, with review points at months three, six and twelve. - Nobody can guarantee what an AI assistant will say; a provider can guarantee effort, reporting and the outcomes it controls. ## How much does GEO cost per month in the UK? Published UK price guides put most monthly retainers between about £800 and £10,000, with enterprise work above that. The ranges below come from UK agencies' own pricing guides, so treat them as what sellers say the market charges, not as an independent survey. | What you buy | Typical UK price | What it usually covers | Source | |---|---|---|---| | Do it yourself with a tool | From about $29 a month, plus your time | Tracking a set of questions; the work itself is yours | Otterly (September 2026) | | Independent consultant | £400 to £850 a day | Audits, advice, schema and content guidance | Jason Burns (August 2026) | | Audit or one-off review | About £500 for a 90-day action plan, up to £5,000 for a full audit | Crawler access, schema, content gaps, a citation baseline | Jason Burns (August 2026) | | Entry retainer | £800 to £2,000 a month | "Basic AI visibility monitoring and lightweight content optimisation" | MarGen (June 2026) | | Programme | £2,000 to £10,000 a month | Several content pieces a month, entity work, a dedicated account manager or strategist, regular strategy calls | MarGen (June 2026) | | Enterprise | £10,000 to £20,000 or more a month | Multi-brand strategy, real-time monitoring, digital PR, a dedicated team | MarGen (June 2026) | [MarGen's guide](https://www.margen.net/geo-agency-pricing-guide-2026-uk/), updated in June 2026, sets out the four monthly bands and says the entry band typically includes two to four content pieces a month, basic schema markup and monthly reporting. [Jason Burns](https://jasonburns.co.uk/insights/how-much-does-geo-cost-uk), an independent UK consultant, gives the day rates and audit prices. [Gripped](https://gripped.io/geo-agency/how-much-does-geo-cost-a-b2b-saas-pricing-guide/), which works with B2B software firms, quotes £3,000 to £8,000 for a standalone audit. The spread is wide because the same three letters cover very different amounts of work. For our own fees, see [what it costs](/what-it-costs). ## Why do quotes for the same work vary so much? Because providers are quoting for different jobs under the same name. Before you compare prices, check which of these each proposal actually includes: - **Monitoring.** Running your buyers' questions through the assistants and reporting what they say. Cheap to deliver with a tool. - **Content.** Pages and articles that answer buyer questions plainly. Cost depends on how many pieces, and whether someone senior interviews your experts. - **Earned coverage.** Trade press commentary, listings, reviews and awards. This is the expensive, slow part, and it matters: [Muck Rack's May 2026 analysis](https://www.globenewswire.com/news-release/2026/05/07/3290268/0/en/generative-pulse-earned-media-consistently-drives-ai-citations-holding-at-84.html) of more than 25 million links cited by ChatGPT, Claude and Gemini found earned media made up 84% of them. - **Website work.** Fixing a site that assistants cannot read or trust. Sometimes quoted separately, sometimes left out entirely. - **Seniority and commitment.** A strategist with sector knowledge costs more than a junior account manager, and some providers ask for minimum terms of three or twelve months. For example, two proposals at £1,000 and £4,000 a month may both be fair prices. They are rarely prices for the same thing. We cover what the lower end usually leaves out in [whether a cheap AI search package will work](/insights/cheap-ai-optimisation-doesnt-work). ## What do AI visibility tracking tools cost? Of the tools we checked, entry plans ranged from $29 to $199 a month, priced mainly by how many questions you track. These are list prices in US dollars as shown on each vendor's pricing page in September 2026. | Tool | Entry price | What the entry plan includes | |---|---|---| | [Otterly.AI](https://otterly.ai/pricing) | $29 a month (Lite) | 15 prompts across ChatGPT, Google AI Overviews, Perplexity and Copilot; other engines as add-ons | | [Ahrefs](https://ahrefs.com/pricing) | $129 a month (Lite) | Brand Radar with 5 custom prompts; Brand Radar AI on its own from $199 a month | | [Semrush](https://www.semrush.com/pricing/) | $199 a month (Starter, SEO plus AI search) | 50 prompts tracked daily across Google, ChatGPT, Perplexity, Gemini and more | A tool tells you where you stand. It does not write pages or earn coverage, so a tool-only budget works for a firm with a marketer who can act on what it shows. Our guide to [finding out whether ChatGPT recommends your firm](/insights/track-ai-recommending-your-business) sets out the manual method, which costs nothing but time. ## How much does an AI visibility audit cost? Published UK audit prices run from about £500 for a short prioritised action plan to £5,000 or more for a full audit ([Jason Burns](https://jasonburns.co.uk/insights/how-much-does-geo-cost-uk)), and [Gripped](https://gripped.io/geo-agency/how-much-does-geo-cost-a-b2b-saas-pricing-guide/) quotes £3,000 to £8,000 for B2B software firms. A useful audit records what the assistants say today for your buyers' questions, checks that AI crawlers can read your site, and ranks the fixes. We offer two entry points: the free [Growth Review](/growth-review), 30 minutes with David Adams and three fixes in writing, and the [Paid Discovery](/discovery), a week of work with every finding ranked and the fixes written so your own team can ship them. [TCC Group](/work/tcc-group) shows what a one-week plan can do: "1 week to a plan TCC's own team built the new site from." ## Is £1,000 a month enough to get recommended by ChatGPT? Sometimes, but only if the rest is already in place. £1,000 a month sits in MarGen's entry band, which buys monitoring and a small amount of content. That can be enough for a firm with a strong website, a clear specialism and existing trade coverage. It is rarely enough for a firm that also needs its site fixed and its reputation built, because earned coverage is the part a small retainer leaves out. If £1,000 a month is the ceiling, spend it on the foundations first: a site that answers your buyers' main questions, accurate listings, and a monthly check of what the assistants say. ## How do we set the right budget for our firm? Start from what a new client is worth and how many you need, not from the price list. Then any quote becomes something you can judge: does it plausibly produce the clients that would repay it? The method takes an afternoon with your finance director and a spreadsheet. ### Step 1: Work out what a new client is worth Use gross profit over the life of a typical client, not first-year fees. For example, if an average client pays £15,000 a year, your gross margin is 50% and clients stay three years, each new client is worth £7,500 a year in gross profit, or £22,500 over the relationship. If you do not know your average client lifetime, look at the last twenty clients you lost and count how long each stayed. ### Step 2: Work out how many extra clients would repay the spend Divide a candidate budget by the value of a client. In the example, £3,000 a month is £36,000 a year. Over a client's lifetime you break even at £36,000 divided by £22,500, which is 1.6, so two new clients. If your finance director wants the money back within the year, it is £36,000 divided by £7,500, which is 4.8, so five. Show both: the lifetime figure is the true return, and the same-year figure is the cash question. ### Step 3: Check the target against your enquiry numbers Turn the client target into enquiries, using your own win rate. If you win one in three qualified proposals, two extra clients means six extra qualified proposals a year, roughly one every two months. For a firm with a clear specialism that is a modest target; in a crowded national market it may take longer. ### Step 4: Choose the level of effort your market needs Run a quick check of who the assistants name for your buyers' questions today, using our guide to [measuring AI share of voice against competitors](/insights/benchmarking-ai-visibility-competitors). If you are already named for some important questions, you are holding ground and a smaller budget may do. If a rival is named for most of them, you are catching up and need foundations plus a year of coverage. If you want to be the default answer in a contested market, expect original research and regular senior commentary. ### Step 5: Add your own team's time and a contingency Nobody can write expert content about your firm without your experts. For example, six hours a month from you and two senior colleagues, at an illustrative charge-out rate of £200 an hour, is £1,200 a month of time that would otherwise be billed. Add a contingency of about a tenth of the external spend for things you cannot predict, such as an awards entry or a website fix. ## What does the money actually pay for? Three jobs, and most wasted spend comes from paying for one and assuming it covers the others. A website that answers buyer questions, in plain language and readable by AI crawlers. Coverage and proof elsewhere, because assistants recommend firms they find described consistently in places they trust. And measurement, the smallest line, without which you cannot tell your board anything at month six. As a starting point, a first-year split of roughly half on website and content, a third on coverage and proof, and the rest on measurement and contingency suits many firms. Shift more to coverage if your website is already strong. ## How should we phase the spend over twelve months? Front-load the foundations, then settle into a steady rhythm. In months one to three, measure where you stand, fix what stops the assistants reading or trusting your site, and write the core pages. In months four to nine, publish every month, earn coverage and track the same questions. In months ten to twelve, compare with the baseline, count enquiries by source and set next year's level. Change can come sooner than a year. Assistants that search the web live can reflect a new page within weeks, while answers drawn from a model's memory change only when it is retrained, which takes months. [ID-Pal](/work/id-pal) is the dated example: "UK buyer questions where AI names ID-Pal: none in May, two in August, four in September 2026." ## Should we hire, use an agency or mix the two? For most founder-led firms with £1m to £20m+ revenue, a mix works best: outside specialists for the work that needs press contacts, website skills and measurement, and one person inside who owns it. One hire rarely brings all of those, and a new in-house marketer will need salary, employer costs, tools and months to build relationships with editors. A mix can also mean an outside plan that your own team delivers. TCC Group took that route: "TCC chose to build it with its own team, and the new tcc.group is live, built to the plan, with nothing about it depending on Fortitude." ## How do we present the budget to the finance director? Present three numbers and one agreement: the cost of doing nothing, the investment (fees, tools, internal time and contingency over twelve months), the break-even target from Step 2, and the review points. At month three, foundations done and a baseline recorded. At month six, named more often for the target questions. At month twelve, enquiries and wins by source against the target. Our guides to [what inaction is costing you](/insights/cost-of-inaction-losing-right-now) and [making the business case to your board](/insights/business-case-ai-optimisation-board) cover both ends. ## What to do this month 1. Pull your average client fee, gross margin and client lifetime, and calculate what a new client is worth. 2. Pick a candidate monthly budget and work out the break-even number of clients on a lifetime and a same-year basis. 3. Ask ChatGPT, Google, Copilot and Gemini ten of your buyers' main questions and note who is named, so you know your position. 4. Put each proposal you have into the table above: which of monitoring, content, coverage and website work does it include? 5. List the internal hours the work will need and who will own it. 6. Agree the budget, contingency and review points with your finance director. ## Frequently asked questions ### What does an AI SEO agency charge? The same as a GEO or AEO agency, because it is the same work under a different name. Expect entry retainers of about £800 to £2,000 a month and programmes of £2,000 to £10,000 a month (MarGen, June 2026). Ask for a line-by-line list of what is included before comparing. ### What share of our marketing budget should go on AI search? There is no reliable published benchmark, and a percentage borrowed from another firm ignores what your clients are worth. Set the amount with the value method above, then compare it with what you spend on channels that stop the moment you stop paying. We make that comparison in [whether being named by ChatGPT is cheaper than Google Ads](/insights/economics-ai-visibility-vs-advertising). ### Can a provider guarantee we will be recommended? No. Nobody can guarantee what an AI assistant will say, so be wary of anyone who promises a place in ChatGPT's answers. A provider can guarantee effort, reporting and outcomes it controls, such as pages published, coverage secured and meetings held. ### Can we pay ChatGPT to be recommended instead? Not inside the answer. ChatGPT has shown labelled ads to UK users on its Free and Go plans since 11 August 2026, and OpenAI says the ads do not change the answer itself. Being named in the answer is still earned. ### Do we need to include the cost of a new website? Only if your current site stops the assistants reading or trusting your firm. If it is slow, thin, hard for crawlers to read or vague about what you do, fixing it usually gives the best return in the first year. See [what a B2B website costs in the UK](/insights/real-cost-of-a-website). ### What if we can only afford a small budget? Spend it on the foundations first and set review points further out, because a smaller budget usually means slower progress. Published ranges tell you what the market charges; your client value tells you what you can afford to pay. For our own fees, see [what it costs](/what-it-costs), or [book a free Growth Review](/growth-review) and we will work through your numbers with you. ## Sources - [MarGen, GEO agency pricing guide 2026: what UK businesses should expect to pay (June 2026)](https://www.margen.net/geo-agency-pricing-guide-2026-uk/) - [Jason Burns, How much does GEO cost in the UK? (August 2026)](https://jasonburns.co.uk/insights/how-much-does-geo-cost-uk) - [Gripped, How much does GEO cost? A B2B SaaS pricing guide (July 2026)](https://gripped.io/geo-agency/how-much-does-geo-cost-a-b2b-saas-pricing-guide/) - [Otterly.AI, Pricing (checked September 2026)](https://otterly.ai/pricing) - [Ahrefs, Pricing (checked September 2026)](https://ahrefs.com/pricing) - [Semrush, Pricing (checked September 2026)](https://www.semrush.com/pricing/) - [Google Search Central, Optimizing your website for generative AI features on Google Search (July 2026)](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide) - [Muck Rack, What Is AI Reading? Earned media drives 84% of AI citations (May 2026)](https://www.globenewswire.com/news-release/2026/05/07/3290268/0/en/generative-pulse-earned-media-consistently-drives-ai-citations-holding-at-84.html) --- Title: How should we organise our website so buyers and AI understand it? URL: https://fortitudemedia.ai/insights/website-architecture-ux-and-ai Author: David Adams Published: 2026-03-31 Last reviewed: 2026-09-24 Topic: Websites that bring work in Short answer: Organise it around the choices your buyers make: a page for each service and each sector, case studies filed under both, articles grouped into topic hubs that link to the relevant service, and people linked to their work. Keep the menu short, addresses readable and link text descriptive. For several regions, use subfolders on one domain with hreflang; for several brands, one group page that ties them together. Clear structure is proven to help Google and very likely helps AI assistants too. Most B2B websites are not designed so much as accumulated. The site launches with a sensible handful of pages. Then a new service arrives and gets a page wherever there is room. A sector push adds a landing page that nothing links to. The blog grows to eighty posts listed newest first. Five years on, nobody could draw the site on a whiteboard. Visitors feel the result. A buyer from a housing association has to guess which menu your social housing work sits under. A prospect finds three articles on one service scattered across different years, unlinked. They leave thinking you do a bit of everything, the opposite of what a specialist firm wants. AI assistants face a similar problem. Google's AI features answer one question by running several related searches at once, which Google calls query fan-out, and [Ahrefs found in March 2026](https://ahrefs.com/blog/ai-overview-citations-top-10) that only 38% of AI Overview citations now come from pages ranking in the top ten, down from 76% in July 2025. A firm with a clear page for each service, sector and common question has more pages that can match those related searches. The structure that helps a buyer find their situation is almost always the one that helps search engines and AI assistants too. ## Key takeaways - Website structure is how pages are grouped, linked and named, and it shapes what buyers and search engines conclude about your expertise. - Organise around the choices buyers make, usually a page for each service and each sector, so they find their situation in two clicks. - Articles grouped by topic and linked to the relevant service page show depth, while a blog listed by date hides it. - Google documents that links and descriptive link text help it understand pages; that this carries through to AI assistants is likely but not proven. - Put insights in a subfolder unless you have a business reason not to, and for several countries use subfolders with hreflang. - A group with several brands needs one place that states how they connect, on the page and in the markup. ## What does "website structure" actually mean? It is the way your pages are grouped, linked and named: what sits in the main menu, which pages sit under which, how pages link to each other in the text, and what the addresses look like. Designers call it information architecture. Think of a filing cabinet against a library. In the cabinet everything sits in the order it was filed, and you find something only if you know where it is. In the library, related works sit together and the catalogue points you from one to the next. Only the library helps a stranger. ## What does a well-structured B2B site look like? Here is one picture for a firm with four services and three sectors. Say a mid-sized law firm offers employment, commercial property, dispute resolution and corporate work, for education, healthcare and manufacturing clients. The main menu has six items, each a hub page, with readable addresses underneath: - **Home** (/) - **Services** (/services/): a hub listing all four - /services/employment-law/, /services/commercial-property/, /services/dispute-resolution/, /services/corporate/ - **Sectors** (/sectors/): a hub listing all three - /sectors/education/, /sectors/healthcare/, /sectors/manufacturing/ - **Our work** (/work/): case studies tagged by service and sector - /work/academy-trust-tupe-transfer/ - **Insights** (/insights/): topic hubs, not a date-ordered blog - /insights/employment-law/ with its guides, such as /insights/employment-law/tupe-for-academy-trusts/ - **People** (/people/): one profile per partner or specialist - **About and Contact** The menu gives the shape. The cross-links in the text do the real work: | Page type | Links to | |---|---| | Service page | Every sector it serves, its case studies, the people who lead it, its topic hub | | Sector page | The services relevant to that sector, sector case studies, sector specialists | | Case study | The service and sector it involved, the people who did the work | | Guide | The service it relates to, its topic hub, the author's profile | | Person | The services and sectors they lead, the guides they wrote | Our own [Who we work with](/for) hub works this way: a page for each sector we serve, each linking to our services, the client work relevant to that sector and the guides written for it. ## Why does structure matter to buyers? Because buyers arrive looking for their own situation, and structure decides how quickly they find it. A finance director at a food manufacturer wants to see your work for food manufacturers. If they cannot find their situation in a couple of clicks, many will assume you do not cover it. On a flat site, a buyer from a multi-academy trust sees a list of practice areas and a news page. On the site above, they click "Education", see how each practice supports schools and trusts, read two education case studies and see the partners who lead that work. Same firm, same expertise, very different first impression. Structure also signals specialism: one long services page makes a firm look like a generalist, however deep its experience. ## Why does structure matter to AI assistants? Because assistants match questions to pages, and a clear structure gives them clear pages to match. When a buyer asks which firms advise multi-academy trusts on employment law, a dedicated page linked from both the education sector page and the employment service page is an easy match. A passing mention in a list is not. Be careful about how far the evidence goes. [Google's link guidance](https://developers.google.com/search/docs/crawling-indexing/links-crawlable) says links help it find pages and understand them, and that "every page you care about should have a link from at least one other page on your site." Google, Bing and the assistants built on their indexes (Gemini, AI Overviews and AI Mode, Copilot, and ChatGPT search, which has drawn on Bing) benefit from that directly. Whether ChatGPT or Claude give weight to your menu position or link counts on their own has not been shown. Our view is that clear structure helps, but treat it as a well-grounded inference, not a proven rule. One limit matters. [Google's guide to AI features](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide), updated July 2026, warns that creating separate pages for every query variation, mainly to manipulate AI answers, breaks its scaled content abuse policy. A page per service and sector is structure. A thin page per keyword is spam. TCC Group shows what poor structure costs. When we reviewed its site, "What failed was underneath: a site organised around internal teams rather than the questions buyers ask, pages that half-answered the same question so neither was cited, and machine-readable data that declared a name, a logo and nothing else." ([TCC Group case study](/work/tcc-group)) ## Should each service have its own page, or one services page? Each service your buyers choose separately should have its own page. One services page with a paragraph per service is hard for a buyer to scan and hard for a search engine or assistant to match to a specific question. The exception is a service nobody buys on its own, which belongs as a section on the page it supports. Then decide which leads the menu, services or sectors. If buyers mainly look for a type of help, such as an audit or a legal specialism, lead with services and show the sectors inside each. If they mainly look for someone who understands their world, lead with sectors. Sector-first often suits buyers who are not specialists themselves. Either way, every service page should link to its sectors and every sector page to its services. Templates often make this harder than it should be, as we explain in [template or custom website: which does a B2B firm need?](/insights/template-vs-custom-website-b2b) ## How should articles and case studies fit in? They should sit with the work they relate to, not in a separate archive listed by date. - **Group articles by topic.** Instead of one blog page with eighty posts newest first, create a few topic hubs, each with a main guide and the articles that go deeper, linked to the relevant service page. We explain how in [how to organise your content so AI assistants see you as the expert](/insights/topic-clusters-ai-understands). - **File case studies under service and sector.** Tag each one with both, so it appears on both pages, and open it with the client type, the problem and the result. - **Connect people to their work.** Profiles should link to the services and sectors people lead and the articles they wrote, which helps buyers see who they would work with and helps machines connect named experts with their subjects. ## How should pages link to each other? Deliberately, with descriptive words, and with the most links pointing to the pages that matter most. - Use descriptive link text. "Read our guide to TUPE transfers for academy trusts" says what the next page is about. Google's guidance names "click here" as an example to avoid. - Link from the text, not only the menu, where a reader needs the next page. - Add breadcrumbs, the short trail such as Home, Services, Employment law, so visitors and crawlers see where each page sits. - Check for orphan pages, which nothing links to. If a page matters, link to it. If it does not, redirect or remove it. To find orphan pages, crawl your site with a crawler tool that follows links from the homepage, then compare its list with your XML sitemap, your CMS page list and the pages receiving visits in Search Console or analytics. Any page on the second lists but missing from the crawl is an orphan. ## What should page addresses look like? Short, readable and consistent with the menu. [Google's URL guidance](https://developers.google.com/search/docs/crawling-indexing/url-structure) recommends readable words rather than ID numbers, with hyphens between words. /services/employment-law/ tells a reader and a crawler what to expect; /?p=456 tells them nothing. Two or three levels deep is enough for most B2B firms. Do not change addresses casually. Each may have links, bookmarks and search positions attached. If a page must move, set up a permanent 301 redirect from the old address to the new one. ## Should our blog be on a subdomain or a subfolder? Put it in a subfolder, such as /insights/, unless there is a business reason not to. Google's [SEO starter guide](https://developers.google.com/search/docs/fundamentals/seo-starter-guide) says of subdomains versus subdirectories, "From a business point of view, do whatever makes sense for your business", so neither is a ranking rule. The practical case for a subfolder is that articles and service pages share one site, one navigation and one set of internal links, which makes the cross-linking above easy. A subdomain such as blog.example.co.uk often ends up on a separate platform, with its own menu, and drifts away from the services it should support. ## How should you structure a site that covers several countries or brands? For several countries, use one domain with a subfolder per region, and tell search engines which version serves which market. [Google's multi-regional guidance](https://developers.google.com/search/docs/specialty/international/managing-multi-regional-sites) lists subfolders as easy to set up and low maintenance, and advises against switching content by visitor IP address, which it calls "difficult and generally not reliable". Mark each regional version with hreflang tags such as en-GB, en-US and en-AU, and [Google's hreflang rules](https://developers.google.com/search/docs/specialty/international/localized-versions) require that "Each language version must list itself as well as all other language versions." | Option | Example | Best for | Watch out for | |---|---|---|---| | Subfolders on one domain | example.com/uk/, example.com/us/ | Most B2B firms: one site, shared authority, easy upkeep | Needs hreflang and a clear region switcher | | Subdomains | uk.example.com | Regions run by separate teams or platforms | Easy to let regions drift apart | | Country domains | example.co.uk, example.com.au | Firms that need a strong local identity | More domains to run and redirect | Keep facts that do not vary by region, such as history, leadership and methods, in one place and share them. arum is a working example: "310 pages rebuilt across three regions, in weeks", with "3 regional versions: UK and Europe, North America, Australia and New Zealand" and "facts stored once so two pages never disagree, and every old address redirected so years of rankings carry over." ([arum case study](/work/arum)) For a group with several brands, the question is whether buyers see them as one offer. If they do, one site with a section per brand is simpler. If the brands sell to different buyers, separate sites can work, but something must connect them. TCC Group's problem was "Three brands, one group. AI could only see the pieces." The plan included "one group page telling the story once." In the markup, Organization data on each site can name the parent and sister brands. ID-Pal needed exactly that after buying NorthRow: "About four hours of structured data work declaring NorthRow part of ID-Pal on both domains." ([ID-Pal case study](/work/id-pal)) ## How do you restructure a site without losing what you have? Plan it on paper first, keep what already works and redirect everything that moves. ### Step 1: List every page you have Export all your pages with their addresses, visits, enquiries and links from other sites. This shows what you must protect. ### Step 2: Map how your buyers choose Write down the services, sectors and problems buyers look for, in their words. This becomes the backbone of the new structure. ### Step 3: Draft the new structure Sketch the menu, the hubs and the pages in each, like the picture above. Decide where every existing page will live, which will merge and which will go. ### Step 4: Plan the redirects Match every old address to its new home or closest equivalent. Nothing that attracts visits or links should end in an error page. ### Step 5: Launch and watch Check search performance, enquiries and error reports weekly for the first two months. Our [technical checklist for AI search](/insights/technical-checklist-ai-ready-website) covers the other checks to run at the same time. ## What to do this month 1. Draw your current site on one sheet of paper, section by section, and note anything you cannot place. 2. Ask three recent clients how they would have looked for your services, and compare their words with your menu. 3. Pick your strongest service and make sure its page links to its sectors, case studies, people and articles. 4. Crawl the site, find the orphan pages, and link to them or redirect them. 5. Decide whether your menu leads with services or sectors, and agree it with your partners or board. ## Frequently asked questions ### How many items should be in the main menu? For most B2B firms, five or six: for example Services, Sectors, Our work, Insights, About and Contact. Everything else can sit inside those sections or in the footer. A long menu makes a firm look unfocused and is hard to use on a phone. ### Should a group with several brands have one website or separate sites? One site if buyers see the brands as parts of one offer, separate sites if each sells to a different buyer. Either way, publish one group page that names every brand and says how they connect, link each brand's site to it, and state the relationship in each site's Organization markup, so search engines and assistants stop treating the brands as strangers. ### Do we need to restructure the whole site at once? No. Many firms build one service or sector properly, with linked case studies, articles and people, and use it as the model for the rest. Restructure everything at once only when piecemeal fixes would take longer. ### How many internal links should an article have? There is no set number. Link wherever a reader would want to go next: the relevant service, a related guide, a supporting case study, the author's profile. Links added only to hit a number make a page harder to read. ### Will changing our page addresses hurt our Google rankings? It can, if old addresses are left broken. Set up a permanent 301 redirect from every old address, update your internal links, and submit the new sitemap in Google Search Console. Expect some movement for a few weeks, and keep the addresses of your strongest pages where you can. ### How long before a new structure shows up in AI answers? Assistants that search the web live, such as ChatGPT search, Perplexity, Copilot and Google's AI features, can reflect new pages within weeks once they are crawled. Answers drawn from a model's own training change only when the model is retrained, which takes months. A clear structure is one of the few website decisions that helps buyers, search engines and AI assistants at once, and it costs far less than a redesign. If you want a site built around how your buyers choose, see how we plan and build a [Website That Sells](/website-that-sells). ## Sources - [Ahrefs, AI Overview citations and top-ten rankings (March 2026)](https://ahrefs.com/blog/ai-overview-citations-top-10) - [Google Search Central, Optimizing your website for generative AI features on Google Search (July 2026)](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide) - [Google Search Central, Link best practices for Google (December 2025)](https://developers.google.com/search/docs/crawling-indexing/links-crawlable) - [Google Search Central, URL structure best practices (December 2025)](https://developers.google.com/search/docs/crawling-indexing/url-structure) - [Google Search Central, SEO starter guide (December 2025)](https://developers.google.com/search/docs/fundamentals/seo-starter-guide) - [Google Search Central, Managing multi-regional and multilingual sites (December 2025)](https://developers.google.com/search/docs/specialty/international/managing-multi-regional-sites) - [Google Search Central, Tell Google about localized versions of your page (September 2026)](https://developers.google.com/search/docs/specialty/international/localized-versions) --- Title: Is being named by ChatGPT cheaper than paying for Google Ads? URL: https://fortitudemedia.ai/insights/economics-ai-visibility-vs-advertising Author: Ashley Kenerson Published: 2026-03-31 Last reviewed: 2026-09-24 Topic: How buying is changing Short answer: In our view, over any period longer than a year, being named by ChatGPT, Gemini, Copilot and Claude usually costs less per client than paid search, because the content and coverage that earn a mention keep working with no fee per recommendation. Paid search wins in the first months. ChatGPT has shown labelled ads to UK users since 11 August 2026, but OpenAI says ads do not change the answer. Work out your cost per client, keep advertising as a baseline and move a fixed share of budget each year. You have been paying Google for clicks for years. The invoice arrives monthly, the leads arrive in roughly the numbers you expect, and when you pause the campaign the phone goes quiet. It works, in the way a taxi meter works. Now someone on your team has pointed out that buyers are asking ChatGPT and Gemini who to hire, and that the answer does not mention you. The founder's question is financial. If we spent the same money on being named by AI assistants instead of buying clicks, would we win more clients, fewer or the same, and how long would it take? The ground is already moving under paid search. [Seer Interactive's November 2025 study](https://www.seerinteractive.com/insights/aio-impact-on-google-ctr-september-2025-update) of 42 organisations found that paid click-through on searches showing an AI Overview fell 68% between June 2024 and September 2025, from 19.70% to 6.34%. The same study found that brands cited inside the AI Overview earned 91% higher paid click-through than brands that were not. The two channels are partners with very different cost curves, and this guide sets out how to split a finite budget between them. ## Key takeaways - Paid search is a rental: every enquiry costs clicks bought at auction, priced by how hard competitors bid. - Being named by AI assistants works more like an asset: the content, coverage and proof that earn a mention keep working, with no fee for a place in the answer. - ChatGPT has sold labelled ads to UK users on its Free and Go plans since 11 August 2026, but OpenAI says ads do not influence the answer. - Paid search usually produces more leads for the same money in the first months, so plan a staged shift rather than a switch. - The channels work together: brands cited in an AI Overview earned 91% higher paid click-through in Seer Interactive's data. - Work out your real cost per client from paid search first; it is the number the other channel has to beat. ## How does paid search really cost you money? Paid search charges you for attention, one click at a time, for as long as you run it. Google auctions the click, you win it, and a stranger lands on your page, most of whom will not become clients. Three things follow: - The cost per client is set by other people. If two competitors bid harder, your price rises without your service changing. - Nothing is left behind. A click that does not convert has no residual value. - The channel stops the day you stop, which is why founders describe paid search as a treadmill. None of that makes it a bad channel: for enquiries next week, it is often the only one that delivers. It is an operating cost, and should be judged as one. ## How do I work out cost per client from Google Ads? Divide what you spent by the clients it produced, not by the leads. Cost per client is the only figure that compares fairly with any other channel. ### Step 1: Pull twelve months of spend and clicks Take total spend and clicks from Google Ads for the last twelve months, so seasonal swings even out. ### Step 2: Count the conversion chain Record how many clicks became enquiries, how many enquiries became first calls and how many calls became clients, using Google Ads conversion tracking and your CRM. ### Step 3: Multiply the leaks together For example, suppose 1 in 20 clicks becomes an enquiry, 1 in 2 enquiries becomes a call, and 1 in 4 calls becomes a client. You then pay for 20 x 2 x 4 = 160 clicks for every client won, so your cost per client is 160 times your average cost per click. The rates here are illustrative; use your own. ### Step 4: Check the price of the terms you rely on Google's [Keyword Planner](https://support.google.com/google-ads/answer/7337243?hl=en), free inside a Google Ads account, shows estimated monthly searches and what it typically costs for your ad to show. Set the location to the United Kingdom, look up the three terms that bring you most clients, note the date and repeat each quarter. ## What is the average cost per click for B2B keywords in the UK? There is no single figure worth quoting. Prices vary widely by term and sector and change with the auction, and we have found no dated, published UK benchmark for B2B terms we would stand behind. The figure that matters is your own: what you pay for the terms that bring clients, and how many clicks it takes to win one. ## How do Google Ads, LinkedIn Ads, ChatGPT ads and being named compare? The three paid channels charge you to be seen; being named in an AI answer is earned. The table sets out how each works for a UK B2B firm today. | | Google Ads | LinkedIn Ads | ChatGPT ads | Being named by AI assistants | |---|---|---|---|---| | How you pay | Per click, at auction | At auction, for clicks or impressions | Mostly per click or per outcome, says OpenAI | No fee per mention; you pay for site, content, coverage and senior time | | Who you reach | People searching Google | People on LinkedIn, targeted by role, company and sector | UK users of ChatGPT's Free and Go plans | Anyone who asks an assistant, on any plan | | When it stops | The day you stop paying | The day you stop paying | The day you stop paying | It keeps working while the evidence stays current | | What is left behind | Landing pages | Some awareness | Some awareness | Pages, coverage, reviews and case studies | | Position relative to the answer | Above and below AI Overviews; ads inside them are not yet sold in the UK | Not in AI answers | Labelled and separate from the answer | Inside the answer | The UK is not among the twelve countries where Google sells [ads inside AI Overviews](https://support.google.com/google-ads/answer/16297775?hl=en), and [OpenAI's ads page](https://openai.com/index/testing-ads-in-chatgpt/) says the Plus, Pro, Business, Enterprise and Education plans show no ads, so a buyer on a company account never sees one. ## What does being named by an AI assistant cost? The cost of being named by ChatGPT, Claude, Gemini or Perplexity is the cost of becoming the obvious answer, paid mostly up front, and then very little per recommendation after that. When a buyer asks an assistant "which firms handle payroll for construction companies in the Midlands", the answer is not sold by auction. The assistant draws on your website, the coverage and directories that mention you, your reviews and case studies, and how consistently they agree. What it cites is overwhelmingly earned: [Muck Rack's May 2026 analysis](https://www.globenewswire.com/news-release/2026/05/07/3290268/0/en/generative-pulse-earned-media-consistently-drives-ai-citations-holding-at-84.html) of more than 25 million links cited by ChatGPT, Claude and Gemini found that 84% were earned media and 0.3% were paid or advertorial. There is no invoice for the mention. ChatGPT now sells ads beside the answer, but OpenAI says [advertising does not influence the answers](https://openai.com/index/expanding-access-to-ai-with-chatgpt-ads/) ChatGPT gives. So the spend is different in kind: a website that says plainly what you do and for whom, in a form AI crawlers can read; articles and case studies that answer buyers' questions; coverage that repeats the same facts about you; and the time of senior people to put their names to it. That looks expensive next to a month of clicks, and in the first year it often is. The difference is month thirteen, when last spring's article, the trade press piece and the case study are all still there, and you are not paying again for each recommendation they produce. How assistants choose which firms to name is explained in [how ChatGPT decides which firms to recommend](/insights/what-is-llm-optimisation). What outside help costs against a pay-per-click budget depends on who does the work: UK market ranges are in [how much AI search optimisation costs in the UK](/insights/ai-search-optimisation-cost-uk), and our prices are on [what it costs](/what-it-costs). A programme is a fixed cost that builds assets; paid search is a variable cost that buys clicks. ## Why do the two cost curves run in opposite directions? In our view, the cost of an enquiry from paid search is flat at best and rising at worst, while the cost of an enquiry from being named starts high and falls. Picture two lines on a chart. The paid line is roughly horizontal: you pay about the same for the hundredth client as for the first, and if the auction heats up, or fewer people click ads because an AI answer sits above them, it tilts upwards. Better pages and tighter keywords help, but the unit is still a click bought at an auction that resets every day. The earned line starts well above it, because you are building assets and getting little back for some months, then bends down as each new piece adds to the evidence, being named brings clients, and clients become case studies. What locks in and what does not is set out in [whether you can afford to wait a year on AI search](/insights/first-mover-advantage-ai). Where the lines cross depends on your sector, your starting point and how consistently you do the work. We have no dataset across UK B2B firms that supports a typical crossover month, so we do not quote one. ## Should I advertise on ChatGPT instead of Google Ads? Not instead, and not yet as a main channel for most B2B firms. ChatGPT ads have been live for UK users on the Free and Go plans since 11 August 2026, and OpenAI says self-service buying is launching across Europe. They reach people at the moment they ask a question, but only on the plans that carry ads, and they sit beside the answer rather than in it. If your buyers use personal accounts, run a small test judged on cost per client; if they use company accounts, most will never see the ad. ## Which channel brings the better customer? An enquiry that arrives because an assistant named you tends to be further along and less price-driven than one that arrives from an advert, because the buyer has already been told you are a credible option. A paid click catches a buyer comparing four suppliers in four tabs, and they arrive cold. An AI recommendation usually comes when the buyer is building a shortlist, and shortlists decide a great deal: the [6sense 2025 Buyer Experience Report](https://6sense.com/newsroom/the-timeline-for-influencing-b2b-buyers-is-shrinking-insights-from-6senses-2025-buyer-experience-report/) found that buying groups had ranked their preferred vendors before first contact in 94% of cases and bought from that first choice 77% of the time. Helm Club, a founder membership community, shows what the channel can become: "1 in 5 new members this year came through an AI recommendation." The full story is in [the Helm Club case study](/work/helm-club). We have seen no study measuring the uplift in deal size across UK B2B firms, so we claim none. More on buyers who arrive pre-sold is in [what your website should do now visitors arrive pre-sold by AI](/insights/death-cold-website-visit-ai-funnel). ## When does AI search break even against paid search? When your cumulative spend on being named, divided by the clients it has brought, falls below your paid cost per client. No general month can be quoted, so measure it for your own firm and plan for a gap of some months in which paid search carries the business. ID-Pal gives a dated example of how being named builds: "UK buyer questions where AI names ID-Pal: none in May, two in August, four in September 2026." Assistants that search the web live can reflect new pages within weeks; answers from a model's memory change only when it is retrained, which takes months. The gap is a real cash cost, and a board that has not been told to expect it will lose its nerve. So stage the shift. For example, a firm might plan: | Year | Paid search share | Building the asset | What decides the next step | |---|---|---|---| | Year one | Around four fifths of the budget | Around a fifth | Whether AI-sourced enquiries start to appear | | Year two | Around three fifths | Around two fifths | Whether the AI-sourced share is rising; if not, find out why | | Year three onwards | A floor against surprises | The main engine | Cost per client by channel, each quarter | These proportions are illustrative; what matters is deciding them in advance and reviewing them each quarter. To put that to a board, see [how to get your board to approve budget for AI search](/insights/business-case-ai-optimisation-board), and for what doing nothing costs, see [what it is costing you not to be recommended](/insights/cost-of-inaction-losing-right-now). ## Why should we keep advertising anyway? Do not switch paid search off; keep it as a baseline while you build the asset, for four reasons. First, the channels help each other: in Seer Interactive's data, brands cited in an AI Overview earned 91% higher paid click-through (7.89% against 4.14%). Second, speed: paid search is the only lever that produces enquiries next week. Third, coverage: some buyers still click adverts, even as AI answers take more of the page, as we explain in [why ranking first on Google does not guarantee a ChatGPT mention](/insights/google-rankings-wont-protect-you). Fourth, diversification: two channels are safer than one. ## What does the comparison look like for a smaller firm? The logic holds at a smaller scale. A firm with a modest budget can still fix its website so it says plainly what it does and for whom, and publish one useful answer to a buyer question each month under the founder's name. In a region or a niche, that is often enough to start being named. ## What to do this month 1. Work out your real cost per client from twelve months of paid search data, using the four steps above. 2. Look up your three most valuable terms in Keyword Planner, set to the United Kingdom, and record the figures with the date. 3. Add "an AI assistant such as ChatGPT" to the "how did you hear about us" options on every enquiry form, and ask the question on every first call. 4. Decide, in writing, what share of next year's budget moves from clicks to building the asset, and what evidence would make you move more. 5. Pick the first three buyer questions your firm should be the answer to, and give each to a named person to write. ## Frequently asked questions ### Can we just keep running Google Ads and ignore AI assistants? You can, and for a while it will work. The risk is that more buyers form their shortlist in an assistant, and you find out late that a competitor has become the name it gives. Treating paid search as the whole plan is a bet that buyers keep clicking ads as AI answers take more of the page. ### What share of our budget should move across in the first year? Enough to build the asset properly, not so much that enquiries collapse first: for example, around a fifth in year one, with more once AI-sourced enquiries show. Decide it with the board. ### How do we know if AI-sourced enquiries are actually arriving? Ask, and check. Add the option to your enquiry form and ask on every first call. Separately, run a fixed list of 20 to 50 buyer questions through the assistants several times a month, counting a win only when it repeats. Helm Club's results are reported the same way: "Qualified calls a week, reported on one page every Monday and checked against Helm's own member records rather than a platform dashboard." Our guide to [tracking whether ChatGPT recommends your firm](/insights/track-ai-recommending-your-business) covers the set-up. ### Does this apply to a firm selling a low-value service? Less strongly. AI recommendations matter most where a buyer chooses a supplier carefully, the purchase is significant and the decision is professional. For a cheap, quick purchase, paid search and price still do most of the work. For most founder-led B2B firms with revenue of £1m to £20m+, the purchase is considered, and the argument applies. ### We are profitable on paid search today. Why change anything? Because that profit depends on paying the auction for as long as you want the enquiries, and on buyers continuing to click ads. The asset you build to be named keeps producing after the spend stops. You can keep the paid channel running while you build it. In our view, adverts are usually cheaper this quarter and being named is usually cheaper over any period that matters to a founder. If you want to know where you stand before you move any budget, [book a free Growth Review](/growth-review): 30 minutes with David Adams, and three fixes in writing afterwards. ## Sources - [Seer Interactive, AIO impact on Google CTR: September 2025 update (November 2025)](https://www.seerinteractive.com/insights/aio-impact-on-google-ctr-september-2025-update) - [Muck Rack, What is AI reading? Earned media drives 84% of AI citations (May 2026)](https://www.globenewswire.com/news-release/2026/05/07/3290268/0/en/generative-pulse-earned-media-consistently-drives-ai-citations-holding-at-84.html) - [OpenAI, Testing ads in ChatGPT (updated August 2026)](https://openai.com/index/testing-ads-in-chatgpt/) - [OpenAI, A milestone in expanding access to AI (August 2026)](https://openai.com/index/expanding-access-to-ai-with-chatgpt-ads/) - [Google Ads Help, About ads and AI Overviews (September 2026)](https://support.google.com/google-ads/answer/16297775?hl=en) - [Google Ads Help, Use Keyword Planner (September 2026)](https://support.google.com/google-ads/answer/7337243?hl=en) - [6sense, 2025 Buyer Experience Report (November 2025)](https://6sense.com/newsroom/the-timeline-for-influencing-b2b-buyers-is-shrinking-insights-from-6senses-2025-buyer-experience-report/) --- Title: Is our website technically ready for ChatGPT and AI search? URL: https://fortitudemedia.ai/insights/technical-checklist-ai-ready-website Author: David Adams Published: 2026-03-31 Last reviewed: 2026-09-24 Topic: Websites that bring work in Short answer: Your site is ready for AI search when the search crawlers behind ChatGPT, Claude, Perplexity, Copilot and Google can get in, and your key pages show their full text without JavaScript. Check robots.txt, your Cloudflare or firewall settings and Bing Webmaster Tools first. Then give each page one clear topic, proper headings, visible dates and accurate structured data. Most fixes are small and need no rebuild, and none guarantees a recommendation. Sooner or later someone on your board or partner group asks whether the website is "ready for AI". The web agency says it is, because the site is fast and ranks well on Google. That answer is often wrong, and not because anyone is being dishonest. A site can do well in Google and still be hard for ChatGPT, Claude or Perplexity to read, because their crawlers fetch pages differently. The difference is not academic. [Vercel and MERJ found in December 2024](https://vercel.com/blog/the-rise-of-the-ai-crawler) that GPTBot made 569 million requests and Claude's crawler 370 million across Vercel's network in a month, together about 20% of Googlebot's 4.5 billion, and that "none of the major AI crawlers currently render JavaScript." A page that builds its words with scripts can look perfect in Google and empty to ChatGPT. The problems we find most often are the same: a security setting blocks AI crawlers without anyone knowing, service pages are built with scripts some crawlers never run, and the homepage says nothing a machine can pin down about who the firm is. Each is fixable, most in days. This is the checklist we use, written for the founder who needs to know what to ask. Hand it to whoever looks after your site. ## Key takeaways - A website is ready for AI search when crawlers can get in, read the full text and tell who you are and what each page covers. - Each AI company now runs separate crawlers for training, for search and for fetching a page when a user asks, and blocking one is not the same as blocking another. - Check robots.txt, Cloudflare or firewall settings and Bing Webmaster Tools first, because Copilot searches Bing and ChatGPT search has drawn on Bing's index. - The main AI crawlers do not run JavaScript, so key text must be in the HTML your server sends. - Clear headings, one topic per page, visible dates and accurate structured data help machines read and quote you correctly. - None of this guarantees a recommendation; it removes the technical reasons an assistant might pass you over. ## What does "AI-ready" actually mean for a website? Three things. AI crawlers are allowed in. They can read the full text of your important pages. And they can tell, without guessing, what each page is about and who your firm is. If any of the three fails, the quality of your content hardly matters. The technical side is the foundation, not the building. Whether an assistant recommends you also depends on what your pages say, how specific your proof is and what other sites say about you. We explain how AI crawlers differ from Google's in [how AI crawlers read your website](/insights/ai-crawlers-vs-google-spiders). ## The checklist to hand to your developer Print this and go through it together. Each line is a yes or a no. | Check | How to test | Who fixes it | Pass or fail | |---|---|---|---| | Search crawlers allowed: OAI-SearchBot, Claude-SearchBot, PerplexityBot, Bingbot, Googlebot | Open yourdomain/robots.txt and read each named group | Developer | Pass / Fail | | Training crawlers (GPTBot, ClaudeBot, Google-Extended) set by a deliberate decision | Same file; decision written down | Founder decides, developer sets | Pass / Fail | | No firewall or CDN block on AI crawlers | Cloudflare: Security Settings, AI bot policies; ask any other host | Developer or host | Pass / Fail | | Site verified in Bing Webmaster Tools, sitemap submitted | Log in, check indexed pages and crawl errors | Developer or marketing | Pass / Fail | | Key pages indexed in Google Search Console | URL Inspection on each key page | Developer or marketing | Pass / Fail | | Key text present without JavaScript | View page source, or fetch as GPTBot (below) | Developer | Pass / Fail | | One H1 and logical H2 and H3 headings | View source or a headings checker | Content owner | Pass / Fail | | Each section opens with a direct answer | Read the first two sentences of each section | Content owner | Pass / Fail | | Organization and Person markup accurate | Rich Results Test or Schema Markup Validator | Developer | Pass / Fail | | Unique titles and descriptions; no stray noindex or nosnippet | A crawler tool, or view source | Developer | Pass / Fail | | Published and updated dates on articles | Look on the page and in the markup | Content owner | Pass / Fail | | Core Web Vitals met | PageSpeed Insights on key pages | Developer | Pass / Fail | | WCAG 2.2 AA basics: contrast, alt text, keyboard use | An accessibility checker and a keyboard test | Developer and content owner | Pass / Fail | ## Can AI crawlers get into your site? Check this first, because it overrides everything else. A crawler is the program a company sends to read web pages. Each AI company now runs several, with different jobs, and your robots.txt file, at the root of your domain, decides which get in. | Company | Training | Search | Fetch when a user asks | |---|---|---|---| | OpenAI | GPTBot | OAI-SearchBot | ChatGPT-User | | Anthropic | ClaudeBot | Claude-SearchBot | Claude-User | | Perplexity | (none named) | PerplexityBot | Perplexity-User | | Google | Google-Extended (a control, not a crawler) | Googlebot | (none named) | | Microsoft | (none named) | Bingbot | (none named) | The split matters. [OpenAI's crawler documentation](https://developers.openai.com/api/docs/bots) says a site can "allow OAI-SearchBot in order to appear in search results while disallowing GPTBot to indicate that crawled content should not be used for training", and that for ChatGPT-User, "robots.txt rules may not apply" because a user started the visit. [Anthropic says](https://support.claude.com/en/articles/8896518-does-anthropic-crawl-data-from-the-web-and-how-can-site-owners-block-the-crawler) blocking ClaudeBot excludes future content from training, while blocking Claude-SearchBot or Claude-User reduces your visibility in Claude's answers. [Perplexity says](https://docs.perplexity.ai/guides/bots) its user fetcher "generally ignores robots.txt rules". Google says Google-Extended controls use in Gemini training and does not affect inclusion in Google Search, which includes AI Overviews. A robots.txt that lets the search and user crawlers in, keeps an admin folder private and sets training crawlers separately looks like this: ```text # Search and user-request crawlers User-agent: OAI-SearchBot User-agent: ChatGPT-User User-agent: Claude-SearchBot User-agent: Claude-User User-agent: PerplexityBot Disallow: /admin/ # Training crawlers: change to "Disallow: /" to opt out of training User-agent: GPTBot User-agent: ClaudeBot Disallow: /admin/ User-agent: * Disallow: /admin/ Sitemap: https://www.example.co.uk/sitemap.xml ``` A crawler follows only the most specific group that names it, so repeat private folders under every group. OpenAI says changes can take about 24 hours to take effect. If your file still names older Anthropic user agents from a 2023 template, replace them with ClaudeBot, Claude-SearchBot and Claude-User. ### Check Cloudflare and your firewall Security services can block AI crawlers before robots.txt is read. Cloudflare [changed its default on 1 July 2025](https://blog.cloudflare.com/content-independence-day-no-ai-crawl-without-compensation/), "changing the default to block AI crawlers unless they pay creators for their content." [Cloudflare's documentation](https://developers.cloudflare.com/bots/additional-configurations/block-ai-bots/) puts the setting under Security Settings, then Configure AI bot policies, with "Block AI bots" on older accounts. It treats search, training and agent crawlers separately, each set to block on all pages, block on pages with ads, or allow. From 15 September 2026, it says, training and agent bots are blocked by default on pages that show ads while search stays allowed. If you use another host or security plugin, ask what its bot protection blocks. ### Check Bing as well as Google [Microsoft says](https://learn.microsoft.com/en-us/copilot/microsoft-365/manage-public-web-access) Copilot "may fetch information from the Bing search service" when it answers, and ChatGPT search has drawn on Bing's index too. A site that Bing cannot crawl, or has never been submitted to it, loses both. Verify the site in Bing Webmaster Tools, submit your sitemap, and check for crawl errors, just as you would in Google Search Console. ## Can they read your pages without running JavaScript? They should. Many modern sites send a nearly empty page and let code in the browser fill in the words. Googlebot runs that code, which helps Gemini and Google's AI features. GPTBot, ClaudeBot and PerplexityBot fetch the page but do not run the code, so they may see little or nothing. ### How to see your site the way GPTBot sees it Open a key page, right-click and choose "View page source", then search for your first paragraph. Or switch off JavaScript in the browser and reload. For an exact test, a developer can run one line that fetches the page with GPTBot's published user agent and counts how often a phrase from your first paragraph appears: ```text curl -s -A "Mozilla/5.0 AppleWebKit/537.36 (KHTML, like Gecko); compatible; GPTBot/1.4; +https://openai.com/gptbot" https://www.example.co.uk/services/ | grep -c "a phrase from your first paragraph" ``` A result of 1 or more means the text is in the HTML. A 0 means crawlers that do not run JavaScript cannot see it. An error page may mean your firewall blocks the name, or that it treats a test from your own computer as an impostor, so check the firewall log before concluding either way. Also watch for text that loads only when a tab or accordion is opened, key facts locked in PDFs or images, and case studies pulled in from another system after the page loads. The fix is usually to have the server send the finished page. We cover platforms in [how your website platform affects whether ChatGPT can read your site](/insights/cms-choice-affects-ai-visibility). ## Is each page clearly about one thing? It should be. Assistants answer specific questions, so they favour pages that deal with one subject clearly. Five pages that each cover one service properly are easier to quote than one page covering five. - **Use headings properly.** One H1 that says what the page is about, then H2 and H3 headings in order. Headings phrased as buyers' questions, such as "How long does an ISO 27001 audit take?", make matching easy. - **Use the right building blocks.** Semantic HTML tags mark the main article, the navigation, lists and tables, which helps crawlers separate content from menus. Comparison data should be a real table, not a picture of one. - **Put the answer first.** Start each section with a plain sentence that answers its heading, then explain. How pages link together matters too, as we explain in [how to organise your website so buyers and AI understand it](/insights/website-architecture-ux-and-ai). ## Does your site say plainly who you are? It should, in words and in structured data. Structured data, often called schema markup, is a set of labels in the page code that states facts in a standard form: this is an organisation called Example Ltd, at this address, with these people. It helps search engines read your facts correctly. Google says its AI features need no special markup, and it stopped showing FAQ rich results on 7 May 2026. OpenAI and Anthropic have not said they read it. Keep it accurate; do not treat it as a ranking trick. The pieces that matter most for a B2B firm are Organization markup on the homepage, Person markup on partner and author pages, Article markup with published and updated dates on guides, and Breadcrumb markup. A cut-down example for a fictional firm: ```html ``` The markup must match what the page says, and the facts must match LinkedIn, Google Business Profile and directories, because conflicting details are a common reason assistants describe firms wrongly. We cover which types are worth the effort in [whether you need schema markup for AI assistants](/insights/structured-data-schema-markup-ai). ## Are your titles, dates and page settings correct? They are easy to overlook and easy to fix. Every important page needs a unique title and meta description. Articles should show when they were published and last updated, on the page and in the markup. A canonical tag should point to the main version of each page. Check that no important page carries a noindex instruction left over from a test site, and that nobody has added nosnippet or a very small max-snippet setting, which limits how much Google can quote, including in AI Overviews. ## Is the site fast, stable and accessible? It should be, because slow and awkward pages lose people. [Google's Core Web Vitals guidance](https://developers.google.com/search/docs/appearance/core-web-vitals) sets three targets: the main content appears within 2.5 seconds, the page responds to a tap or click in under 200 milliseconds (Interaction to Next Paint), and the layout shift score stays under 0.1. For accessibility, the current standard is WCAG 2.2, which [the W3C notes](https://www.w3.org/WAI/standards-guidelines/wcag/) is also an approved ISO standard, ISO/IEC 40500:2025. Level AA asks, among other things, for a contrast ratio of at least 4.5 to 1 for normal text, descriptive alt text on meaningful images, and pages that work with a keyboard. Alt text also gives crawlers words where there would otherwise be none. More in [whether your website's speed and accessibility affect enquiries](/insights/website-accessibility-speed-matter). ## What to do this month 1. Ask your developer to read robots.txt and the Cloudflare or firewall settings, and confirm the search crawlers can reach the site. 2. Decide, in writing, whether to allow GPTBot and ClaudeBot for training, and set robots.txt to match. 3. Verify the site in Bing Webmaster Tools and submit your sitemap. 4. Run the view-source or GPTBot test on your homepage, main service page and best case study. 5. Fix the H1, headings, answer-first openings and any stray noindex on those three pages. 6. Add or correct Organization and Person markup, and validate it before it goes live. ## Frequently asked questions ### What is the difference between GPTBot and OAI-SearchBot? GPTBot collects content that may be used to train OpenAI's models. OAI-SearchBot indexes pages so they can appear in ChatGPT's search answers. Blocking GPTBot keeps your content out of training but leaves you eligible for ChatGPT search; blocking OAI-SearchBot removes you from those answers. ### Can we block AI training but still appear in ChatGPT search? Yes. Set GPTBot to "Disallow: /" and leave OAI-SearchBot allowed, as OpenAI's documentation describes. The same split works for Anthropic: block ClaudeBot for training and allow Claude-SearchBot. For most B2B firms we recommend allowing both, because a training block means future models know less about you. ### How do I check whether Cloudflare is blocking ChatGPT from my site? In the Cloudflare dashboard, open your domain and go to Security Settings, then Configure AI bot policies (or Block AI bots on older accounts), and check how search, training and agent crawlers are set. Then look at the security events log for blocked requests from GPTBot, OAI-SearchBot or ChatGPT-User. If you do not manage Cloudflare yourself, ask whoever does to send a screenshot. ### Does ChatGPT use Bing to find websites? OpenAI says ChatGPT search uses third-party search providers without naming them, and ChatGPT search has drawn on Bing's index. Microsoft Copilot searches Bing directly. So being crawled and indexed by Bing matters for both, which is why Bing Webmaster Tools is on the checklist. ### Should we block AI crawlers to protect our content? For most B2B firms, no. If buyers ask ChatGPT and Perplexity who to hire, blocking their search crawlers removes you from the answers. The exception is content you sell, such as paid reports, which belongs behind a login, because user-triggered fetchers may not obey robots.txt. ### Can we do this on WordPress, Squarespace or Wix? Mostly. WordPress gives the most control. Wix has a robots.txt editor. Squarespace offers [a tick box that blocks a list of AI crawlers](https://support.squarespace.com/hc/en-us/articles/360022347072), including GPTBot and ClaudeBot, rather than direct editing, so check what it covers before you tick it. We compare the options in [template or custom website: which does a B2B firm need?](/insights/template-vs-custom-website-b2b) ### How often should we repeat this check? Run the full checklist once a year and after any redesign, platform change or new security set-up. Check crawler access and key pages every quarter, because plugins, hosting changes and CDN defaults can quietly undo earlier work. Live-search assistants can reflect a fix within weeks; answers from a model's training change only when it is retrained. The technical side is the part of AI search most firms can fix quickly, and it is often what stands between good content and being named. For a view of how your site reads to AI crawlers and buyers, [book a free Growth Review](/growth-review); if the site needs rebuilding so the content is in the HTML, that is [Website That Sells](/website-that-sells). ## Sources - [Vercel and MERJ, The rise of the AI crawler (December 2024)](https://vercel.com/blog/the-rise-of-the-ai-crawler) - [OpenAI, Overview of OpenAI crawlers (September 2026)](https://developers.openai.com/api/docs/bots) - [Anthropic, Does Anthropic crawl data from the web, and how can site owners block the crawler? (April 2026)](https://support.claude.com/en/articles/8896518-does-anthropic-crawl-data-from-the-web-and-how-can-site-owners-block-the-crawler) - [Cloudflare, Content Independence Day: no AI crawl without compensation (July 2025)](https://blog.cloudflare.com/content-independence-day-no-ai-crawl-without-compensation/) - [Cloudflare Docs, Block AI bots (July 2026)](https://developers.cloudflare.com/bots/additional-configurations/block-ai-bots/) - [Microsoft Learn, Manage public web access in Microsoft Copilot (August 2026)](https://learn.microsoft.com/en-us/copilot/microsoft-365/manage-public-web-access) - [Google Search Central, Understanding Core Web Vitals and Google search results (December 2025)](https://developers.google.com/search/docs/appearance/core-web-vitals) - [W3C Web Accessibility Initiative, WCAG 2 overview (September 2026)](https://www.w3.org/WAI/standards-guidelines/wcag/) --- Title: Should I still pay a marketing agency now that AI can do the work? URL: https://fortitudemedia.ai/insights/ai-disrupting-marketing-agency-model Author: Ross Williams Published: 2026-03-31 Last reviewed: 2026-09-24 Topic: How buying is changing Short answer: Keep paying for judgement and outcomes, and stop paying agency rates for production. AI tools now handle much of the drafting, design variation and reporting that agencies charged for by the hour. What is still worth buying from outside is positioning, earned coverage, being named by AI assistants and qualified meetings, with part of the fee tied to results you can count. Split your retainer into its parts, bring production in-house with one owner, and renegotiate the rest on outcomes. The agency report arrives on the first Monday of the month. It is thirty pages long, beautifully designed, and does not contain a single new customer. You have paid the retainer for two years. When you ask what it has produced, the answer is a list of activity: posts published, campaigns optimised, keywords tracked. Meanwhile your operations director has started using ChatGPT to draft the tender responses that used to go to a copywriter. Many founders of B2B firms with £1m to £20m+ revenue will recognise some of this. The agency was hired when production was slow and marketing was opaque, so paying for time made sense. Now much of the production is fast and cheap, and the question of what you are paying for is hard to avoid. Buyers have moved at the same time. In [G2's April 2026 survey of 1,076 software buyers](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html), 51% said they now start their research in an AI chatbot, up from 29% a year earlier. So as well as asking what an agency costs, ask whether anything it does makes you the firm those assistants name. We run a growth firm, so we have a stake in this, which is a reason to be straight about it. Here is what AI has changed, what is still worth paying for, how an in-house hire compares, and how to renegotiate. ## Key takeaways - AI tools now produce first drafts, design variations, ad variants and routine reports quickly, so paying agency rates for that work no longer makes sense. - AI has not made judgement cheap: knowing what to say, to whom, whether it is true, and answering for whether it wins work. - Buyers now ask AI assistants who to hire, and becoming the firm they name is new work most traditional agencies were not set up to do. - A good retainer is split into parts, each with a named owner and a monthly number, so you can see what each part produces. - Tie part of any fee to outcomes the provider controls, such as qualified meetings held, and never to a promised place in ChatGPT's answers. - A marketing manager costs more than the salary: employer National Insurance, pension and your management time come on top. ## What has AI actually changed? AI has made production fast and cheap, and it has made clients better informed. Those were the two things the old agency model depended on: weeks of senior time to produce a strategy, creative and copy, and clients who could not easily judge what that time was worth. **First drafts are no longer scarce.** A founder can ask Claude or ChatGPT for a positioning analysis and a messaging framework and have something usable in an afternoon. It will not match a strong strategist, but it shows what the hours were buying. The same is true of first-draft copy, design variations, ad variants and reports. The scarce thing is knowing which draft is right and why. **Timelines have compressed.** Work that moved through people and revision rounds one after another can now move in parallel. Websites show it most clearly. For [arum](/work/arum), which advises banks, governments and major brands on collections technology, the result was "310 pages rebuilt across three regions, in weeks." That removes the case for long projects priced on effort. **Routine optimisation is increasingly automated.** Google says its Performance Max campaigns use "Google AI across bidding, budget optimisation, audiences, creatives, attribution and more" ([Google Ads Help](https://support.google.com/google-ads/answer/10724817?hl=en-GB)). If your monthly report is mostly bid and budget changes, check how much of it describes something the platform did by itself. Someone still has to set the goal, supply good inputs and check that the leads are real, and that part is worth paying for. ## What has AI not changed? AI has not made judgement cheap, and judgement is where the value in marketing always was. Knowing what a firm should be known for, and by whom, comes from understanding the market, the buyers and the competitors, and from having done it before. Knowing whether content is true and useful, rather than merely fluent, takes someone who knows the sector. And someone has to be willing to be measured on whether the work produces enquiries. AI has also created a new category of work. Buyers ask ChatGPT, Copilot, Gemini, Claude and Perplexity who to hire. ChatGPT alone has more than a billion weekly users ([OpenAI, August 2026](https://openai.com/index/expanding-access-to-ai-with-chatgpt-ads/)), and [Forrester found in January 2026](https://www.forrester.com/blogs/b2b_buyers_make_zero_click_buying_number_one/) that business buyers are four times as likely as consumers to use Microsoft Copilot. What those assistants say depends on what they can read about you across the web. Making a firm the clear, well evidenced answer draws on positioning, plain writing, earned coverage and technical clarity at once, and most traditional agencies have not built it. Our guide to [which assistants buyers actually use](/insights/ai-search-landscape-chatgpt-perplexity-google-claude) explains what each one reads. ## Can I replace my marketing agency with ChatGPT? You can replace much of the production, and for many firms that is the right move. You cannot replace positioning, earned coverage, sector judgement or someone accountable for results, because a chatbot produces text and answers for none of it. | Task | Can AI tools produce a usable first draft? | Who should own it | Worth agency rates? | |---|---|---|---| | First-draft copy, social posts, ad variants | Yes | One person in-house, using AI tools | No | | Design variations and formatting | Yes, for most routine work | In-house, or a designer on call | No | | Standard monthly reporting | Largely, from your own analytics | In-house | No | | Positioning: what you are the obvious choice for | A starting point only | The founder, with an experienced outside view | Yes, once, then a yearly review | | Content that shows real expertise | Drafts only; the substance comes from your people | Named experts, with an editor | Yes, for editing and structure | | Coverage in the titles your buyers read | No | A specialist with editor relationships | Yes | | Being named by AI assistants | No | A specialist, measured monthly | Yes | | Qualified meetings | No: a person should write and approve every message | A specialist, paid partly on meetings held | Yes, tied to results | ## What is still worth paying for? Pay for the four things that produce customers and that are hard to do in-house: positioning, being named by AI assistants, earned authority and qualified meetings. Expect a smaller build, a monthly programme and a small team that uses AI tools openly. ### Positioning the firm Deciding what you are the obvious choice for, and saying it in plain words, is the highest-value work in marketing and the hardest to do from inside. It is worth a proper fee once and a review each year. ### Being named by AI assistants This combines a website that says plainly what you do and for whom, content that answers buyer questions, consistent facts across every source that mentions you, and named experts. How its cost compares with paid search is set out in [whether being named by ChatGPT is cheaper than Google Ads](/insights/economics-ai-visibility-vs-advertising). ### Earned authority Coverage in the titles your buyers read, guest articles, awards and directory listings that repeat the same facts about you. It needs relationships and persistence, which an in-house team rarely has time for. ### Qualified meetings Meetings are measurable, so they should be priced at least partly on results. Our [Pipeline That Fills](/pipeline-that-fills) shows how that can work. Every message is researched, written and approved by a person. Month one is warm-up, and the criteria for a qualified meeting are agreed with the client first. The guarantee covers months two and three: if fewer than six qualified meetings are held across both, month four's management fee is waived. Fees are on [what it costs](/what-it-costs). The principle applies to any provider: a guarantee on meetings held is one the provider can keep, because it controls the effort. Researched outbound reaches buyers before any assistant is asked, and it works alongside being named. ## What should a marketing retainer include in 2026? A retainer should buy a few named outcomes, each with an owner and a monthly number, instead of a bundle of hours. | Part of the retainer | What good looks like | Warning sign | |---|---|---| | Strategy and positioning | A written view of who you are for, reviewed each year | A new strategy deck every quarter | | Content | Answers to buyer questions, with named authors and sourced figures | A count of posts published | | Coverage | Named target titles and pieces actually placed | "Outreach sent" | | Website | Monthly changes tied to enquiries | "Maintenance" with nothing changing | | Being named by AI assistants | A fixed list of 20 to 50 buyer questions, each asked several times a month in a clean session, with a win counted only when it repeats | A screenshot of one good answer | | Measurement | Enquiries and meetings by source, checked against your own records | Impressions, reach and clicks | | Media management | Goals set by a person, results checked against real leads | Changes the platform made itself | Measurement is where most retainers fall down. For [Helm Club](/work/helm-club), the number that matters is "Qualified calls a week, reported on one page every Monday and checked against Helm's own member records rather than a platform dashboard." Ask for something that plain from anyone you pay. Our [example monthly client report](/samples/monthly-report) shows the format we use. ## Should we hire a marketing manager or use an outside team? Hire when you have steady work for one person and someone senior to direct them. Use an outside team when you need several skills at once, or results sooner than a new hire could be recruited and trained. Many founder-led firms end up with both: one owner inside and specialists outside. The salary is only the start. The [National Careers Service](https://nationalcareers.service.gov.uk/job-profiles/marketing-manager) puts a marketing manager's pay at £30,000 for a starter to £65,000 for someone experienced. Employers also pay National Insurance at 15% on earnings above £5,000 a year in 2026 to 2027 ([GOV.UK](https://www.gov.uk/guidance/rates-and-thresholds-for-employers-2026-to-2027)), and at least 3% of qualifying earnings into a workplace pension ([GOV.UK](https://www.gov.uk/workplace-pensions/what-you-your-employer-and-the-government-pay)). For example, a marketing manager on £50,000 costs £56,750 a year in salary and employer National Insurance alone. Pension, recruitment, training and your time managing them come on top, and one person rarely covers writing, design, PR and outbound equally well. | | In-house marketing manager | Traditional agency retainer | Specialist team paid partly on outcomes | |---|---|---|---| | What you pay for | Salary and on-costs, whether or not the work lands | Hours and activity | Named outcomes, part of the fee tied to results | | Range of skills | One person's | Broad, often more than you use | Deep in the areas that produce customers | | Time to start | Months to recruit and train | Weeks | Weeks | | Management load on you | High | Medium | Low, if the reporting is clear | | Best when | Steady work and a senior person to direct it | Large firms with complex needs | Founder-led firms that want results and one owner inside | So which is cheaper? Compare the yearly total of each option with the enquiries and meetings it will produce over the same twelve months. Our fees are on [what it costs](/what-it-costs), and our [comparison with hiring a marketing team](/vs/hiring-a-marketing-team) sets out the trade-offs. ## How do I negotiate performance-based pricing with an agency? Ask what the retainer will produce in enquiries and meetings over the next two quarters, and renegotiate on the answer. If the agency says its value is strategic thinking, good: price the strategic thinking as strategic thinking, and stop paying for production at the same rate. ### Step 1: Split the retainer into its parts Separate strategy, production, coverage, website, measurement and media management, and put a fair current price against each. ### Step 2: Decide what to bring in-house Usually production and routine reporting, given to one named person with AI tools. ### Step 3: Define the outcome in writing Agree what counts before any fee depends on it: the job titles and company sizes that qualify, a real need, and a meeting that was held rather than merely booked, checked against your own records. ### Step 4: Tie part of the fee to results the agency controls Ask the agency to re-price what remains and tie a share of the fee to those outcomes. Nobody can guarantee what an AI assistant will say, so tie AI search work to effort and reporting, and the meetings part to meetings held. If the agency will not re-price, look for a provider that will, and ask it the same questions. Be fair: a good agency that has done real work for you deserves the chance to adapt. What you should not accept is a report describing activity in place of a customer. ## How can you tell a good provider from a cheap one? A good provider tells you what it measures, names the people doing the work, and accepts being judged on enquiries. A cheap one sells volume, and volume without specific, evidenced, named expertise will not make you the firm ChatGPT names. Ask how they would know, in six months, whether the work had succeeded. A list of outputs is a poor answer. A list of buyer questions, who is named for them now and the enquiries expected is a good one. More on this in [how to compare providers](/insights/how-to-compare-ai-optimisation-providers) and [why cheap AI search packages rarely work](/insights/cheap-ai-optimisation-doesnt-work). ## What to do this month 1. List every marketing task the firm pays for, in-house or out, and mark each as production, judgement or measurement. 2. Give the production tasks to one person for a week with ChatGPT, Claude or Copilot, and note what was good enough. 3. Ask your agency, in writing, what the retainer will produce in enquiries and meetings over the next two quarters. 4. Write down what counts as a qualified enquiry and a qualified meeting, and check last quarter's numbers against your own records. 5. Put twenty of your buyers' questions to ChatGPT, Google's AI Mode, Copilot and Claude, and note whether anything the agency has done has made you the answer. 6. Decide what to bring in-house, re-price or stop, and give each remaining piece of work an owner and a number. If you run an agency yourself, read [how an agency gets named when clients ask AI](/insights/ai-optimisation-marketing-creative-agencies). ## Frequently asked questions ### How much does a marketing agency cost per month in the UK? Fees vary widely with scope, seniority and whether media spend is included, so compare what each fee buys rather than the headline figure. Ask every agency to split its quote into its parts and say which are tied to results. Production should cost much less than it did, because the draft is no longer the expensive part, while strategy and judgement cost about what they did. Our own fees are on [what it costs](/what-it-costs). ### What about our website, which the agency built and hosts? Make sure you own the domain, the content and the accounts, and ask for the source. Then improve the site monthly rather than rebuilding it every few years. Whoever looks after it should keep the key text in the page's HTML, because the main AI crawlers do not run JavaScript; [how AI crawlers read your website](/insights/ai-crawlers-vs-google-spiders) covers the checks. ### Our agency says it now offers AI services. Is that enough? Ask what they mean. Using AI tools to work faster is sensible. Making your firm the one ChatGPT, Copilot and Gemini recommend is different work, and the test is whether they can show you today which firms are named for your buyers' questions and how they would change that. ### Can an agency guarantee that ChatGPT will recommend us? No. Nobody can guarantee what an AI assistant will say, so be wary of anyone offering a place in ChatGPT's answers. ChatGPT has shown labelled ads to UK users on its Free and Go plans since 11 August 2026, and [OpenAI says](https://openai.com/index/testing-ads-in-chatgpt/) the ads do not change the answer itself, so being named is still earned. A provider can guarantee effort, reporting and outcomes it controls, such as meetings held. ### What should we keep in-house permanently? Ownership. Someone inside the firm should own the questions your buyers ask, the facts the web states about you, and the monthly numbers. They need not produce everything themselves. They should know what is being produced and why. The agency model is changing because the scarce thing has changed: production is cheap, and judgement, evidence and outcomes are not. For a second opinion on where your marketing spend is going and what it should produce, book a free [Growth Review](/growth-review): 30 minutes with David Adams and three fixes in writing. ## Sources - [G2 via PR Newswire, Half of B2B software buyers now start their research with AI chatbots (April 2026)](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html) - [Google Ads Help, About Performance Max campaigns (checked September 2026)](https://support.google.com/google-ads/answer/10724817?hl=en-GB) - [OpenAI, A milestone in expanding access to AI (August 2026)](https://openai.com/index/expanding-access-to-ai-with-chatgpt-ads/) - [Forrester, B2B buyers make zero-click buying number one (January 2026)](https://www.forrester.com/blogs/b2b_buyers_make_zero_click_buying_number_one/) - [OpenAI, Testing ads in ChatGPT (updated August 2026)](https://openai.com/index/testing-ads-in-chatgpt/) - [National Careers Service, Marketing manager job profile (checked September 2026)](https://nationalcareers.service.gov.uk/job-profiles/marketing-manager) - [GOV.UK, Rates and thresholds for employers 2026 to 2027 (September 2026)](https://www.gov.uk/guidance/rates-and-thresholds-for-employers-2026-to-2027) - [GOV.UK, Workplace pensions: what you, your employer and the government pay (September 2026)](https://www.gov.uk/workplace-pensions/what-you-your-employer-and-the-government-pay) --- Title: Should our content answer the questions our customers actually ask? URL: https://fortitudemedia.ai/insights/content-around-customer-questions Author: Ross Williams Published: 2026-03-31 Last reviewed: 2026-09-24 Topic: Content that gets quoted Short answer: Yes. Buyers now put their questions to AI assistants first, and assistants reach for pages that already answer them. Collect questions from sales calls, lost deals, tenders, support and Search Console, and start with recommendation and comparison questions, because those are the answers that name firms. Put the direct answer in the first two sentences under a heading that is the question, and build about 13 connected pages on one subject in three months. Your sales team answers the same twenty questions every week. How long does it take. What does it cost. What happens if it goes wrong. Why should we use you rather than the firm down the road. They answer them well, from experience, and the answers win work. Then you look at your website and none of those questions are on it. The website talks about your values, your approach and your commitment to excellence. That gap costs more now than it did two years ago, because buyers put those same questions to an AI assistant before they pick up the phone. In [G2's April 2026 survey of 1,076 software buyers](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html), 51% said they now start their research in an AI chatbot, up from 29% a year earlier. An assistant answering "how long does a factory fit-out take for a food manufacturer" looks for a page that answers that question. If your page talks about excellence and a competitor's page answers the question, the competitor gets named. So yes, your content should answer the questions customers actually ask, in their words. This guide covers where to find them, which ones get firms named, how to write answers an assistant can quote, and a plan for the first three months. ## Key takeaways - Build pages around the questions buyers ask on sales calls, in support tickets and in tenders, in the buyer's own words. - A prompt to ChatGPT or Google AI Mode is longer than a search and carries context such as firm size, sector and location, so pages need to answer the follow-up questions as well as the headline one. - Recommendation and comparison questions are the ones where assistants name firms; plain definitions are often answered without naming anyone. - Give the direct answer in the first two sentences under a heading that is the question, then the reasoning, then an example. - FAQ markup no longer produces anything visible in Google, so write FAQ sections for readers and follow-up questions, not for markup. - At one good page a week, three months gives about 13 connected pages on one subject; order and linking matter more than volume. ## Why do AI assistants favour content built around questions? Because a question is what they are given and an answer is what they have to produce. They reach for sources that already look like an answer to the question in front of them. When a buyer asks ChatGPT "which accountants specialise in dental practices in the Midlands", a page headed "Accountants for dental practices: what to look for and who we work with" fits the shape of the answer. A page headed "Our services" does not. Where the answer sits matters too: an analysis of 1.2 million ChatGPT answers by Kevin Indig, [reported by Search Engine Land](https://searchengineland.com/chatgpt-citations-content-study-469483) in February 2026, found 44% of citations came from the first 30% of a page. Questions also come in families. "What is a managed service agreement", "what should be in one", "how much should one cost" and "what happens when one ends" belong together. Google's AI Mode and AI Overviews use what Google calls query fan-out, running several related searches for one question ([Google Search Central, July 2026](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide)), so a site that answers the whole family has more chances to be picked up. How to organise those pages is covered in [how to organise content so AI assistants see you as the expert](/insights/topic-clusters-ai-understands). ## How is a buyer's prompt different from a Google search? A search is a few words. A prompt is a paragraph with context, several questions at once and a follow-up. Google says questions in AI Mode have been [two to three times longer than traditional searches](https://blog.google/around-the-globe/google-europe/united-kingdom/ai-mode-search-uk/) since it reached the UK in July 2025. Example: the same need, written three ways. | Where | What the buyer types | What your page needs to answer | |---|---|---| | Google search | outsourced payroll provider uk | Who you are and what you do | | ChatGPT prompt | We're a 60-person engineering firm in Leeds moving payroll out of house after our payroll manager retires. Which UK providers should we look at, what should we ask them, and what usually goes wrong in the first quarter? | Who the service suits, how to choose, what to ask, and the common problems in the first months | | AI Mode follow-up | Which of those handle salary sacrifice pensions and give us a named account manager? | The specific features and service details a shortlist is decided on | A page that only says what you do answers the search. The prompt and the follow-up need pages on who it suits, how to choose, what goes wrong and what it costs. ## How do we find out what questions buyers ask ChatGPT about our industry? OpenAI does not publish what people ask, so no tool can show you the prompts. You can get close by combining what buyers tell you with the public signals of what they search. Start inside your own firm, because you have better data than any keyword tool. ### Your sales team and call recordings Ask each salesperson for the ten questions they hear most on first calls and the five objections at proposal stage, in the buyer's words. "Will this disrupt our production line" is a buyer's question; "implementation concerns" is not. If you record calls with a tool such as Gong or Fathom, search the transcripts for question marks and for phrases like "how do you" and "what happens if". And ask new clients directly: "what did you ask ChatGPT or Google before you called us?" ### Lost deals and tenders CRM notes on lost deals show the questions you answered badly or too late. Tender clarification questions are gold: they are written by buyers, in their words, about exactly what worries them. ### Support inbox, onboarding and account reviews Questions that arrive after the sale, such as "can we add another site", show what nobody explained, and prospects ask them too. Tag a month of tickets and the patterns show. ### Search data and the questions Google shows Filter Google Search Console for queries starting with how, what, why, which, can, should or does. Look at Google's "People also ask" boxes for your main subjects, and use a tool such as AlsoAsked, which maps those boxes into families of related questions. ### Ask the assistants themselves Put a buyer question to ChatGPT, Gemini, Claude, Copilot and Perplexity, and to Google AI Mode, and note the follow-up questions they suggest and which firms they name. That shows you who owns the answer today. For a routine version, see [how to find out whether ChatGPT is recommending our firm](/insights/track-ai-recommending-your-business). Put everything in one list. In our experience most firms end up with sixty to a hundred questions. Group them by subject and by type. ## Which questions matter most? Recommendation and comparison questions first, because those are the answers in which assistants name firms. Cost and process questions next, how-to questions after that, and plain definitions last. A plain definition such as "what is a fractional finance director" is often answered from the model's own memory, without naming or citing anyone. A recommendation question such as "which fractional finance directors work with SaaS firms in the UK" has to name someone. [Peec AI's study of 30 million sources, reported by Search Engine Land in March 2026](https://searchengineland.com/ai-search-engines-cite-reddit-youtube-and-linkedin-most-study-473138), found review platforms such as G2 appeared often in recommendation queries, which is why reviews and third-party lists matter alongside your own pages. | Question type | Example | Does the answer name firms? | Priority | |---|---|---|---| | Recommendation | "Which IT support firms suit a 50-person accountancy practice?" | Yes, directly | First | | Comparison and decision | "In-house or outsourced IT support for a 50-person firm?" | Often | First | | Cost and process | "What does outsourced IT support cost, and what changes the price?" | Sometimes, and it cites sources | Second | | How-to | "How do we move from on-site servers to the cloud?" | Rarely a supplier, but it cites the source | Third | | Definition | "What is a managed service agreement?" | Usually not | Fold into other pages or a glossary | The "names firms" column is our experience from running buyer questions for clients, not a published study. Definitions still matter as building blocks: put a short, clear one inside the pages that need it, or in a [glossary that AI assistants can use as a reference](/insights/glossary-knowledge-base-ai-references). For example, a commercial insurance broker's first pages might be "professional indemnity versus public liability for consultants", "which brokers specialise in cover for design firms", "how much professional indemnity cover does a design firm need" and "what happens if we under-insure", rather than "our approach to client service". ## Should we publish a "best [service] firms in the UK" page on our own website? We would not publish a ranking with your own firm at number one. Buyers see through it, and assistants draw heavily on what others say about you. Publish fair comparison pages instead, and work to appear on lists and review sites that other people run. A fair comparison page names the alternatives, says who each suits, and admits when a competitor is the better fit. It answers the question a buyer is asking when weighing options, which is why it gets used. For [Helm Club](/work/helm-club), a founder membership community, "We built the structured data AI engines read, then a comparison page for every question a founder asks when weighing up peer groups: Helm against YPO, Vistage and the rest." The result: "Every comparison page we have published ranks first for its question on Google and the AI engines. The pages not yet built are the ones where a competitor still writes the answer." ## How do we structure an answer so an assistant can use it? Answer first, reasoning second, example third, on every page and in every section. The title should be the question in the buyer's words. The first paragraph should answer it in one or two plain sentences that make sense lifted out and read alone. Then explain why, then give an example a reader can picture. For [ID-Pal](/work/id-pal), the rebuilt pages followed exactly this shape: "Five industry pages rebuilt: fund administration, wealth management, legal, accountancy and estate agency. Each opens with the question a compliance lead types, names the regulator, and answers inside the first ninety words." We take apart a page that does this well in [what an article that ChatGPT actually quotes looks like](/insights/anatomy-of-article-cited-by-ai). A weak page opens with "Understanding professional indemnity", followed by a paragraph saying the subject is important and many firms find it confusing. Nothing has been answered. A strong page opens with "What does professional indemnity insurance cover?", followed by "Professional indemnity insurance covers the cost of claims made against your firm for advice, designs or services that caused a client financial loss. It pays legal costs and any compensation awarded, up to the limit you choose." Then the explanation, the exceptions and an example. Be specific: "it covers claims for negligent advice, including legal defence costs" beats "it protects your business". Cover the different ways buyers ask the same thing through your headings and FAQ answers, written naturally. Do not add lines listing variant phrasings; that reads as the keyword stuffing described in [whether ChatGPT and Google will quote content written for SEO](/insights/thought-leadership-vs-keyword-stuffing). ## Do FAQ sections still help now Google has dropped FAQ rich results? Yes, for readers and for follow-up questions, but not because of the markup. Google limited FAQ rich results to well-known government and health sites in August 2023 and [stopped showing them altogether on 7 May 2026](https://www.searchenginejournal.com/google-drops-faq-rich-results-from-search/574429/). FAQPage markup is still valid; it simply produces nothing visible in Google. Nor is an FAQ a shortcut to ChatGPT. [SE Ranking's November 2025 study, reported by Search Engine Journal](https://www.searchenginejournal.com/new-data-top-factors-influencing-chatgpt-citations/561954/), found pages with FAQ sections averaged 3.8 ChatGPT citations against 4.1 for pages without, probably because FAQs sit on simpler support pages. Write a short FAQ where buyers have real follow-up questions, answer each one directly, and keep any markup accurate without expecting it to rank you. ## How should we plan the first three months? One good answer page a week, in a deliberate order, beats four thin posts a week in a random order. At that pace, three months gives about 13 pages on one subject, linked to each other. | Month | Pages | What to write | |---|---|---| | 1 | About 4 | Recommendation and comparison pages for your chosen subject: who it suits, options compared, when a competitor is the better fit | | 2 | About 4 or 5 | Cost and process pages: what it costs and what changes the price, how it works, what goes wrong and how to avoid it | | 3 | About 4 or 5 | How-to and objection pages drawn from lost deals and tender questions, with short definitions folded in where needed | Link each new page to the ones before it and to the service page it supports. Where existing posts already hold the answers, rewrite them to this shape rather than starting again. For how much output a firm like yours needs over a year, see [how often we need to publish for AI assistants to notice us](/insights/how-often-publish-content-ai). ## How do we tell whether it is working? Keep a fixed list of 20 to 50 of your buyers' questions, and ask each one several times a month in ChatGPT, Gemini, Claude, Copilot and Perplexity and in Google's AI features, in a clean session. Answers vary from run to run, so count a win only when it repeats. Record whether you are named, which page is cited and who else is named. The questions where you are not named are your next pages, and where a competitor is named, their page shows what a good answer looks like. Also ask new enquiries what they read before they called. ## What to do this month 1. Ask your sales team and account managers for the questions they hear most, in the buyer's words, and add tender clarification questions and lost-deal notes from the CRM. 2. Pull question-shaped queries from Search Console and the "People also ask" questions for your main subject. 3. Group the list by subject and type, and pick the one subject where you most want to be named. 4. Write the first recommendation or comparison page for that subject, with the question as the title and the answer in the first two sentences. 5. Run ten of your buyers' questions in the main assistants and record who is named, as a baseline. ## Frequently asked questions ### What prompts do B2B buyers use when asking ChatGPT to recommend a supplier? Longer ones than you might expect. They usually describe the firm (size, sector, location), the trigger ("our payroll manager is retiring"), and ask for a shortlist with reasons, then follow up on specific features, price or risk. Write pages that answer each of those parts, not only the service name. ### How many answer pages do we need before ChatGPT starts naming us? There is no fixed number, and one page on a subject rarely does it alone. In our experience a connected group on one subject, answering the recommendation, comparison and cost questions, is what starts to show. Assistants that search the web live can reflect a new page within weeks; answers from a model's memory change only when it is retrained, which takes months. For ID-Pal: "Where a page has been rebuilt, ID-Pal appears. Where it has not yet, it does not, which tells us exactly what to build next." ### What if a competitor already answers the question well? Answer it better, or move on. If their page is thin, general or out of date, a more specific answer with a real example and a named author gives an assistant a better source. If their page is excellent and the question is minor, pick a question nobody has answered well. ### How do we keep answer pages current? Review the pages that bring in work every quarter and the rest at least yearly, update figures and examples, and show a visible updated date. Cost pages need the most frequent checks. [How old is too old for AI assistants](/insights/content-freshness-recency-ai) sets out the cycle. ### What if the questions buyers ask are not the ones we want to answer? Answer them anyway. If buyers keep asking whether they can do it themselves, write the page that explains when they can and when they should not. That page is trusted because it is not a sales pitch, and skipping the awkward questions leaves them to a competitor. The questions are already in your building. The work is writing them down and answering them where an assistant can read them. To see which questions your buyers ask and who is named today, start with a [free Growth Review](/growth-review): 30 minutes with David Adams and three fixes in writing. The rest of our [content strategy and authority guides](/insights/pillar/content-strategy-authority) pick up from there. ## Sources - [G2 via PR Newswire, Half of B2B software buyers now start their research with AI chatbots (April 2026)](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html) - [Search Engine Land, Kevin Indig's analysis of 1.2 million ChatGPT answers (February 2026)](https://searchengineland.com/chatgpt-citations-content-study-469483) - [Google, AI Mode in Search arrives in the UK (July 2025)](https://blog.google/around-the-globe/google-europe/united-kingdom/ai-mode-search-uk/) - [Google Search Central, Optimizing your website for generative AI features on Google Search (July 2026)](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide) - [Search Engine Land, Peec AI study of 30 million sources cited by AI search engines (March 2026)](https://searchengineland.com/ai-search-engines-cite-reddit-youtube-and-linkedin-most-study-473138) - [Search Engine Journal, Google drops FAQ rich results from Search (May 2026)](https://www.searchenginejournal.com/google-drops-faq-rich-results-from-search/574429/) - [Search Engine Journal, SE Ranking data on the top factors influencing ChatGPT citations (November 2025)](https://www.searchenginejournal.com/new-data-top-factors-influencing-chatgpt-citations/561954/) --- Title: Should we build a glossary so AI assistants use us as the reference? URL: https://fortitudemedia.ai/insights/glossary-knowledge-base-ai-references Author: Ashley Kenerson Published: 2026-03-31 Last reviewed: 2026-09-24 Topic: Content that gets quoted Short answer: Yes, if buyers in your field need terms explained and nobody credible explains them well yet. First ask ChatGPT, Perplexity and Google AI Mode to define each term, and skip terms a regulator, gov.uk or a trade body already owns. Give a term its own page only when you can say something substantial about it, and group the rest on hub pages. Expect citations and early visits more than enquiries, so end every entry with when a buyer needs a specialist and link to the work you sell. Every sector has a vocabulary that buyers only half understand. Your clients ask what a schedule of condition is, what separates a lease assignment from a sublet, whether a managed service is the same as outsourcing. Your people answer those questions every week, and the answers are nowhere on your site. When a buyer asks ChatGPT, Perplexity or Google instead, the assistant explains the term, and when it searches the web to do so, it cites whoever explained it best. Usually that is a large firm, a trade body or a regulator. Sometimes it could be you. Definition questions are where AI answers appear most. In a [US study by Pew Research Center published in July 2025](https://www.pewresearch.org/short-reads/2025/07/22/google-users-are-less-likely-to-click-on-links-when-an-ai-summary-appears-in-the-results/), 60% of Google searches that began with a question word such as "what" or "how" produced an AI summary, against 8% of one- or two-word searches. A glossary is the simplest way to answer those questions on your own site. It also has limits most advice skips: assistants define common terms from memory without citing anyone, hundreds of one-paragraph pages can do more harm than good, and being cited for a definition is a long way from being recommended for the work. ## Key takeaways - A glossary earns citations where terms are specialist, UK-specific or recently changed; for common terms, assistants usually answer from memory. - Before writing an entry, ask ChatGPT, Perplexity and Google AI Mode to define the term, and skip terms a regulator, gov.uk or a trade body already owns. - Give a term its own page only when you can fill the entry without padding, and group short, related terms on hub pages. - Definition questions rarely produce a supplier recommendation, so every entry should say when a buyer needs a specialist and link to the matching service. - DefinedTerm markup keeps definitions unambiguous, but Google shows no rich result for it and says its AI features need no special markup. - Link each glossary term from your articles once, at first use, with the term itself as the anchor text. ## Do AI assistants actually use glossary definitions? Sometimes. For a common term, ChatGPT or Claude will usually answer from what the model learned in training, and in our experience often cite nobody. They search and cite more when the term is specialist, UK-specific or recently changed, or when it comes with a practical question such as "do I need one?". That second group is where a glossary earns its keep. Google's AI Overviews and AI Mode also split a question into related searches, which Google calls query fan-out, and cite pages that answer the parts. [Ahrefs found in March 2026](https://ahrefs.com/blog/ai-overview-citations-top-10) that only 38% of AI Overview citations came from pages in Google's top ten, down from 76% in July 2025. So a precise entry can be cited without ranking first for the main query, though ranking still helps. Reputation matters too. In an [SE Ranking study reported by Search Engine Journal in November 2025](https://www.searchenginejournal.com/new-data-top-factors-influencing-chatgpt-citations/561954/), the number of different websites linking to a domain was the strongest predictor of ChatGPT citations. A small firm's definitions can be cited, and are far more likely to be when trade bodies, partners and the trade press link to them. ## Who owns this definition today? Check before you write. For each term, ask ChatGPT (with search on), Perplexity and Google AI Mode "What is [term]?" in a clean session, and note who each one cites. Answers vary from run to run, so ask at least twice and record what repeats. | What you find | What to do | |---|---| | gov.uk, legislation.gov.uk, a regulator or a trade body is cited | Skip the standalone entry; if buyers need the practical angle, cover it on a hub page and link to the official source | | A national firm, publisher or software vendor gives a generic answer | Write a clearer, UK-specific entry with a real example | | A direct competitor is cited | Make it a priority, and note what their entry lacks | | Nobody is cited, or the answer is vague or wrong | A strong candidate, if buyers actually ask it | Most lists shrink once the officially owned terms come off. ## Do glossary pages help SEO or count as thin content? They help when each entry has substance, and count as thin when it does not. A one-paragraph definition that stops there offers nothing the assistant could not say itself, and the method in our [guide to auditing existing content](/insights/audit-existing-content-ai-readiness) would score it as shallow. Google's [guide to its generative AI features](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide), updated in July 2026, says there is "no requirement to break your content into tiny pieces for AI to better understand it". It adds that creating separate content for every variation of how people search, mainly to manipulate rankings, breaks its scaled content abuse policy. A glossary of 150 near-empty pages is that pattern. So test every term: can you write the full entry below, with an example and a real distinction, without padding? If yes, it earns a page. If not, it goes on a hub page with its neighbours. [Why ChatGPT ignores our website when we have hundreds of pages](/insights/ai-penalises-thin-content) explains the wider risk. ## Should each glossary term have its own URL? Only the terms that pass that test. | Format | Use it when | Watch out for | |---|---|---| | Own page per term | Buyers ask about the term on its own, you have plenty to say, and it leads towards work you sell | Overlapping pages; merge them | | Hub page of related terms | Terms are short and usually met together, such as the clauses in a commercial lease | Give each term its own heading and anchor so it can be linked to and quoted | | Definition inside a guide | The term only matters inside one process | Define it at first use and link to the hub | For example, a commercial property firm might give "schedule of condition" and "dilapidations" their own pages and put "rent deposit" and "service charge cap" on a lease terms hub. ## What does a good entry look like? In our experience an entry that gets drawn on has seven parts, and we now add an eighth. - The term as its title, as people say it, with the acronym if there is one. - A one-line definition that could stand alone in an assistant's answer. Put it first: in [Kevin Indig's analysis of 1.2 million ChatGPT answers](https://searchengineland.com/chatgpt-citations-content-study-469483), reported by Search Engine Land in February 2026, 44% of citations came from the first 30% of a page. - A plain explanation: what it is, why it exists, how it works. As long as the term needs. - How it differs from the neighbouring terms buyers confuse it with. - An example from your sector, labelled as one. - Related terms, linked. - A named author, a date and a link to the authoritative source, such as the statute or professional body. - When you need a specialist: the point at which a buyer should get help, linked to the matching service or comparison page. Take a firm of commercial property solicitors writing up "schedule of condition". The one-liner: a record of a property's state at the start of a lease, usually photographs and a written description, used to limit the tenant's repair obligations at the end. Then why tenants want one, how it is prepared and attached to the lease, and what happens without one. Then an example: a tenant taking a five-year lease of an older unit who, without a schedule, could have to return it in better condition than they found it. Then links to "dilapidations", "full repairing and insuring lease" and "break clause", the partner's name and the date. Finally the specialist paragraph: taking a lease on an older building, have the schedule agreed before you sign, with a link to the firm's lease work. [What an article that ChatGPT actually quotes looks like](/insights/anatomy-of-article-cited-by-ai) covers the same structure for longer pages. ## How do I get ChatGPT to use my firm's definition of a term? You cannot make it, but you can make yours the clearest, most current and most referenced version it finds. Assistants that search the web live can pick up a new page within weeks; answers from a model's memory change only when it is retrained, which takes months. 1. **Say what generic definitions do not:** the UK rule, the practical threshold, the mistake buyers make. 2. **Use one wording everywhere.** Two versions of a definition on your site give an assistant a reason to trust neither. Our rebuild for [arum](/work/arum) followed that rule: "Proper structure and markup on every page, a plain summary AI assistants can read, facts stored once so two pages never disagree, and every old address redirected so years of rankings carry over." 3. **Cite the authority.** In a [Princeton and IIT Delhi study presented at KDD 2024](https://arxiv.org/abs/2311.09735), citing sources raised a page's visibility in AI-generated answers by about 30%. It was tested on a research engine, so treat it as directional. 4. **Make it readable.** ChatGPT search needs OAI-SearchBot allowed, and GPTBot, ClaudeBot and PerplexityBot do not run JavaScript, so keep the definition in the HTML. See [how AI crawlers read your website](/insights/ai-crawlers-vs-google-spiders). 5. **Get it referenced** by trade bodies, partners and journalists who need the definition. ## Does a glossary bring in enquiries or just traffic? Mostly citations and early-stage visits. A buyer asking what a term means is still working out the problem, and definition prompts rarely produce a supplier recommendation. Being understood is not the same as being recommended either. At the start of our work with [ID-Pal](/work/id-pal), "AI could describe ID-Pal. It never recommended it." "Across 36 recorded AI answers to the questions ID-Pal's buyers actually ask, ID-Pal was named three times, and only for its brand or its Salesforce integration. AI knew who it was, and had no reason to recommend it." Part of the fix was pages that answer buyers' questions directly, and the result is dated: "UK buyer questions where AI names ID-Pal: none in May, two in August, four in September 2026." So judge glossary pages by the visits they pass to service pages as well as by their own traffic. ## What schema markup should a glossary page use? DefinedTerm for each term and DefinedTermSet for the glossary, both core [schema.org](https://schema.org/DefinedTerm) types. Google shows no rich result for either: neither appears in [Google's list of the structured data it supports](https://developers.google.com/search/docs/appearance/structured-data/search-gallery), updated in June 2026, and Google says its AI features need no special markup. OpenAI and Anthropic have not said whether they read it. Add it for accuracy, and do not expect it to lift rankings. On a hub page, give each term its own DefinedTerm with an @id pointing to its anchor. Keep the description word for word the same as the visible definition. More in [do we need schema markup for AI assistants to understand our site?](/insights/structured-data-schema-markup-ai) ## How do we build the glossary? ### Step 1: List the terms buyers ask about Ask your fee-earners, sales team and support inbox which terms clients ask them to explain, and add the terms in your proposals and engagement letters. [Building content around the questions customers ask](/insights/content-around-customer-questions) is the method. ### Step 2: Test, rank and decide page or hub Run the ownership test, then rank what is left. Tier one is the 25 to 40 terms every buyer meets early, tier two the terms that matter once a project is under way, tier three the specialist ones. Publish tier one first, over two to three months. ### Step 3: Write each entry to the eight-part shape Give each term to the person who explains it best in person, and have a writer shape their explanation. Never define a term using other undefined terms. ### Step 4: Link it into everything else - Link each glossary term once per article, at its first use, with the term itself as the anchor text. - Add a "Terms used on this page" box to long guides. - End every entry with a link to the deeper guide or service page on its subject. - Restate definitions elsewhere in the glossary's exact wording. [How to organise our content so AI assistants see us as the expert](/insights/topic-clusters-ai-understands) covers the site-level pattern. ### Step 5: Maintain it on a calendar Add terms as buyers raise them, review every entry yearly and update at once when a rule changes, with a visible date and a line on what changed. [Seer Interactive found in July 2026](https://www.seerinteractive.com/insights/study-content-recencys-impact-on-ai-visibility-in-2026) that 75% of pages cited by AI assistants had been updated in the previous 12 months, in four consumer and finance industries. See [whether ChatGPT prefers newer content](/insights/content-freshness-recency-ai). ## How can we tell whether it is working? Keep a fixed list of 20 to 50 questions: your published definitions plus the buyer questions they should lead to. Ask them in ChatGPT, Perplexity and Google AI Mode several times a month in a clean session, and count a win only when it repeats. Search Console's generative AI reports, [live for all sites since 31 August 2026](https://developers.google.com/search/blog/2026/06/gen-ai-performance-reports), show your impressions in AI Overviews and AI Mode. [How to find out whether ChatGPT is recommending your firm](/insights/track-ai-recommending-your-business) has the full method. ## What to do this month 1. Ask five colleagues who deal with clients for the ten terms they explain most often, and combine the lists. 2. Run the ownership test on your top 40 terms and strike out the ones an official body owns. 3. Decide for each remaining term whether it earns its own page or belongs on a hub. 4. Write the first five entries to the eight-part shape and publish them in a proper glossary section. 5. Link those terms from your existing guides at first use, and add a "Terms used on this page" box to your three longest guides. 6. Put a quarterly review in the diary. ## Frequently asked questions ### How many terms do we need? Start with 25 to 40 candidates and let the ownership test cut the list. Then grow as buyers raise new terms, with no target count of pages. Forty precise, current, well-linked entries do more than three hundred thin ones. ### Should acronyms have their own entries? No. Put the acronym in the title of the full term, such as "Full repairing and insuring lease (FRI)", and let one page answer both. Separate entries split the value and create duplicates that assistants have to choose between. ### How is a glossary different from an FAQ page? An FAQ page answers scattered questions in one place. A glossary gives each term its own page or anchor, so an assistant can match a specific question to a specific passage. [Google stopped showing FAQ rich results on 7 May 2026](https://www.searchenginejournal.com/google-drops-faq-rich-results-from-search/574429/), so neither format earns a special display; both are judged on their answers. ### Does the glossary need to be public? Yes. An assistant cannot draw on what it cannot read, and neither can Google. Definitions behind a login or in a PDF download do nothing for being named. Keep them on public pages, and check that robots.txt and your CDN let the AI search crawlers in. ### We are a small firm. Will anyone take our definitions seriously? Yes, if they are better and other sites point to them. Links from other websites were the strongest predictor of ChatGPT citations in the SE Ranking study, so offer your definitions to trade bodies, partners and journalists. Small firms do best on specialist, UK-specific terms that large sites cover vaguely. If you want to know which terms in your field are still unowned, and whether assistants describe your firm without recommending it, [book a free Growth Review](/growth-review): 30 minutes with David Adams and three fixes in writing. Expert content that gets a firm named is part of [Reputation That Wins](/reputation-that-wins). ## Sources - [Pew Research Center, Google users are less likely to click on links when an AI summary appears in the results (July 2025)](https://www.pewresearch.org/short-reads/2025/07/22/google-users-are-less-likely-to-click-on-links-when-an-ai-summary-appears-in-the-results/) - [Ahrefs, AI Overview citations from top-ten pages (March 2026)](https://ahrefs.com/blog/ai-overview-citations-top-10) - [Search Engine Journal, New data: top factors influencing ChatGPT citations (November 2025)](https://www.searchenginejournal.com/new-data-top-factors-influencing-chatgpt-citations/561954/) - [Google Search Central, Optimizing your website for generative AI features on Google Search (July 2026)](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide) - [Google Search Central, Structured data markup that Google Search supports (June 2026)](https://developers.google.com/search/docs/appearance/structured-data/search-gallery) - [Search Engine Land, Kevin Indig's study of 1.2 million ChatGPT answers (February 2026)](https://searchengineland.com/chatgpt-citations-content-study-469483) - [Aggarwal and others, Princeton and IIT Delhi, study of visibility in AI-generated answers (KDD 2024)](https://arxiv.org/abs/2311.09735) - [Seer Interactive, Content recency's impact on AI visibility in 2026 (July 2026)](https://www.seerinteractive.com/insights/study-content-recencys-impact-on-ai-visibility-in-2026) --- Title: Should we redesign our website or keep improving the one we have? URL: https://fortitudemedia.ai/insights/website-redesign-vs-continuous-iteration Author: David Adams Published: 2026-03-31 Last reviewed: 2026-09-24 Topic: Websites that bring work in Short answer: For most B2B firms, keep improving the site you have. A rebuild on a fixed cycle leaves the site out of date for much of the time and puts search history at risk when page addresses change. Rebuild only when the platform, the structure or the positioning is wrong, give every old address a permanent redirect, then fund a monthly routine with a named owner so buyers, Google and AI assistants always see the firm as it is today. Your website is four or five years old. A partner says it looks dated, a new salesperson is embarrassed to send prospects to it, and an agency has proposed a full rebuild. Last time, the project took longer than planned, cost more than quoted, and within two years the new site was drifting out of date again. The choice in front of you is less about design than about how you fund and run the website: as a big project every few years, or as working equipment that gets a little attention every month. That matters more now, because buyers increasingly ask ChatGPT, Gemini or Copilot who to hire before they visit your site, and those assistants read what your website says. In a [Seer Interactive study published in July 2026](https://www.seerinteractive.com/insights/study-content-recencys-impact-on-ai-visibility-in-2026), 75% of the pages AI assistants cited had been updated in the previous 12 months. It covered consumer and finance sites rather than B2B services, but the direction is clear: a site frozen since launch works against you. This guide compares the two approaches, the signs that a redesign really is the answer, and how to rebuild without losing your search history. ## Key takeaways - For most B2B firms, steady monthly improvement serves the business better than a full rebuild on a fixed cycle. - A site left alone between redesigns drifts out of date, and buyers, Google and AI assistants all read that version. - A redesign is justified when the platform, the structure or the positioning is wrong, and a tired colour scheme is none of those. - The biggest risk in a redesign is changed page addresses, so give every old address a permanent redirect and keep it for at least a year, as Google advises. - Growth-driven design and website retainers both fund a website as ongoing work, and the scope matters more than the label. - Judge either approach by enquiries and by whether buyers and AI assistants describe you correctly, not by how the homepage looks. ## Why do most firms end up redesigning every few years? Because websites are bought as projects and then left alone. The site is built, everyone is pleased at launch, and then the work stops. Nobody owns it, changes pile up, and by the third or fourth year it describes a firm that no longer quite exists. Starting again is the only option that feels big enough. In our experience many B2B firms rebuild every three to five years, and some agencies advise a shorter cycle: [Bigfork, a UK web agency](https://www.bigfork.co.uk/insights/all-you-need-to-know-about-b2b-website-redesigns), suggests thinking about a redesign every two to three years as a rule of thumb. The cycle persists for a financial reason. A redesign is easy to approve as a one-off project with a clear deliverable. A monthly budget feels like a running cost with no end date, so it is the first thing cut. ## What goes wrong in the years between redesigns? The website slowly stops telling the truth about your firm: people who have left, case studies from years ago, services in the words you used before you changed direction. AI assistants then repeat that picture to buyers. We set out the costs, and how to size them, in [what an outdated website costs a B2B firm](/insights/hidden-cost-static-website). ## What are the risks of a full redesign? A redesign fixes a lot at once and puts a lot at risk at once. - **Changed page addresses.** If old addresses are not redirected (automatically forwarded) to new ones, links from other sites break, Google loses track of pages it had ranked, and AI assistants find nothing where they expect a page. [Google's guidance on site moves](https://developers.google.com/search/docs/crawling-indexing/site-move-with-url-changes) (updated August 2026) says to expect ranking fluctuations while it recrawls, and that a small to medium-sized site "can take a few weeks for most pages to move". - **Lost content.** Pages quietly doing useful work, such as an old guide that still brings enquiries, get dropped because they do not fit the new design. - **Text that crawlers cannot read.** The main AI crawlers, GPTBot, ClaudeBot and PerplexityBot, fetch pages but do not run JavaScript ([Vercel and MERJ, December 2024](https://vercel.com/blog/the-rise-of-the-ai-crawler)). If the new site only shows its text after scripts run, those crawlers see an empty page. - **Months of distraction.** Briefing, design reviews, copy and approvals take senior time from the people who win and deliver the work. - **A fresh start that ages from day one.** Unless the way the site is run changes, the next redesign is already coming. We hear it often enough to have a page for it: [we rebuilt the website and nothing changed](/sound-familiar/we-rebuilt-the-website-and-nothing-changed). ## How do we keep our Google rankings during a website redesign? Keep every page that earns visits, links or enquiries, give every old address a permanent redirect to its closest new equivalent, and measure before and after. Plan for a temporary dip and check that it closes. This is the checklist we use. ### Step 1: Crawl the old site and list every address Use a crawling tool, your sitemap, analytics and server logs to list every old address, with the visits, links and enquiries each brings, so you know which pages must survive. ### Step 2: Benchmark rankings, enquiries and AI answers Record rankings for your main terms, visits and enquiries by page, and what AI assistants say about you: a fixed list of 20 to 50 buyer questions, each asked several times in a clean session in ChatGPT, Gemini, Copilot and Perplexity, counting a result only when it repeats. Our guide to [finding out whether ChatGPT recommends your firm](/insights/track-ai-recommending-your-business) sets out the method. ### Step 3: Map every old address to a new one Match each old address to its closest new page, one to one where you can, and never send everything to the homepage. Use permanent (301 or 308) redirects, point internal links straight at the new addresses, and keep the redirects "generally at least 1 year", in Google's words, or indefinitely if you can. ### Step 4: Launch and test on the day Test every redirect. Check that robots.txt and any "noindex" settings left from the test site are not blocking search engines or AI crawlers, submit the new sitemap in Search Console, and send a test enquiry through every form. ### Step 5: Recheck at one, four and twelve weeks | When | What to check | What good looks like | |---|---|---| | One week | Redirect errors, missing pages, forms, crawler access | No broken old addresses; every form delivers | | Four weeks | Rankings and visits for the benchmarked pages | Any dip is closing | | Twelve weeks | Enquiries by page, and the same buyer questions in the same assistants | Enquiries at or above the benchmark; assistants describe you correctly | Assistants that search the web live (ChatGPT search, Perplexity, Copilot, Google's AI features) can reflect a rebuilt page within weeks. Answers from a model's own memory change only when it is retrained, which takes months. This is how we ran the rebuild for [arum](/work/arum), which advises banks, governments and major brands on collections technology: "310 pages rebuilt across three regions, in weeks." "215 articles, research pieces and news items moved across, going back to 2014." "Every old address redirected, so years of rankings carried over." And at the end: "When arum's team sent a final list of 77 changes, we checked each one against the live site before touching a word." ## What does continuous improvement look like in practice? It means treating the website like working equipment: a budget each month, one person responsible, and improvements made steadily rather than saved up. A sensible monthly rhythm: - **Add something useful.** One or two pages or articles that answer questions buyers actually ask. - **Update something old.** Refresh a case study, a service page or a team profile. - **Fix what the data shows.** The pages people leave quickly, the forms they abandon, the pages that load slowly. Once a quarter, review how enquiries arrive and what AI assistants say about you. Once a year, check the structure still fits how you sell, and plan any new section or visual refresh then. For example, take an accountancy practice with three offices and a site built three years ago. Rather than rebuild, it spends a year adding a page for each of its four specialist sectors, updating every partner profile, answering one client question a month and refreshing the homepage. By the end of the year the site describes the firm accurately and has never been offline or lost a page. If you are not sure your current site is doing its job, [how to tell whether your website is working](/insights/evaluate-whether-website-working) gives a test you can run in an afternoon. ## What are growth-driven design and a website retainer? Growth-driven design is continuous improvement with a method attached: launch a smaller, sound site quickly, then improve it in regular cycles based on what visitors do. HubSpot teaches it as a [course in its Academy](https://academy.hubspot.com/courses/growth-driven-design), and many agencies sell it under that name. A website retainer is a way of paying: a fixed monthly fee to an agency or freelancer for an agreed amount of work. | Term | What it usually means | What to check before you sign | |---|---|---| | Redesign or rebuild | A new site, often on a new platform with new page addresses | A page list, a redirect plan and a plan for after launch | | Growth-driven design | A smaller launch, then improvement in regular cycles | Who sets priorities, which data they use, what counts as done | | Website retainer | A monthly fee for agreed hours or tasks | Whether it covers new pages, writing and design, or only fixes | | Maintenance plan | Hosting, backups, security and software updates | It keeps the site running; it does nothing to keep it accurate | A retainer that only covers hosting and patches is a maintenance plan by another name. To keep the site accurate, the scope must include writing and publishing, with someone inside the firm supplying the facts. ## How does each approach affect whether AI assistants name you? Steady improvement usually serves you better, because AI assistants favour sources that are clear, current and consistent with what others say about you. Across about 17 million citations, [Ahrefs found in July 2025](https://ahrefs.com/blog/do-ai-assistants-prefer-to-cite-fresh-content) that pages cited by AI assistants were on average 25.7% fresher than Google's organic results. Seer's 2026 study points the same way for the monthly path: among cited pages that showed both dates, 72% had been updated in the past year, but only 42% had first been published in it. Updating an existing page counts. Freshness has to be real. Google's guidance on [helpful, people-first content](https://developers.google.com/search/docs/fundamentals/creating-helpful-content) warns against changing page dates to seem fresh when the content has not substantially changed. Update what the page says, then the date. We cover how old is too old in [whether ChatGPT prefers newer content](/insights/content-freshness-recency-ai). ## What are the signs we need a new website rather than updates? A redesign is justified when the problem is in the foundations: a platform your team cannot edit, a structure that no longer matches how you sell, or a change of name. Tired colours and an old font are fixable on the site you have. | What you see | Usually fixable on the current site | Usually a reason to rebuild | |---|---|---| | The design looks dated | Refresh colours, type and images a section at a time | Only if the platform cannot change its templates | | Only the agency can edit pages | Sometimes, with training or a simpler editor | The platform is unsupported or locked to one supplier | | The structure no longer fits how you sell | Add a section for a new sector or service | The menu and page addresses are built around an old model | | A merger or new name | Small wording changes | Usually, with work to tell Google and AI assistants the old and new names are one firm | | Key text only appears after JavaScript runs | Sometimes, by switching on server-side rendering | Often, if the whole site is built that way | If your platform is the question, read [how your CMS affects whether AI assistants can read your site](/insights/cms-choice-affects-ai-visibility) and [whether a template or a custom site fits a B2B firm](/insights/template-vs-custom-website-b2b). Even when a rebuild is right, aim for a foundation that is easy to improve every month, or the cycle simply restarts. More is in our [website design and performance hub](/insights/pillar/website-design-performance). ## How do you compare the costs fairly over five years? Compare over five years, including costs that never appear on an invoice. Put your own numbers against each line: | Cost line | Redesign every few years | Sound site, improved monthly | Where to find your figure | |---|---|---|---| | Build | One build now, another in year four or five | None if the site is sound, or one foundation build | Last invoice and the new proposal | | Running costs | Hosting and maintenance | Hosting, maintenance and a monthly budget | Current contracts | | Senior time | Heavy during each build | A few hours a month for review | Diaries from the last project | | Migration risk | Each rebuild risks rankings and links | Page addresses stay stable | Search Console around the last launch | | Months the site is accurate | Mostly the year after each launch, in our experience | Every month | A date check of your main pages | Example: if five partners each spend 20 hours on briefings, workshops and approvals during a rebuild, that is 100 hours of senior time before anyone has written a page. Price those hours at your own charge-out rate and add them to the build line. Fill it in with your finance director. Our guide to [what a B2B website costs in the UK](/insights/real-cost-of-a-website) covers sourced market figures for builds and running costs. ## What to do this month 1. **Check the site against the firm.** List every service, sector, office and senior person, and note where the site is wrong or silent. 2. **Decide whether the foundations are sound.** Use the signs table above. 3. **Set a monthly budget and an owner.** One named person with a modest monthly amount beats a large budget spent every four years. 4. **Fix the three most-visited pages first.** Usually the homepage, your main service page and the team page. 5. **Start a fixed list of buyer questions.** Record what the assistants say each month, so you have a benchmark whichever path you choose. 6. **If a rebuild is agreed, crawl and benchmark first.** Complete steps 1 and 2 of the checklist before any design work starts. ## Frequently asked questions ### How often should a B2B website be fully redesigned? Only when the platform, structure or positioning is wrong. Some agencies suggest every two to three years, and many firms rebuild every three to five because the site was left alone. With monthly improvement and an annual review, a well-built site stays current much longer. ### How long does a B2B website redesign take? Agency estimates vary with size and complexity. Bigfork puts a redesign at "anywhere between 4 to 14 weeks or more" (October 2024), and [Rudo, another UK agency](https://rudo.co.uk/insights/guides/b2b-website-redesign-timeline/), gives 10 to 16 weeks from kick-off to launch (April 2026), with content preparation the most common cause of delay. Allow about twelve weeks after launch before you judge the result. ### How much does a website retainer cost per month in the UK? It depends on scope, from maintenance-only plans to retainers that include new pages, writing and design. Compare what is included before you compare fees, because a retainer that only covers updates will not keep the site accurate. Our guide to [the real cost of a B2B website](/insights/real-cost-of-a-website) gives sourced UK market figures, and our own fees are on [what it costs](/what-it-costs). ### We have just paid for a new site. What should we do now? Set a monthly budget and name an owner before the launch glow fades. Check that every old address redirects and no useful content was dropped. Then add something useful every month; habits are easiest to set in the first year. ### Who should own the website day to day? One named person with authority to make changes and a small monthly budget, often a marketing manager or operations lead working with an outside team for design and technical work. What matters is that someone is accountable for keeping it accurate. A website that improves a little every month usually does more for your firm than one rebuilt every few years and ignored in between. If you want a site built to be improved rather than replaced, see how [Website That Sells](/website-that-sells) works, or book a free [Growth Review](/growth-review) and leave with three fixes in writing. ## Sources - [Seer Interactive, Study: content recency's impact on AI visibility in 2026 (July 2026)](https://www.seerinteractive.com/insights/study-content-recencys-impact-on-ai-visibility-in-2026) - [Google Search Central, Site moves with URL changes (August 2026)](https://developers.google.com/search/docs/crawling-indexing/site-move-with-url-changes) - [Google Search Central, Creating helpful, reliable, people-first content (December 2025)](https://developers.google.com/search/docs/fundamentals/creating-helpful-content) - [Vercel, The rise of the AI crawler (December 2024)](https://vercel.com/blog/the-rise-of-the-ai-crawler) - [Ahrefs, Do AI assistants prefer to cite fresh content? (July 2025)](https://ahrefs.com/blog/do-ai-assistants-prefer-to-cite-fresh-content) - [Bigfork, All you need to know about B2B website redesigns (October 2024)](https://www.bigfork.co.uk/insights/all-you-need-to-know-about-b2b-website-redesigns) - [Rudo, B2B website redesign timeline (April 2026)](https://rudo.co.uk/insights/guides/b2b-website-redesign-timeline/) --- Title: Should we turn our webinars and podcasts into written content? URL: https://fortitudemedia.ai/insights/video-podcast-transcripts-ai-content Author: Ross Williams Published: 2026-03-31 Last reviewed: 2026-09-24 Topic: Content that gets quoted Short answer: Yes, and put the video on YouTube too. YouTube was the most-cited website in Google AI Overviews in Ahrefs' 2026 study, and Gemini can process video directly, but AI engines mostly cite YouTube through its transcripts and descriptions. Give each video a question-led title, a full description, chapters and corrected captions, and publish an edited, named, question-led transcript page on your own site. In our experience, one 45-minute recording yields a main article and several shorter pieces for about a day of editing. You have a webinar recording from March that your sales team still sends to prospects. One of the partners records a podcast every fortnight. There is a conference talk on YouTube with a few hundred views. All of it is good, and all of it took real effort. Yet when a buyer asks ChatGPT or Google who understands their problem, the recordings rarely count, because almost none of what was said exists as edited text anyone can quote. Video itself is not invisible to AI. YouTube was the most-cited website in Google's AI Overviews in [Ahrefs' study of more than 3 million US queries](https://ahrefs.com/blog/most-cited-domains-ai-overviews/), updated in September 2026, with 22.9% of citations among the 50 most-cited sources. But the citations go to YouTube's page, they depend on the words around the video, and the buyer may never reach your site. So publish each recording twice: on YouTube, described well enough to be cited, and on your own site as an edited page an assistant can read and quote. The expertise has already been said out loud; the work is turning it into text worth quoting. ## Key takeaways - YouTube was the most-cited website in Google AI Overviews in Ahrefs' 2026 study, so a well-described video can be cited, especially in Google's answers. - AI engines mostly cite YouTube through its transcripts and descriptions; Gemini can process video directly, but the words on and around the video do most of the work. - Publish each recording on YouTube with a question-led title, a full description, chapters and corrected captions, and link it to a transcript page on your own site. - A raw automated transcript rarely gets quoted: edit out the filler, name the speakers and structure the page around buyer questions. - In our experience, one well-edited 45-minute recording yields a main article and several shorter pieces for about a day of editing. - Start with the recordings that match the work you most want to win, whatever their view counts. ## Does ChatGPT watch YouTube videos or read their transcripts? It depends on the assistant. Google documents that its [Gemini models can process video](https://ai.google.dev/gemini-api/docs/video-understanding), using both the audio and the visual streams, and can take a YouTube link as input. OpenAI has not said that ChatGPT search watches videos, and ChatGPT, Claude and Perplexity build their answers mainly from text they can read. In practice, text decides most citations even for video. In a [Peec AI study of 30 million sources reported by Search Engine Land in March 2026](https://searchengineland.com/ai-search-engines-cite-reddit-youtube-and-linkedin-most-study-473138), YouTube ranked among the five most-cited domains across ChatGPT, Google AI Mode, Gemini, Perplexity and AI Overviews, and "dominates video citations via transcripts and descriptions". A video with a vague title, no description and uncorrected captions gives an assistant little to work with. On your own site, GPTBot, ClaudeBot and PerplexityBot fetch pages but do not run JavaScript ([Vercel and MERJ, December 2024](https://vercel.com/blog/the-rise-of-the-ai-crawler)), so a transcript loaded by a script or player is invisible to them. Put the text in the page's HTML. ## Why are our videos and podcasts not getting us recommended? Usually for three reasons. The words are not in a usable form: auto-captions with mangled names, no description, a podcast that exists only as audio on Spotify. Nobody is named, so there is no expert for an assistant to credit. And the recording carries a label rather than a question, such as "Client webinar, March 2026", which matches nothing a buyer types. An assistant answering "who can help a manufacturing firm with an ERP migration?" is comparing sources it can read and attribute. If your explanation lives in an unedited recording and a competitor's lives on a clear web page, the competitor is the one who gets named. For example, a monthly hour-long webinar and a fortnightly 45-minute podcast add up to more than thirty hours of expert conversation a year, more than most marketing teams could write in that time. ## How do I get our YouTube videos cited in Google AI Overviews? Treat each YouTube page as a text page with a video on it. Nobody can guarantee a citation, but these six elements give Google and the assistants something to cite. | Element | What to do | Why | |---|---|---| | Title | The question the video answers, in the buyer's words | Matches the question and the related searches Google runs for it | | Description | The answer in the first two lines, then who is speaking, the key points and a link to the transcript page | AI engines cite YouTube largely through descriptions and transcripts | | Chapters | Timestamps in the description with question-led labels | Readers and Google can jump to the part that answers a specific question | | Captions | Upload corrected captions instead of relying on automatic ones | Auto-captions often get names and technical terms wrong | | Speaker | Name and role in the description and on screen | Expertise needs a name to be credited | | Link | A link from every video to its transcript page on your site | Sends viewers and crawlers to the page you control | YouTube's own rules for chapters: the [first timestamp must start at 00:00, you need at least three in ascending order, and each chapter must be at least 10 seconds long](https://support.google.com/youtube/answer/9884579?hl=en-GB). On captions, [YouTube warns](https://support.google.com/youtube/answer/6373554?hl=en-GB) that automatic captions "might misrepresent the spoken content" and encourages creators to add professional captions first. Your edited transcript is most of the way to a corrected caption file. ## Should we host our webinars on YouTube or on our own website? Both. Upload to YouTube for reach and citations, and embed the video on a page on your own site that carries the edited transcript. Google says it [may index a video both on your page and on the platform's own page](https://developers.google.com/search/docs/appearance/video), and both can appear in its video features. | Option | What you gain | What you lose | |---|---|---| | YouTube only | Reach, and the platform Google's AI Overviews cite most | Your site gets no text, no named expert and often no visit | | Own site only | Full control of the page, the text and the enquiry | YouTube's reach and its citations | | Both, linked to each other | Citations on YouTube, a quotable page on your site, and a path between them | A little extra work per recording | ## Is a raw transcript good enough? No. Pasting a raw automated transcript under the video changes very little. It reads like this: "So, um, the thing with cash flow forecasting is, like, most firms, they do it monthly, right, and, you know, that's fine until it isn't." It is complete, unreadable and hard to quote. Assistants lift sentences that state a point plainly in a paragraph that makes sense on its own, and spoken language almost never does that without editing. Unedited transcripts fail in three ways: filler buries the point, nobody is named, and a wall of text with no headings gives an assistant no way to find the part that answers a question. Editing fixes all three. [What an article that ChatGPT actually quotes looks like](/insights/anatomy-of-article-cited-by-ai) covers what makes a page quotable. ## How do we turn a recording into content that gets quoted? This is the process we use for a 45 to 60 minute recording. In our experience it takes about a working day once the editor has done a few. ### Step 1: Transcribe it Use an automated tool. Microsoft Teams can produce a transcript with each speaker's name and timestamps and export it as a .docx or .vtt file; dedicated services such as Otter.ai, Descript and Rev work from any recording; and OpenAI's open-source Whisper model can be run in-house. Consistency matters more than the choice of tool. Export an editable document with speaker labels and timestamps. ### Step 2: Clean it up Remove filler words, repeated phrases and false starts. Fix names, product terms and technical vocabulary the tool got wrong. Keep the speaker's voice; the aim is a readable page that still sounds like them. Allow an hour or two for a 45-minute recording. ### Step 3: Name the speakers and set the scene At the top of the page, say who is speaking, what their role is, and where and when the recording was made. For example: "[Name], [role] at [your firm], speaking at our client webinar on [date]." An assistant can only credit an expert you have named. [What E-E-A-T is and why AI assistants care about it](/insights/eeat-why-ai-cares-more) explains why. ### Step 4: Structure it around questions Mark where the conversation changes subject. Each change becomes a heading, and the best headings are the questions a buyer would ask: "How long does an ERP migration take for a mid-sized manufacturer?" rather than "Section 3". Under each heading, move the speaker's clearest answer to the top. ### Step 5: Add what was implied, and pull out the quotes Add a sentence wherever a reader arriving cold would be lost. Pull out direct quotes that stand alone and attribute them. In a [Princeton and IIT Delhi study presented at KDD 2024](https://arxiv.org/abs/2311.09735), adding quotations from relevant people raised a source's visibility in AI-generated answers by about 41%; it was tested on a research engine, so treat it as directional. If the speaker quoted a figure, add the source if you can verify it and cut it if you cannot. ### Step 6: Build the FAQ from the questions asked Audience questions and a podcast host's questions are the ones real people ask, in real words. Collect them into a "Frequently asked questions" section with a short, direct answer to each. ### Step 7: Publish it properly Give it its own page, titled with the main question the recording answers, laid out as below. Link to two or three related articles and link back from them. Do not publish it only on a third-party platform. ### Step 8: Cut the derivatives From one edited transcript you will usually get a main article, a short summary, a handful of standalone question-and-answer pieces and a run of quotes for LinkedIn. Each should link back to the full page. ## What should the transcript page look like? In this order, so the answer comes first and the detail follows: 1. The question as the title, then the speaker's name, role and the recording date. 2. The embedded video. 3. A short summary: the answer in three to five sentences. 4. Chapters, each a question linked to its timestamp. 5. The edited transcript, under the same question headings. 6. The audience questions as an FAQ, then links to related pages. Add VideoObject markup. Google requires a name, a thumbnail and an upload date, recommends a description, duration and embed link, and uses Clip entries for [key moments in its video results](https://developers.google.com/search/docs/appearance/structured-data/video). Schema.org's VideoObject also has a transcript property. Google does not list it among the properties it uses, so the visible transcript is what counts; the property simply states it for any system that reads it. ## Will publishing a podcast transcript on our site cause duplicate content problems? No. Google's guidance on duplicate pages says [some duplicate content "is normal and it's not a violation" of its spam policies](https://developers.google.com/search/docs/crawling-indexing/canonicalization). Your edited transcript also differs from YouTube's automatic captions and from any platform's machine transcript. The one thing to avoid is the same transcript at several addresses on your own site: publish it once and link to it. ## What makes transcripts better than most written content? Real experience, said out loud by a named person. A partner answering a client's question gives the awkward case, the point where it usually goes wrong, the opinion they would not have put in writing. Google's [guide to its generative AI features](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide), updated in July 2026, says content offering "unique expert or experienced takes" will influence presence in its AI search more than anything else, and a well-edited transcript is exactly that. Recordings are also dated, which helps with questions where the answer changes; [whether ChatGPT prefers newer content](/insights/content-freshness-recency-ai) covers why. And when the same partner appears across ten pages on one subject, the site builds a visible body of expertise. [How to organise our content so AI assistants see us as the expert](/insights/topic-clusters-ai-understands) shows how to group those pages. ## Which recordings should we start with? The ones that match the work you most want to win. For example, a managed IT firm whose best work is cyber security for professional services firms should start with its recordings on that, whatever else is in the archive. A four-question filter: - Does it answer a question a buyer would type into ChatGPT or Google? If not, it goes to the bottom. - Is the speaker someone whose name we want attached to this subject? If yes, it moves up. - Is it still true? A recording about a regulation that has since changed needs editing or dropping. - Does it strengthen a subject where we already have written content? Do three or four in the first month. ## What does it take, and what do we get back? The real cost is editing time: about a day per recording for an editor who knows the subject, plus a short YouTube pass. The expert's time has already been spent, which is the advantage over commissioning an article from scratch. The recording still needs to be worth transcribing, though: a rambling webinar does not become good content through editing. If you are starting a series now, plan each episode around one question a buyer would ask. ## What to do this month 1. List every recording from the last two years, with its subject, speaker and date. 2. Mark the five that best match the services you want to be recommended for, and check they are still accurate. 3. Fix the YouTube title, description, chapters and captions for those five. 4. Transcribe them and give one person ownership of the editing. 5. Edit and publish the first transcript page in the layout above, with VideoObject markup and a link from the video. 6. Add the questions each recording answers to your fixed list of buyer questions, and check them monthly in ChatGPT, Gemini and Google AI Mode. ## Frequently asked questions ### What is the best AI transcription tool for podcasts in 2026? The one that handles your speakers and vocabulary. Test two or three on the same recording and compare how they cope with names, technical terms and crosstalk, and whether they export an editable file with speaker labels and timestamps. The transcription built into Teams is often enough for meetings and webinars run there; dedicated tools or Whisper suit podcasts recorded elsewhere. Every tool still needs a human edit. ### How long should a published transcript be? As long as the useful content, and no longer. Do not pad it, and do not cut it to a summary and call it a transcript. If the recording covered three separate subjects, split it into three pages, each titled with one question. ### Should we include timestamps? Yes, at the start of each main section rather than on every line. They help readers jump to what they need, match your YouTube chapters and show the page is a record of a real event. ### What if the recording has several speakers? Name each one at first mention with their role, and label every change of speaker. If a guest is from another firm, say so and link to them. ### Can we publish transcripts of events we did not host? Only with permission. Ask the organiser before publishing a talk from someone else's conference, and get podcast guests' agreement to publish, with the chance to check their quotes. If you are not sure whether your recordings are worth the effort, check who is being named for the questions they answer. [Book a free Growth Review](/growth-review): 30 minutes with David Adams and three fixes in writing. Turning a firm's expertise into content that gets it named is the work of [Reputation That Wins](/reputation-that-wins). ## Sources - [Ahrefs, The 50 most-cited websites in Google AI Overviews (June 2026, updated September 2026)](https://ahrefs.com/blog/most-cited-domains-ai-overviews/) - [Search Engine Land, AI search engines cite Reddit, YouTube and LinkedIn most: study (March 2026)](https://searchengineland.com/ai-search-engines-cite-reddit-youtube-and-linkedin-most-study-473138) - [Google AI for Developers, Gemini API video understanding (September 2026)](https://ai.google.dev/gemini-api/docs/video-understanding) - [YouTube Help, Use automatic captioning (accessed September 2026)](https://support.google.com/youtube/answer/6373554?hl=en-GB) - [Google Search Central, Video structured data (September 2026)](https://developers.google.com/search/docs/appearance/structured-data/video) - [Google Search Central, Optimizing your website for generative AI features on Google Search (July 2026)](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide) - [Vercel, The rise of the AI crawler (December 2024)](https://vercel.com/blog/the-rise-of-the-ai-crawler) - [Aggarwal and others, Princeton and IIT Delhi, study of visibility in AI-generated answers (KDD 2024)](https://arxiv.org/abs/2311.09735) --- Title: Template or custom website: which does a B2B firm need? URL: https://fortitudemedia.ai/insights/template-vs-custom-website-b2b Author: David Adams Published: 2026-03-31 Last reviewed: 2026-09-24 Topic: Websites that bring work in Short answer: A template is enough for a new or small firm with one clear service, and many firms should start on one. A firm that competes on expertise across several services or sectors usually outgrows it, because it needs a page for each service, sector and proof point, control over structured data and crawler settings, and text that AI crawlers can read without JavaScript. An adapted theme on WordPress or Webflow is often the middle step. Moving is quicker than most founders expect when the content is ready. Most founder-led firms start on a template, and that is usually the right call. When there are four of you and one service, a clean Squarespace site or a well-made WordPress theme gets you online in days for very little money. Nobody in the early days has time to think about how the site should be organised. The trouble starts later. The firm grows to thirty or forty people, adds a second and third service, wins work in new sectors and hires senior people with real reputations. The website still has the shape of the four-person firm: a homepage, an about page, a services page with five short paragraphs, a blog and a contact form. It looks tidy. It just does not explain the business any more. That gap matters more now. Buyers increasingly compare you with firms an AI assistant named alongside you, and the assistant's shortlist moves decisions: in [G2's April 2026 survey of 1,076 B2B software buyers](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html), 69% said they had chosen a different vendor from the one they first planned because of chatbot advice. This guide compares the three routes, template, adapted theme and custom build, and shows how to tell which one your firm needs now. ## Key takeaways - A template is a sensible start for a new or small firm with one service, and a poor long-term fit for firms that compete on expertise across several. - An adapted theme on WordPress or Webflow is often the middle step: more structure and control than a template, less cost than a custom build. - The biggest limit of a template is structure, because B2B buyers need a page for each service, sector and type of problem. - AI site builders such as Lovable, Framer and Bolt are the new templates: fast and cheap, but check that your text arrives in the page before relying on one. - The six signs you have outgrown a template show up in the sales process before they show up in analytics. - Moving takes weeks rather than months when the content is ready, and keeping or redirecting every old address protects your rankings. ## What is the difference between a template, an adapted theme and a custom build? A template is an off-the-shelf design you fill with your words. An adapted theme is a bought design reshaped by a developer to fit your structure. A custom build is designed from your services, sectors and proof outward. Most firms need one of the last two once they compete on expertise. | | Template | Adapted theme | Custom build | |---|---|---|---| | What it is | A hosted builder (Squarespace, Wix) or an AI site builder, with a ready-made design | A theme on WordPress or Webflow, extended with your own page types and sections | Structure, design and page types built around how your buyers choose | | Cost | Lowest: a subscription and your time | Middle: theme plus developer time | Highest up front | | Time to launch | Days | A few weeks | Weeks if the content is ready, longer if it is not | | Control | What the platform exposes | Good, within the theme's limits | Full | | AI readability | Text usually in the page; structured data and crawler settings limited | Good, if the theme sends text in the HTML | As good as you specify: text in the HTML, full structured data, your crawler rules | | Upkeep | The platform handles hosting and updates | Theme and plugin updates, some developer time | Needs an owner inside the firm and a developer or partner | For UK market prices, see [how much a B2B website costs in the UK](/insights/real-cost-of-a-website); ours are on [what it costs](/what-it-costs). The better comparison is cost against the work the site wins or loses. ## When is a template the right choice? When the firm is small, the offer is simple and speed matters more than standing out. A good template, well written, is better than a custom site that takes six months and says nothing. We would rather see a new firm launch on a template this month than wait a year for something perfect. Templates suit firms that have just started and whose proposition is still settling, that offer one clear service to one type of client, that win most work through referrals so the site mainly confirms they are real, or that need something live while a proper site is planned. Take an example. Say two former Big Four tax specialists set up a boutique advising owner-managed businesses on selling up. For the first year, a clean template with a clear homepage, a page on how they work, two partner profiles and a contact page is plenty. The question changes when they add a second service, hire a third partner and start wanting enquiries from people who have never heard of them. ## Are AI site builders the new templates? Yes, in the sense that matters: tools such as Lovable, Framer and Bolt produce a working site quickly and cheaply, often from a written brief. They are a reasonable way to get live. The check before relying on one is whether your words arrive in the page itself. This matters because [Vercel and MERJ found in December 2024](https://vercel.com/blog/the-rise-of-the-ai-crawler) that the main AI crawlers, including GPTBot, ClaudeBot and PerplexityBot, fetch JavaScript files but do not run them. A site that builds its text in the visitor's browser can look perfect to you and empty to ChatGPT. Builders differ, and change quickly. [Lovable's documentation](https://docs.lovable.dev/tips-tricks/seo-geo), for example, says apps created from 13 May 2026 are rendered on the server, while older apps rely on pre-rendered pages served only to verified search and AI crawlers, so "Third-party SEO scanners and other unverified agents see the regular single-page app." Test your own site rather than the tool's claims: open a key page, choose "View page source", and search for your first paragraph. ## Why does structure decide it? Because buyers look for their own situation, and a template's structure rarely has a place for it. Templates assume home, about, services, blog and contact. A firm offering four services across three sectors needs a page for each service and sector, case studies filed under both, profiles linked to the work people lead, and guides grouped by topic. You can build some of that on most platforms, but a template pulls towards fewer, shorter pages and a blog listed by date. We set out the structure in [how to organise your website so buyers and AI understand it](/insights/website-architecture-ux-and-ai). ## Does it matter if our website looks like our competitors'? Yes, more than most founders expect. A buyer comparing three firms is looking for reasons to choose one. If all three sites use similar layouts, similar stock photography and the same phrases about partnership and excellence, the buyer has nothing to separate them on except price and gut feel. The fix is mostly about what you say and can prove: who you serve, specific case studies, real people and a plain account of how you work. A custom build makes it easier to design the site around that material rather than squeeze it into someone else's layout. We cover the signals that win trust in [what makes a buyer trust your website enough to get in touch](/insights/psychology-of-trust-website-conversions). ## What can you not control on a template platform? It depends on the platform and the plan. Self-hosted WordPress gives the most control, Webflow a good deal, and hosted builders less, though they have opened up. - **Plugins and themes.** [WordPress.com](https://wordpress.com/support/plugins/) now allows plugins on all its paid plans, Personal, Premium, Business and Commerce, so the old rule that hosted WordPress cannot be extended no longer holds. - **Structured data**, the labels that tell machines your firm's name, services and people. [Wix lets you add your own JSON-LD markup](https://support.wix.com/en/article/adding-structured-data-markup-to-your-sites-pages) to static pages, up to five markups per page. [Squarespace offers code injection](https://support.squarespace.com/hc/en-us/articles/205815908-Using-code-injection) on its Core, Plus and Advanced plans, which a developer can use for markup. - **Crawler access.** [Wix has a robots.txt editor](https://support.wix.com/en/article/editing-your-sites-robotstxt-file). Squarespace instead offers [a tick box that blocks a list of AI crawlers](https://support.squarespace.com/hc/en-us/articles/360022347072), without direct editing. - **Speed.** Many themes load scripts, fonts and effects on every page whether you use them or not. - **Page delivery**, meaning whether the words are in the page when it arrives or filled in later by scripts. - **Addresses and redirects**, which matter when you reorganise the site or move away. None of these is always a deal breaker, but each is a place where you rely on the platform's choices. We compare platforms in [how your website platform affects whether ChatGPT can read your site](/insights/cms-choice-affects-ai-visibility), and [whether your website is technically ready for AI search](/insights/technical-checklist-ai-ready-website) lists what to ask your developer. ## How do you know if you have outgrown your template? You will usually see it in the sales process before you see it in the analytics. The signs we look for are: - Your sales team sends documents after every call because the site does not explain the offer. - New services or sectors have been squeezed onto existing pages rather than given their own. - Prospects say they did not realise you did something you have done for years. - You have avoided adding content because the platform makes it awkward. - When you ask ChatGPT or Gemini what your firm does, the answer is vague or out of date. - Your developer's answer to technical questions is "the platform doesn't allow that". Two or three of these together suggest the site is holding you back. That does not automatically mean a custom build. Sometimes a restructure on a more flexible platform is enough. ## How long does moving off a template take? Weeks rather than months, if the content is ready. Most of the time goes on structure and words, not design and build. arum is the clearest example we have: "310 pages rebuilt across three regions, in weeks", and "Every old address redirected, so years of rankings carried over." ([arum case study](/work/arum)) Planning can be faster still. For TCC Group it took "1 week to a plan TCC's own team built the new site from", and "TCC chose to build it with its own team, and the new tcc.group is live, built to the plan, with nothing about it depending on Fortitude." ([TCC Group case study](/work/tcc-group)) The slowest projects are those where nobody inside the firm owns the content. A sensible move follows five steps. ### Step 1: List every page you have Note which pages attract visits, links or enquiries. These are what you must protect. ### Step 2: Plan the new structure Build it around your services, sectors and proof, and decide where each old page will live. ### Step 3: Write the content before the design The words are what buyers and AI assistants use, so settle them first. ### Step 4: Build and redirect Set up a permanent 301 redirect from every old address to its new home, and check that key pages send their text in the HTML. ### Step 5: Watch the first two months Check search performance, enquiries and error reports weekly, and fix problems quickly. Jamie Waller, CEO and Executive Chair of arum, put the result this way: "For most people, our website is the first time they meet arum. It now does us justice." ## What to do this month 1. List the services, sectors and types of client you serve today, and compare the list with the pages on your site. 2. Ask your sales team which questions they answer on every first call that the website should answer instead. 3. Open three key pages, choose "View page source", and check your main text is there. 4. Ask your developer, in writing, which changes the current platform will not allow. 5. Ask ChatGPT and Gemini what your firm does, and note what is missing or wrong. 6. Decide whether the gaps can be fixed on the current platform or a move is due, and set a date for the decision. ## Frequently asked questions ### How much does a custom website cost compared with a template in the UK? A template costs a subscription and your time, an adapted theme adds developer time, and a custom build costs most up front, with running costs in every case. Our guide to [how much a B2B website costs in the UK](/insights/real-cost-of-a-website) sets out sourced market ranges and the costs firms forget. Judge the price against what the site wins: a cheap site that loses one good client a year is expensive. ### Is Squarespace good enough for a consultancy website? For a small consultancy with one clear service, often yes. It is quick, looks competent and, in our experience, sends its text in the page. The limits appear as you grow: custom structured data only through code injection on the Core, Plus and Advanced plans, no direct robots.txt editing, and a structure built for fewer, simpler pages. If several of the six signs apply, it is time to plan a move. ### Is Webflow a template or a custom build? Either. Webflow is a platform: you can start from one of its templates, adapt a theme, or build a fully custom design. It gives more control over structure and code than most hosted builders while staying easy to edit, which is why many B2B firms use it as the middle route. Judge it against the technical checklist rather than the sales demo. ### Can a professional services firm use a website built with Framer or Lovable? Yes, if the site passes the same checks as any other. Your text must be in the page source, AI crawlers must be allowed in, you need structured data, clean addresses and redirects, and your team must be able to edit pages without the tool's original builder. Test those before launch, because a quick build that AI crawlers cannot read costs more than it saves. ### How long does it take to move from Squarespace to a custom site? Usually weeks once the content is agreed, depending more on the number of pages and the state of your case studies than on design. The part not to rush is redirects: list every Squarespace address and map it to its new home before you switch. ### Will we lose our Google rankings if we move? You should not lose much if the move is handled carefully. Keep the addresses of your strongest pages where you can, redirect every other old page to its closest new equivalent, and carry the content across rather than starting from scratch. Some movement in the first few weeks is normal. Big losses usually come from missed redirects or deleted content. A template is a fine place to start and a poor place to stay once your firm competes on expertise. If you want a site built around how your buyers choose, with the structure AI assistants need, see how we approach it in [Website That Sells](/website-that-sells), or ask for a free [Growth Review](/growth-review) first. ## Sources - [G2, New G2 research: half of B2B software buyers now start their research with AI chatbots (April 2026)](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html) - [Vercel and MERJ, The rise of the AI crawler (December 2024)](https://vercel.com/blog/the-rise-of-the-ai-crawler) - [Lovable Documentation, search and AI crawler visibility (September 2026)](https://docs.lovable.dev/tips-tricks/seo-geo) - [WordPress.com Support, Plugins (August 2026)](https://wordpress.com/support/plugins/) - [Wix Help Center, Adding structured data markup to your site's pages (September 2026)](https://support.wix.com/en/article/adding-structured-data-markup-to-your-sites-pages) - [Squarespace Help Center, Using code injection (April 2026)](https://support.squarespace.com/hc/en-us/articles/205815908-Using-code-injection) - [Squarespace Help Center, Request that AI models exclude your site (January 2026)](https://support.squarespace.com/hc/en-us/articles/360022347072) - [Wix Help Center, Editing your site's robots.txt file (September 2026)](https://support.wix.com/en/article/editing-your-sites-robotstxt-file) --- Title: We rank first on Google. Why does ChatGPT not mention us? URL: https://fortitudemedia.ai/insights/google-rankings-wont-protect-you Author: Ashley Kenerson Published: 2026-03-31 Last reviewed: 2026-09-24 Topic: How buying is changing Short answer: Ranking helps but does not guarantee a mention. ChatGPT search, Copilot and Google's AI features start from a search index, so a strong position gets your page read. The assistant then names the firms whose pages answer the question plainly and whose claims other sources confirm. In March 2026 only 38% of AI Overview citations came from top-ten pages, so answer the follow-up questions too, check you rank on Bing, and rewrite your ranking pages as answers. Your marketing manager has a slide that says you rank first for the terms that matter. It took three years and a good deal of money to get there, and it still brings enquiries. Then a prospect mentions, almost in passing, that they asked ChatGPT for a shortlist and your firm was not on it. A competitor you regard as smaller was. [ID-Pal](/work/id-pal), the identity verification firm, found itself in a version of this in May 2026. Across the questions its buyers actually ask, the assistants knew who it was and still passed it over: "AI could describe ID-Pal. It never recommended it." After pages were rebuilt to answer those questions, the count moved: "UK buyer questions where AI names ID-Pal: none in May, two in August, four in September 2026." Google's results page ranks pages; AI assistants answer questions and name firms. Ranking helps, because most assistants start from a search index, but it is no guarantee. [Ahrefs found in March 2026](https://ahrefs.com/blog/ai-overview-citations-top-10) that only 38% of AI Overview citations came from pages ranking in the top ten, down from 76% in July 2025, because Google now runs several related searches for one question and cites pages from all of them. ## Key takeaways - A Google ranking gets your page read by AI assistants; whether you are named depends on how clearly it answers the question and whether other sources agree. - Only 38% of AI Overview citations came from top-ten pages in Ahrefs' March 2026 data, so answering the follow-up questions matters as much as ranking for the main one. - ChatGPT search has drawn on Bing and Microsoft Copilot runs on it, so a firm that is weak or missing on Bing may never be fetched by either. - Pages built to win a keyword often read as vague or salesy to an assistant, and lose to a plainer page. - Links still matter, because they shape which pages a search index surfaces for the assistant to read in the first place. - Measure rankings and mentions side by side, on a fixed list of buyer questions. ## Why does ChatGPT recommend my competitor when we rank higher on Google? Because a ranking is a position in a list of links, and a recommendation is a name in a sentence. The assistant uses search results to find pages, then decides for itself which firms it can confidently name, and a clearer page from a lower-ranked firm can win that decision. Ask Google for "employment law solicitors Manchester" and you get a list to choose from. Ask ChatGPT "which firms should I talk to about an employment tribunal in Manchester" and it searches, reads what it fetches alongside what it already knows, and names three or four firms with reasons. The assistant has already chosen. | | A Google ranking | Being named by an AI assistant | |---|---|---| | What you get | A position in a list of links | Your name in a sentence, with a reason | | Who chooses | The searcher, from the list | The assistant, before the buyer sees anything | | What it starts from | Google's index and its ranking signals, links among them | A search index (Google's or Bing's for most assistants), across the question and its follow-ups | | What decides the outcome | How well the page matches the query and how far other sites vouch for it | Whether the page answers the question plainly and whether other sources say the same | So a firm can hold the top position and not be named, while a firm at position seven with a clearer site and a couple of trade press profiles is named instead. ## What are AI assistants actually reading about you? They read your website, the sites that mention you, the reviews and directories that describe you, and anything your named people have published, and they look for agreement. They have two sources: training data, a snapshot of the web taken at some point in the past, and live retrieval, where ChatGPT search, Copilot, Perplexity, Gemini and Google's AI features fetch current pages using crawlers that behave differently from Google's (see [how AI crawlers read your website](/insights/ai-crawlers-vs-google-spiders)). Links still count at the retrieval stage. Live retrieval starts from a search index, and links help decide which pages that index surfaces for the assistant to read. An [SE Ranking study reported in November 2025](https://www.searchenginejournal.com/new-data-top-factors-influencing-chatgpt-citations/561954/) found the number of referring domains was the strongest predictor of whether ChatGPT cited a page, and [AirOps found in April 2026](https://searchengineland.com/chatgpt-citations-ranking-precision-length-study-474538) that pages in the top position of the search results ChatGPT retrieved were cited 58.4% of the time, against 14.2% for pages in position ten. Once it has the page, the assistant is trying to establish a few facts: - What does this firm do, in plain words, and for what kind of client? - Where does it operate and at what scale? - Who are the people, and what are they known for? - Do other sources agree, or does the picture contradict itself? - Is the page an honest answer to the question, or a page built to catch a keyword? A page that ranks but fails those tests gets read and not named. There is more on the underlying judgement in [what E-E-A-T is and why AI cares about it](/insights/eeat-why-ai-cares-more). ## Does ranking on Bing matter for ChatGPT? Yes, more than most firms assume. OpenAI says ChatGPT search uses third-party search providers, Bing has been reported as one of them, and a small [Seer Interactive test in February 2025](https://www.seerinteractive.com/insights/87-percent-of-searchgpt-citations-match-bings-top-results) found 87% of ChatGPT search citations matched Bing's top results. Microsoft Copilot, which many buyers have at work, runs on Bing. A firm that ranks well on Google but is thin or missing on Bing may never be fetched by ChatGPT search or Copilot. | Assistant | Where it starts looking | Crawler that must be allowed | |---|---|---| | Google AI Overviews, AI Mode, Gemini | Google's index | Googlebot | | ChatGPT search | Third-party search providers, reportedly including Bing | OAI-SearchBot | | Microsoft Copilot | Bing | Bingbot | | Perplexity | Not published in detail | PerplexityBot | | Claude | Not published in detail | Claude-SearchBot | Three checks cover most of it. Verify the site in Bing Webmaster Tools, submit your sitemap and see which key pages Bing has indexed. Search Bing for your three most important buyer questions. Then check that robots.txt and your CDN let OAI-SearchBot and Bingbot in: [OpenAI says](https://help.openai.com/en/articles/9237897-chatgpt-search) a site must allow OAI-SearchBot to appear in ChatGPT search, and [Cloudflare changed its default](https://blog.cloudflare.com/content-independence-day-no-ai-crawl-without-compensation/) to block AI crawlers in July 2025, so check the setting. ## Why do keyword-built pages read badly to an assistant? Because the things that win a ranking and the things that make a clear answer are different, so the pages built hardest for position are often the least likely to be cited. For example, take a typical ranking page: 2,500 words on "cloud backup for accountancy firms", the phrase in every heading, an introduction that circles the topic for three paragraphs, a generic list of benefits and a call to action every few hundred words. Years of link building keep it at the top. Now read it as an assistant would, trying to answer "who provides cloud backup for a twelve-partner accountancy practice in the North West". The page does not say who the firm serves or at what size. It does not say what makes its approach different. It does not name a person. It reads as marketing. The assistant moves on to a competitor's page that says, in plain sentences, "we provide backup and disaster recovery for accountancy practices with five to fifty staff, mostly in the North West, and here is how a typical engagement runs". The second page might never rank first. It answers the question, so it gets named. The same pattern explains [why ChatGPT never quotes most blogs](/insights/why-ai-ignores-most-blog-posts). To see this for your own firm, record the same things for each buyer question: | Column | What to record | Why it matters | |---|---|---| | Buyer question | The question in the buyer's own words | Assistants answer questions | | Google and Bing position | Your position on each | Whether you are likely to be fetched | | Named or not | Yes or no in ChatGPT, Copilot, Gemini, Perplexity and the AI Overview, over several runs | Answers vary between runs | | Who was named instead | The firms the assistant chose | Your real competitors | | What their page says | Their first hundred words | Usually the reason they won | ## Do AI Overviews cite pages outside the top ten results? Yes, most of the time. In [Ahrefs' March 2026 data](https://ahrefs.com/blog/ai-overview-citations-top-10), 38% of AI Overview citations came from top-ten pages, 31.2% from positions 11 to 100 and 31.0% from beyond the top 100. The reason is what Google calls query fan-out. [Google explains](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide) that its AI features generate a set of related searches to answer one question, then draw on pages from all of them. A page that answers a follow-up question well can be cited even if it ranks nowhere for the main search. That is why a competitor who has written plain answers to the questions buyers ask next can appear in the AI Overview above your first-place link. AI Mode goes further, as we explain in [what Google AI Mode is and how it differs](/insights/end-of-ten-blue-links). ## What are the signs that AI is already costing you enquiries? Look for these in the monthly report, and add "an AI assistant" as an option on your enquiry form: - Rankings are stable but visits are falling. - Click-through rates for the same positions are lower than a year ago. - Enquiries from search are fewer, or weaker, without an obvious cause. - Prospects say they "looked you up" in ways your analytics cannot explain. If two or more are true, the ranking is already worth less than the slide suggests. We cover the results page itself in [why traffic falls when rankings have not](/insights/zero-click-search-your-business). ## What actually protects you? Being the clearest, best evidenced answer to the questions your buyers ask. ### Expertise that is specific to you A method, a point of view, original research or a very specific description of who you serve is hard to substitute. Be narrow enough that only you could have written it, and put a named person's name to it: assistants attach expertise to people as well as firms, and "the team" is not someone they can recognise. ### Evidence from other people Trade press profiles, directories, reviews, awards and case studies with named clients tell an assistant that someone other than you says you are what you claim. Consistency matters more than volume: the same facts, in the same words, across several sources. ### A website that answers questions Every service page should say, in its first hundred words, what you do, for whom, where and at what scale, with the key text in the page's HTML so crawlers that do not run JavaScript can read it. ID-Pal's work followed this pattern: "Five industry pages rebuilt: fund administration, wealth management, legal, accountancy and estate agency. Each opens with the question a compliance lead types, names the regulator, and answers inside the first ninety words." The result, in the case study's words: "Where a page has been rebuilt, ID-Pal appears. Where it has not yet, it does not, which tells us exactly what to build next." ### Channels you own A referral network and a direct relationship with your market do not depend on how strangers find you. Researched outbound, where a person writes and approves every message, reaches buyers before any assistant is asked. ## What should we do with the rankings we already have? Use them to guide the shift. Your rankings tell you which questions your market asks and which of your pages the search engines already trust, and those are the pages to rewrite first. ### Step 1: Rewrite your best-ranking pages as answers Put the direct answer at the top, name the people behind the work and link to the evidence. Keep the foundations that serve both systems: fast pages, clean structure, a sitemap and no crawler blocks. ### Step 2: Answer the follow-up questions List the questions a buyer asks after the first one, such as cost, timescale, alternatives and risks, and make sure a page on your site answers each plainly. That is how you get cited through query fan-out. ### Step 3: Earn coverage instead of buying links Stop paying for link volume and put the money into coverage that says something specific about you, in titles your buyers read. ### Step 4: Measure both lines Keep rankings and visits on one line and mentions on another. Google's Search Console added [generative AI performance reports](https://developers.google.com/search/blog/2026/06/gen-ai-performance-reports) for AI Overviews and AI Mode in June 2026, available to all sites by 31 August 2026. For the other assistants, run a fixed list of 20 to 50 buyer questions several times a month in a clean session, and count a win only when it repeats. ID-Pal's measurement works the same way: "The same buyer questions run every month in ChatGPT, Gemini and Claude, in a fresh chat. A win only counts when it repeats, and losses are reported too." ## What to do this month 1. Verify your site in Bing Webmaster Tools, submit your sitemap and check that robots.txt and your CDN allow Bingbot and OAI-SearchBot. 2. Open your monthly search report and look for the symptoms: stable positions, falling visits, lower click-through. 3. Rewrite the first hundred words of your three highest-ranking service pages so they state plainly what you do, for whom, where and at what size. 4. Build the table above for twenty buyer questions, and note who is named where you are not. 5. List the people in the firm whose names should be attached to its expertise, and check what the web currently says about each. 6. Move any spend that exists purely to buy links towards one piece of coverage in a title your buyers read. ## Frequently asked questions ### Should we keep paying for SEO if ChatGPT ignores rankings? Yes, because ChatGPT does not ignore rankings: ChatGPT search, Copilot and Google's AI features all start from a search index, so being found there is the first step to being named. What to stop is chasing positions for their own sake with keyword-built pages and bought links. Our guide to [whether you still need SEO now buyers ask ChatGPT](/insights/seo-vs-llm-optimisation) sets out which parts to keep. ### Which pages on my website does ChatGPT actually read? OAI-SearchBot crawls pages for ChatGPT search, and ChatGPT-User fetches a page when a user's request needs it. Your server logs show which pages each crawler requests. The main AI crawlers do not run JavaScript, so text that only appears after scripts run may never be read. For clicks, look for visits referred from chatgpt.com in your analytics, and note which pages they land on. ### Can one page be written for both Google and the assistants? Mostly, yes. A page that answers the buyer's question directly, names the people involved and links to evidence tends to do well in both. Drop the habits of ranking-first writing, such as keyword repetition, long introductions and heavy sales language, and the same page serves both. ### We are a small firm. Does this help us or hurt us? It can help. Being named rewards clarity and specific expertise, which a small firm can have in abundance. A four-partner practice that says plainly what it does and for whom can be named ahead of a national firm whose site says everything and therefore nothing. ### How long does it take to start being named? Assistants that search the web live, such as ChatGPT search, Perplexity, Copilot and Google's AI features, can reflect a new page or article within weeks; answers drawn from a model's own memory change only when it is retrained, which takes months. ID-Pal's case study records it: "Claude names it for UK fund administrators, on a page rebuilt in August, and links NorthRow to ID-Pal in the same answer." Measure monthly rather than weekly. A first-place ranking was hard won and it is still worth something: it gets you read. It no longer guarantees you are the firm the assistant names. To see the gap for yourself, [book a free Growth Review](/growth-review): 30 minutes with a founder, with your site, two competitors and what Google and ChatGPT say about all three checked beforehand, and three fixes in writing afterwards. ## Sources - [Ahrefs, AI Overview citations from top-ten pages fell to 38% (March 2026)](https://ahrefs.com/blog/ai-overview-citations-top-10) - [Search Engine Land, AirOps study of ChatGPT citations, ranking and page length (April 2026)](https://searchengineland.com/chatgpt-citations-ranking-precision-length-study-474538) - [Search Engine Journal, New data: top factors influencing ChatGPT citations (November 2025)](https://www.searchenginejournal.com/new-data-top-factors-influencing-chatgpt-citations/561954/) - [Seer Interactive, 87% of SearchGPT citations match Bing's top results (February 2025)](https://www.seerinteractive.com/insights/87-percent-of-searchgpt-citations-match-bings-top-results) - [OpenAI Help Center, ChatGPT search (checked September 2026)](https://help.openai.com/en/articles/9237897-chatgpt-search) - [Google Search Central, Optimizing your website for generative AI features on Google Search (July 2026)](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide) - [Google Search Central Blog, Generative AI performance reports in Search Console (June 2026)](https://developers.google.com/search/blog/2026/06/gen-ai-performance-reports) - [Cloudflare, Content Independence Day: no AI crawl without compensation (July 2025)](https://blog.cloudflare.com/content-independence-day-no-ai-crawl-without-compensation/) --- Title: What are Google AI Overviews and what do they mean for my firm? URL: https://fortitudemedia.ai/insights/google-ai-overviews-changing-search Author: David Adams Published: 2026-03-31 Last reviewed: 2026-09-24 Topic: Getting named by AI assistants Short answer: A Google AI Overview is an answer Google writes at the top of the results from several websites, with the sources credited beside it; AI Mode is the conversational version. When one appears, people click ordinary results about half as often, but cited pages earn far more clicks than uncited ones. Google credits pages that are indexed, quotable and answer the follow-up questions, so open each section with a clear claim and track who is credited for your buyers' questions. Somebody on your team has noticed it already. Google a question about your sector and, before any of the familiar results, there is a paragraph or two written by Google itself, with a few source links beside it. Your rankings report still says page one. Your traffic report says fewer people are arriving. Both are true at the same time. That paragraph is a Google AI Overview: Google writing an answer from several websites instead of simply listing them. It changes clicks. In [Pew Research Center's study of US browsing in March 2025](https://www.pewresearch.org/short-reads/2025/07/22/google-users-are-less-likely-to-click-on-links-when-an-ai-summary-appears-in-the-results/), people clicked a traditional result on 8% of visits when an AI summary appeared, against 15% when none did. For a founder-led B2B firm, the question to ask is no longer only "where do we rank?" but "are we one of the sources Google names when it writes the answer?" This guide explains what AI Overviews and AI Mode are, how Google picks the pages it credits, what they do to B2B traffic, how to see them in Search Console, and what to change. For why ranking alone no longer protects you, see our guide on [ranking first on Google but not being mentioned by ChatGPT](/insights/google-rankings-wont-protect-you). ## Key takeaways - A Google AI Overview is an answer Google writes at the top of the results from several websites, with the sources credited beside it; AI Mode is a separate, conversational search built the same way. - Both run several related searches for one question, so pages that answer the follow-up questions get credited, not only pages ranking for the original search. - When an Overview appears, people click ordinary results about half as often (Pew, 2025), but being cited inside it brings far more clicks than not being cited (Seer Interactive, 2026). - Google says there are no special requirements or markup for AI features: pages need to be indexed and eligible for a snippet. - Pages are credited when each section opens with a clear, specific claim that other sources do not already make. - Search Console now reports AI Overview and AI Mode impressions separately, and a setting lets you opt out of AI features without affecting ordinary rankings. ## What is a Google AI Overview? An AI Overview is a summary Google writes and places above the normal results, composed from several pages, with the pages it drew on shown as links beside the text. It is different from the older featured snippet, which lifted a sentence or two from one page. AI Overviews reached the UK in August 2024. They do not appear on every search. Pew found an AI summary on 18% of the searches in its March 2025 sample, and on 60% of searches that started with a question word such as "how" or "what". [Semrush's tracking of more than 10 million keywords](https://searchengineland.com/google-ai-overviews-surge-pullback-data-466314), reported by Search Engine Land in December 2025, found Overviews on 6.5% of queries in January 2025, just under 25% at the July peak and under 16% by November. Both studies are American, and the share moves month to month. ## What is the difference between AI Overviews and AI Mode? AI Overviews sit on top of ordinary results and answer in a paragraph. AI Mode, launched in the UK in July 2025, is a separate tab that works like a conversation: the buyer asks a longer question, gets a fuller answer and asks follow-ups. [Google says](https://blog.google/around-the-globe/google-europe/united-kingdom/ai-mode-search-uk/) early UK users typed AI Mode questions two to three times longer than traditional searches. Both features, in [Google's own words](https://developers.google.com/search/docs/appearance/ai-features), "may use a 'query fan-out' technique, issuing multiple related searches across subtopics and data sources", which lets them show a wider set of links than classic search. | | AI Overviews | AI Mode | |---|---|---| | Where it appears | Above the normal results, on some searches | A separate tab, chosen by the user | | Format | A short written answer with source links | A longer, conversational answer with follow-up questions | | Typical question | Short or medium searches, often "how" or "what" | Longer, more detailed questions | | Sources | Several credited pages | Usually a wider set, from more related searches | | How to be credited | Be indexed, eligible for a snippet and clearly quotable | The same, plus answers to the follow-up questions a buyer asks next | The same pages are eligible for both, because both draw on Google's index and ranking systems. We have not seen a reliable study comparing which pages each credits. Our guide to [what AI Mode is and how it differs from AI Overviews](/insights/end-of-ten-blue-links) covers AI Mode in more depth. ## How does Google decide which pages to credit? Google starts with its normal judgement of which pages are relevant and trustworthy, then uses the passages it can lift cleanly. Google says there are "no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary": a page must be indexed and eligible to show with a snippet. Ranking still helps, but ranking for the original search is no longer the whole story. [Ahrefs found in March 2026](https://ahrefs.com/blog/ai-overview-citations-top-10) that 38% of AI Overview citations came from pages in the top ten for that search, down from 76% in July 2025. The shift reflects the related searches Google now runs for each question, which is why covering the follow-up questions matters as well as ranking. From what we see across client sites, the pages that get credited share five traits: - Each section makes one clear claim near the top, so the point can be lifted without the rest. - Proper headings, short lists and, where they help, tables. - Something the other sources do not already say. Google has little reason to credit five pages making the same point. - Visible dates and current facts, on subjects where facts change. - Depth across several connected pages, not one isolated post. For example, imagine a firm of chartered surveyors with a long, well-written page on dilapidations that opens with three paragraphs of history before it says what a landlord can and cannot claim. Google may find nothing it can quote cleanly and credit a competitor with a shorter, blunter page instead. A real version of the fix comes from ID-Pal, an identity verification firm we work with: "Five industry pages rebuilt: fund administration, wealth management, legal, accountancy and estate agency. Each opens with the question a compliance lead types, names the regulator, and answers inside the first ninety words." The [ID-Pal case study](/work/id-pal) reports what changed in AI answers afterwards. ## What do AI Overviews do to your website traffic? Fewer people click through when an Overview answers the question, and the drop depends on the kind of question. Being one of the credited sources softens it considerably. Pew found clicks on traditional results fell from 15% to 8% of visits when a summary appeared, and people clicked a link inside the summary on only 1% of visits. [Seer Interactive's April 2026 study](https://www.seerinteractive.com/insights/aio-impact-on-google-ctr-2026-update) of 53 brands found that pages cited inside an Overview earned about 120% more organic clicks per impression than pages that were not cited, though still fewer than on searches with no Overview at all. | Kind of question | Example | Likely effect on clicks, in our experience | |---|---|---| | How-to and definitions | "How do I serve a Section 21 notice?" | Heaviest loss: the Overview is often the whole answer | | Comparison and research | "Defined benefit versus defined contribution for a small employer" | Moderate: buyers still want to read further | | Supplier selection | "Employment law firms for manufacturers in the Midlands" | Smallest: buyers still want to see the firms | For a B2B firm this is mixed news. The early research your buyers do increasingly happens without a visit to anyone's site, so whoever Google credits shapes how the buyer thinks about the problem. The later stage, where they choose who to talk to, still produces clicks, and those clicks are worth more. Our guide to [why traffic falls while rankings hold](/insights/zero-click-search-your-business) covers what to measure instead of raw visits. ## Do AI Overviews appear for B2B service searches in the UK? Yes, and increasingly for commercial searches, not just "how to" questions. In the Semrush data, the share of Overview-triggering queries that were informational fell from 91% in January 2025 to 57% in October, while commercial queries rose from 8% to 18%. We have not found a UK-only study of how often Overviews appear on B2B service searches. The reliable answer for your firm is to search your own buyers' questions from a UK connection and record what appears. ## How do you get your firm cited in AI Overviews and AI Mode? Write every page so a hurried reader could find the answer in the first two sentences of each section, then build the reasoning underneath. That single habit does more than anything else. ### Write for extraction Use headings that state the question, open each section with the claim, and put the supporting detail after it. Compare: "Marketing automation has evolved significantly, and today's platforms offer sophisticated multi-channel orchestration across email, SMS and push notifications from a unified interface." Against: "Modern marketing automation platforms do three things: send across email, SMS and web from one place; trigger messages from customer behaviour; and report on which channel produced the result." The second is the one Google can lift. ### Answer the follow-up questions Because Google runs several related searches for each question, publish a connected set of pages rather than one enormous guide: a definition, a comparison, the cost or return, a sector version and a how-to. Each can be credited for a different part of the answer. Our guide to [organising content so AI assistants see you as the expert](/insights/topic-clusters-ai-understands) explains how, and why a page for every phrasing breaks Google's spam policies. ### Say something the other sources do not If every page on choosing a payroll provider says "it depends on your needs", Google uses that point once and moves on. Offer a framework, a scoring method or a ranked list of questions to ask. Comparison pages work well here. For Helm Club, a founder peer group, "Every comparison page we have published ranks first for its question on Google and the AI engines. The pages not yet built are the ones where a competitor still writes the answer." See the [Helm Club case study](/work/helm-club). ### Keep structured data accurate, and keep pages current Structured data helps Google read your facts correctly, but Google says its AI features need no special markup, so treat it as housekeeping, not a ranking trick. Our guide to [schema markup for AI assistants](/insights/structured-data-schema-markup-ai) covers what to add. Show the date each page was reviewed, and update it when a regulation, price or product it mentions changes. ## How do you see AI Overview traffic in Search Console? Use the Search generative AI performance reports, which [Google added to Search Console](https://developers.google.com/search/blog/2026/06/gen-ai-performance-reports) from 3 June 2026 and rolled out to all sites by 31 August 2026. They show impressions for AI Overviews, AI Mode and Discover's AI features, broken down by page, country and device. The ordinary Performance report still counts AI feature traffic within the overall "Web" figures, and in Google Analytics visits from AI Overviews and AI Mode appear as ordinary organic search. So the new reports are the only place Google separates them out. Third-party tools, such as Ahrefs Brand Radar, now track which sites are cited in AI Overviews and AI Mode across large sets of prompts, which is useful for trends and competitor comparison. To track your own market, keep a fixed list of 20 to 50 buyer questions, search each several times a month in a clean browser, and record whether an Overview appeared, who was credited and whether you were. Our guide to [finding out whether ChatGPT and Google recommend your firm](/insights/track-ai-recommending-your-business) sets out the routine. ## Can you stop Google using your content in AI Overviews without losing rankings? Yes. Search Console has a Search generative AI control, under Settings, that removes your content from AI Overviews, AI Mode and Discover's AI features. [Google says](https://support.google.com/webmasters/answer/16908024?hl=en) it "isn't used as a ranking or inclusion signal affecting other parts of Search", and exclusion generally takes a few days. For finer control, Google's preview settings apply to its AI features: nosnippet, data-nosnippet (to protect a single passage) and max-snippet limit what Google can show from a page, without blocking Googlebot. There is no separate crawler for AI Overviews, and blocking Google-Extended, which governs use in Gemini training, does not remove you from them. For most B2B firms, opting out gives up a place where buyers form their shortlist, so fix what Google says about you rather than withdrawing. ## What to do this month 1. Take your five most-visited service or advice pages and rewrite the opening of each section so the answer comes first. 2. Open the Search generative AI report in Search Console and note which pages already earn AI Overview or AI Mode impressions. 3. Search 20 questions your buyers ask and record who Google credits in each Overview and in AI Mode. That is your baseline. 4. Choose one subject you want to be credited for and plan a connected set of four or five pages around its follow-up questions. 5. Add the date each page was last reviewed, and correct anything on those pages that is no longer true. ## Frequently asked questions ### How is an AI Overview different from a featured snippet? A featured snippet lifts a passage from a single page and shows it in a box. An AI Overview is new text Google writes by combining several pages, with each contributing page credited. Overviews reward pages that make several clear, distinct points, because each can be credited separately. ### How much traffic will we lose? It depends on what your visitors were searching for. Firms whose traffic came from how-to and definition questions see the biggest falls; firms whose traffic comes from comparison and supplier questions see much smaller ones. Check the new Search Console report and your own core questions over several months. ### Can a page outside the top ten be credited? Yes. In Ahrefs' March 2026 data, most AI Overview citations came from pages outside the top ten for the original search, a shift that reflects the related searches Google now runs. Ranking well for the main question still helps. ### Should we block Google's AI crawlers? There is no separate AI Overviews crawler to block. Blocking Googlebot removes you from Google altogether, and blocking Google-Extended affects Gemini training, not Overviews. If you want out of AI features alone, use the Search Console control. Our guide to [how AI crawlers read your website](/insights/ai-crawlers-vs-google-spiders) explains which crawler does what. ### Do links from other sites still matter? Yes. Google's AI features draw on the same index and ranking systems as ordinary results, and links from reputable sites remain part of how Google judges which pages to trust. Mentions and coverage elsewhere also help ChatGPT and the other assistants decide whether to name you. ### Will Overviews replace ordinary search results? Unlikely. Where someone wants a particular website or wants to browse options, Google still shows ordinary results, and the share of searches with an Overview has risen and fallen rather than climbing steadily. Expect them on most explainable questions, which describes much of what your buyers ask early on. AI Overviews and AI Mode are two of the places buyers now get answers, alongside ChatGPT, Copilot and Gemini, and the same well-organised content usually works for all of them. To see who Google and ChatGPT credit for your buyers' questions today, [book a free Growth Review](/growth-review): 30 minutes with David Adams and three fixes in writing. ## Sources - [Pew Research Center, Google users are less likely to click on links when an AI summary appears in the results (July 2025)](https://www.pewresearch.org/short-reads/2025/07/22/google-users-are-less-likely-to-click-on-links-when-an-ai-summary-appears-in-the-results/) - [Search Engine Land, Google AI Overviews surge and pullback: Semrush data (December 2025)](https://searchengineland.com/google-ai-overviews-surge-pullback-data-466314) - [Google, AI Mode in Search comes to the UK (July 2025)](https://blog.google/around-the-globe/google-europe/united-kingdom/ai-mode-search-uk/) - [Google Search Central, AI features and your website (December 2025)](https://developers.google.com/search/docs/appearance/ai-features) - [Ahrefs, AI Overview citations from top-ten pages (March 2026)](https://ahrefs.com/blog/ai-overview-citations-top-10) - [Seer Interactive, AIO impact on Google CTR: 2026 update (April 2026)](https://www.seerinteractive.com/insights/aio-impact-on-google-ctr-2026-update) - [Google Search Central Blog, Search generative AI performance reports in Search Console (June 2026)](https://developers.google.com/search/blog/2026/06/gen-ai-performance-reports) - [Search Console Help, Search generative AI control (September 2026)](https://support.google.com/webmasters/answer/16908024?hl=en) --- Title: What does an article that ChatGPT actually quotes look like? URL: https://fortitudemedia.ai/insights/anatomy-of-article-cited-by-ai Author: Ashley Kenerson Published: 2026-03-31 Last reviewed: 2026-09-24 Topic: Content that gets quoted Short answer: An article AI assistants quote is first found: it ranks in Google or Bing for the question or one of its sub-questions. Then it makes a claim easy to lift: the buyer's words in the title and headings, a direct answer in the opening lines, one topic per section, a figure of your own with its source, a named author and a visible date. A 2026 study found 44% of ChatGPT citations came from the first 30% of a page, so put the answer and the key figure there. Your firm and a competitor both have a competent article on the same subject, and both rank somewhere on Google. When a buyer asks ChatGPT, Perplexity or Google's AI Overviews the question, the competitor is quoted and you are not, and it happens across the site. The difference is rarely the writer's talent. It is how the piece is built. In [Kevin Indig's analysis of 1.2 million ChatGPT answers](https://searchengineland.com/chatgpt-citations-content-study-469483), reported by Search Engine Land in February 2026, 44.2% of citations came from the first 30% of a page, and cited passages were nearly twice as likely to use clear definitions of the "X is" kind. Assistants do not read your article the way you wrote it. They look for a claim they can lift, near the top, stated plainly. This guide takes the article apart element by element, with the evidence for each, a real example and an annotated page to brief a writer against. ## Key takeaways - An assistant can only quote a page it finds, so the article must rank in Google or Bing for the question or a related search. - Put the direct answer and the key figure in the first third of the page: in a 2026 study, 44% of ChatGPT citations came from the first 30% of a page. - Match headings to the words buyers use: in an April 2026 study, heading-to-question match was the strongest on-page factor in ChatGPT citations. - Write definitive sentences and one clear summary line per section; those are the sentences that get lifted. - A specific figure or example of your own, with its source, gives an assistant something the other pages lack. - Studies disagree on ideal length, so write as long as the question needs and no longer. - Add the buyer's follow-up questions because readers need them; one large study found FAQ sections did not raise ChatGPT citations. ## What is an assistant doing when it reads your page? Three things, in order: finding candidate pages, reading them and citing. Most advice skips the first step, where many good articles fail. **Finding.** ChatGPT search, Perplexity, Copilot and Google's AI features run web searches, and Google's often run several related searches for one question, which Google calls query fan-out. Your page has to turn up in those results. In an [AirOps study reported by Search Engine Land in April 2026](https://searchengineland.com/chatgpt-citations-ranking-precision-length-study-474538), pages in the top search position were cited by ChatGPT 58.4% of the time, against 14.2% for pages in position ten. Ranking helps without guaranteeing anything: [Ahrefs found in March 2026](https://ahrefs.com/blog/ai-overview-citations-top-10) that only 38% of AI Overview citations came from top-ten pages, because the fan-out searches reach pages that rank for the sub-questions. Crawlers must also be let in: ChatGPT search needs OAI-SearchBot, and Perplexity asks sites to allow PerplexityBot. **Reading and citing.** The assistant scans the candidates for a clear statement, evidence that it is reliable, and something the other pages lack, favouring the top of the page and plain sentences. When it names a source, it is telling the buyer where a claim came from. So ask of every article: can an assistant find it, and can it lift a reliable, distinctive claim and attribute it to us? ## The nine elements and the evidence behind each | Element | Evidence | Type | |---|---|---| | 1. The buyer's question as the title and headings | Heading-to-question match was the strongest on-page factor (AirOps, April 2026); question-style headings on their own did not help (SE Ranking, November 2025) | Studies, mixed | | 2. The answer in the first paragraph | 44.2% of citations from the first 30% of a page; definitive "X is" phrasing favoured (Kevin Indig, February 2026) | Study | | 3. A short list of key points near the top | Puts the main claims in the first third of the page | Fortitude observation | | 4. One topic per heading | Pages that answered the main query narrowly beat broad guides; 4 to 10 subheadings did best (AirOps) | Study | | 5. Depth: mechanisms, exceptions, trade-offs | Google says non-commodity content with "unique expert or experienced takes" matters most (Google, July 2026); studies disagree on length | Google guidance | | 6. A figure or example of your own | Adding statistics raised visibility in AI answers by about 34%, quotations by about 41% (Princeton and IIT Delhi, KDD 2024) | Study, research engine | | 7. A named author and credible references | Citing sources raised visibility by about 30% (same study); Google stresses first-hand expertise | Study and Google guidance | | 8. Buyer follow-up questions at the end | No citation lift: pages with FAQ sections averaged 3.8 ChatGPT citations against 4.1 without (SE Ranking) | Study; include for readers | | 9. A visible date and last review | AI-cited pages were 25.7% fresher than Google's organic results (Ahrefs, July 2025) | Study | Most of these studies are American, several are run by vendors that sell AI-visibility tools, and all are correlational. Treat them as direction, not law. ## Should the title and headings be questions? They should use the buyer's words, and a question is often the most natural way to do that: "How long does a commercial lease renewal take?" rather than "Lease renewals: our guide". The evidence supports matching over punctuation: AirOps found pages with the strongest heading-to-question match were cited 41.0% of the time against roughly 30% for weaker matches, while an [SE Ranking study of 216,524 pages](https://www.searchenginejournal.com/new-data-top-factors-influencing-chatgpt-citations/561954/), reported in November 2025, found question-style headings on their own averaged fewer citations than plain ones. So write titles and headings in the words buyers type, as questions where that reads naturally. A title that promises "everything you need to know" promises nothing in particular. ## What should the first paragraph do? Answer the question in the first two or three sentences, then spend the rest of the piece on why, how and when not. The most common failure we see is an article that takes four hundred words to reach the point, pushing its best sentence out of the part of the page assistants quote most. Say the question is how long a lease renewal takes. The first paragraph says: typically four to six months from first notice to signing if both sides engage, longer if the landlord disputes terms, and here is what controls the timing. That is quotable. "Lease renewals are an important part of commercial property" is not. Follow the opening with a short list of key points, and end each section with one sentence that states its point cleanly, in the definitive form the Indig study found assistants favour. ## How should the headings be organised? One topic per heading, with the answer in the section's first sentence: "What delays a lease renewal?", "Can the landlord refuse?" Each becomes a unit an assistant can use on its own. Keep the hierarchy clean and sections to a few hundred words; a two-thousand-word section under one heading is as hard to use as no headings at all. AirOps also found pages that answered the main query narrowly outperformed broad, comprehensive guides, so resist cramming three questions into one article. ## How much depth and length does an article need? As much as the question needs. Depth is how expertise shows: a piece that says renewals take four to six months is useful; one that explains which steps run in parallel, what the landlord's surveyor can and cannot do, when the timetable slips and when relocating is cheaper is far more likely to be named. In our experience depth has four parts: the mechanism, the exceptions, the trade-offs, and more than one angle. Length is where the studies disagree. AirOps found pages of 500 to 2,000 words performed best for ChatGPT citations, and pages over 5,000 words were cited less often than pages under 500. SE Ranking found articles over 2,900 words averaged 5.1 citations against 3.2 for those under 800. Neither supports a single word count, so do not set one. [Whether ChatGPT prefers long articles or short ones](/insights/long-form-vs-short-form-ai-prefers) goes through the evidence in full. ## What makes an article distinctive enough to quote? Something the other candidate pages do not have. When two pages are equally well built, the one with a finding, pattern or comparison of its own wins. In a [Princeton and IIT Delhi study presented at KDD 2024](https://arxiv.org/abs/2311.09735), adding statistics raised a page's visibility in AI-generated answers by about 34% and adding quotations from relevant people by about 41%. It was tested on a research engine, so treat it as directional. It needs no research budget. A pattern from your own work, stated with its basis, is enough. Say where every figure came from and when: "in the trade body's 2025 survey" or "across the renewals we handled over the past three years". [Does publishing our own research help ChatGPT recommend us?](/insights/original-research-data-ai-trust) covers how to produce it from what you already have. ## Does the author matter, and how many sources should we cite? Yes, and as many as your claims need. A byline with a name, a role and a line of experience signals that someone accountable knows the subject; "by the team" signals nothing. So do the marks of experience in the text: examples only a practitioner would know and what usually goes wrong. [What makes ChatGPT treat our content as expert rather than marketing](/insights/what-makes-expert-quality-content) sets out those marks. On sources, the Princeton and IIT Delhi study found citing credible sources raised visibility by about 30%, but it set no ideal number, and neither do we. Cite, link and date a source for every figure and every claim a sceptical reader would question. ## Should an article end with FAQs? Yes, for readers, and do not expect a citation boost from the format itself. SE Ranking found pages with FAQ sections averaged 3.8 ChatGPT citations against 4.1 without, and [Google stopped showing FAQ rich results on 7 May 2026](https://www.searchenginejournal.com/google-drops-faq-rich-results-from-search/574429/). The case for a closing set of questions is that buyers have follow-ups, and Google's AI features run related searches for them. A page that answers the next three questions a buyer asks, in their words, serves the reader and gives the fan-out searches something to find. [Do we need schema markup for AI assistants to understand our site?](/insights/structured-data-schema-markup-ai) gives the practical view on marking them up. ## Does the date matter? Yes. [Ahrefs found in July 2025](https://ahrefs.com/blog/do-ai-assistants-prefer-to-cite-fresh-content) that pages cited by AI assistants were on average 25.7% fresher than Google's organic results, with ChatGPT showing the strongest preference for recent pages. Show a publication date, a "last reviewed" date and a line on what changed, and change the content when you change the date. [Does ChatGPT prefer newer content?](/insights/content-freshness-recency-ai) covers how often to refresh. ## What does a real cited page look like? Our work for [ID-Pal](/work/id-pal), an identity verification firm, is the clearest example we can share. The starting point: "Across 36 recorded AI answers to the questions ID-Pal's buyers actually ask, ID-Pal was named three times, and only for its brand or its Salesforce integration. AI knew who it was, and had no reason to recommend it." The rebuilt pages follow the shape in this guide: "Five industry pages rebuilt: fund administration, wealth management, legal, accountancy and estate agency. Each opens with the question a compliance lead types, names the regulator, and answers inside the first ninety words." That is element 1 (the buyer's question), element 2 (the answer first) and element 7 (the authority named) in one sentence. The result: "Gemini now names ID-Pal for UK wealth managers. Claude names it for UK fund administrators, on a page rebuilt in August, and links NorthRow to ID-Pal in the same answer." Overall, "UK buyer questions where AI names ID-Pal: none in May, two in August, four in September 2026." ## An annotated page, element by element For example, a commercial property firm's article on lease renewals, built to the nine elements: | Part of the page | What it says | Element | |---|---|---| | Title | "How long does a commercial lease renewal take?" | 1 | | Byline and dates | Partner's name, role and years in practice; published and last reviewed dates | 7, 9 | | First paragraph | "Typically four to six months from first notice to signing if both sides engage..." | 2 | | Key points | Five one-sentence takeaways | 3 | | Section heading | "What delays a lease renewal?" opening with the answer | 4 | | Section body | The steps, what the landlord's surveyor can do, when relocating is cheaper | 5 | | Own finding | A pattern from the firm's own renewals, with its basis and date | 6 | | References | The statute and a dated trade body survey, linked | 7 | | Closing questions | "Can the landlord refuse to renew?" and four more follow-ups | 8 | ## How do I get my article cited in Google AI Overviews? Be indexed, eligible for a snippet and useful for the question or one of its sub-questions. Google says [its AI features need no special markup or files](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide), and that separate pages for every variation of a query, made mainly to manipulate rankings, break its scaled content abuse policy. So build one strong page per real question, cover the follow-ups, and put the answer where it can be lifted. [What Google AI Overviews mean for your firm](/insights/google-ai-overviews-changing-search) covers the wider picture. ## What makes Perplexity choose a source to cite? Perplexity [says it searches the internet in real time](https://www.perplexity.ai/help-center/en/articles/10352895-how-does-perplexity-work) when asked a question and gives each answer numbered citations. Its crawler, PerplexityBot, "is designed to surface and link websites in search results on Perplexity", so [Perplexity recommends allowing it](https://docs.perplexity.ai/guides/bots) in robots.txt. Beyond that, the same elements apply, and community sites weigh heavily: in [Profound's 2025 study of 680 million citations](https://www.tryprofound.com/blog/ai-platform-citation-patterns), Reddit made up 46.7% of Perplexity's top ten sources, so a firm discussed on relevant forums and review sites has more ways in. ## What to do this month 1. Score your three most important service articles against the nine elements. 2. Search Google and Bing for each article's question and two follow-ups, and note where each page ranks. 3. Rewrite the first paragraph of each to answer the question directly, with the key figure inside the first third. 4. Add one specific figure or example from your own work to each, with its source and date. 5. Add a byline, a last-reviewed date and the three follow-ups buyers ask most. 6. Add each article's question to a fixed list of buyer questions and check it monthly in ChatGPT, Perplexity and Google AI Mode. ## Frequently asked questions ### Does the article need original research to be quoted? No, but it needs something distinctive. A pattern from your own work, a comparison nobody else has assembled or a clear statement of where standard advice fails is enough. What does not work is a page that could sit on a competitor's site unchanged. ### Do AI assistants prefer lists and tables over paragraphs? They prefer whatever states the point most clearly. Use a list for list content, a table when options are compared and prose where the reasoning is the point. Google says there is no need to break content into tiny pieces for AI, and we have seen no reliable study showing a blanket preference for lists. ### Should we write differently for AI assistants than for people? Mostly no. What helps an assistant helps a reader: the answer first, one topic per section, plain headings, examples, honest limits. The one difference is placement. People may read to the end, but assistants quote the top of the page most, so nothing important should wait until the final third. ### Can a mediocre article on a big firm's site beat our excellent one? Sometimes. SE Ranking found links from other websites were the strongest predictor of ChatGPT citations, which favours established sites. But a more specific page on a smaller site can win on specialist questions, and the Princeton and IIT Delhi study found lower-ranked pages gained most from these changes. [Why most blog posts never get quoted](/insights/why-ai-ignores-most-blog-posts) covers the common failures. ### How long before a rewritten article gets cited? Assistants that search the web live can pick up a rewritten page within weeks; answers from a model's memory change only when it is retrained, which takes months. Count a win only when it repeats. If you would like to know how your articles compare with your competitors' today, [book a free Growth Review](/growth-review): 30 minutes with David Adams and three fixes in writing. Writing the expert content that gets a firm named is part of [Reputation That Wins](/reputation-that-wins). ## Sources - [Search Engine Land, Kevin Indig's study of 1.2 million ChatGPT answers (February 2026)](https://searchengineland.com/chatgpt-citations-content-study-469483) - [Search Engine Land, AirOps study of ChatGPT citations, ranking and length (April 2026)](https://searchengineland.com/chatgpt-citations-ranking-precision-length-study-474538) - [Search Engine Journal, New data: top factors influencing ChatGPT citations (November 2025)](https://www.searchenginejournal.com/new-data-top-factors-influencing-chatgpt-citations/561954/) - [Ahrefs, AI Overview citations from top-ten pages (March 2026)](https://ahrefs.com/blog/ai-overview-citations-top-10) - [Ahrefs, Do AI assistants prefer to cite fresh content? (July 2025)](https://ahrefs.com/blog/do-ai-assistants-prefer-to-cite-fresh-content) - [Aggarwal and others, Princeton and IIT Delhi, study of visibility in AI-generated answers (KDD 2024)](https://arxiv.org/abs/2311.09735) - [Google Search Central, Optimizing your website for generative AI features on Google Search (July 2026)](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide) - [Perplexity, Perplexity crawlers (accessed September 2026)](https://docs.perplexity.ai/guides/bots) --- Title: What does ChatGPT tell candidates about working at my firm? URL: https://fortitudemedia.ai/insights/ai-search-recruitment-talent Author: Ross Williams Published: 2026-03-31 Last reviewed: 2026-09-24 Topic: Press, reviews and being trusted Short answer: Candidates ask ChatGPT, Gemini, Claude, Copilot and Perplexity whether your firm is a good place to work, what the pay is like and what to expect at interview. The assistants answer by summarising review sites such as Glassdoor, LinkedIn, news coverage, public records, your careers page and what your leaders publish, and they report the patterns. To change the answer, fix the causes of repeated complaints first, then invite honest reviews from all staff, put your leaders in public and make your careers page specific. Your recruiter has finally found the operations director you have spent six months trying to hire. The message goes out on a Tuesday. By Thursday there is no reply, and there never will be. One likely reason: the candidate typed your firm's name into ChatGPT, asked whether it is a good place to work, read the answer and moved on. For years, employers controlled most of what a candidate saw: the advert, the careers page, the recruiter's pitch, the interview. Candidates could read Glassdoor, but few read all of it. Now an assistant reads it for them, alongside LinkedIn, the trade press, Companies House and whatever your leaders have said in public, and turns it into a few paragraphs. ChatGPT alone has more than a billion weekly users ([OpenAI, August 2026](https://openai.com/index/expanding-access-to-ai-with-chatgpt-ads/)), and LinkedIn, where your staff's tenure and your leaders' posts sit, appears in about 11% of AI answers ([Semrush, March 2026](https://www.semrush.com/blog/linkedin-ai-visibility-study/)). The candidate sees the pattern, not the pitch. This guide covers what the assistants read about you as an employer, which signals do the most damage, how to check the answer and how to change it. Most of the work sits outside marketing: whether the firm is a good place to work, and whether that shows. ## Key takeaways - Candidates ask AI assistants about employers before replying to a recruiter, before applying, before interviews and when comparing offers. - The assistants build their answer from review sites, LinkedIn, news coverage, public records, your careers page and what your leaders publish, and they look for patterns across all of them. - The signals that do most damage are a repeated theme in reviews, visible churn, unexplained redundancies, a leadership story that reads badly and a gap between your stated values and what employees say. - You may invite every employee to leave an honest review, but paying for, requiring or drafting reviews breaks the platforms' rules and produces a pattern that is easy to spot. - The durable fix is to deal with the causes of complaints first, then make the evidence public through your staff, your leaders and a specific careers page. - Assistants that search the web can reflect new evidence within weeks, while answers drawn from a model's memory change only when it is retrained, which takes months. ## What do candidates ask ChatGPT about an employer? They ask whether the firm is a good place to work, what the pay and culture are like, whether it is stable and what to expect at interview. The questions change with the moment, and at each one the assistant's summary carries more weight than anything you send. | Moment | Typical question | What the answer leans on | |---|---|---| | Before replying to a recruiter | "Is [firm] a good place to work?" | Review themes, LinkedIn tenure, recent news | | Before applying | "What is the pay like for a [role] at [firm]?" | Salary data on review and job sites, your adverts, your careers page | | Before an interview | "What should I know before interviewing at [firm]? What will they ask?" | Your website, your leaders' posts and articles, interview reviews, trade press | | Comparing offers | "Compare [firm] and [competitor] on culture, progression and stability" | Everything above, for both firms side by side | | After an offer | Friends and family ask the same questions and report back | The same sources, read by someone else | Before an interview, candidates use the assistants as a briefing service: what the firm does, who leads it, recent news and what to ask. Your own material is one input among many, and often the least trusted, because the candidate knows the careers page was written to attract them. The assistants gather evidence slightly differently, as we explain in [which AI assistants people actually use and how they differ](/insights/ai-search-landscape-chatgpt-perplexity-google-claude), but they reach for similar sources. ## What do the assistants read to form their answer? They read review sites, forums, LinkedIn, news coverage, public records, your own careers and about pages and anything your leaders have published, and they weigh patterns across those sources more than any single item. Most of it is written by other people: across all topics, earned media made up 84% of the more than 25 million links cited by ChatGPT, Claude and Gemini in [Muck Rack's May 2026 analysis](https://www.globenewswire.com/news-release/2026/05/07/3290268/0/en/generative-pulse-earned-media-consistently-drives-ai-citations-holding-at-84.html). | Source | Who writes it | What it tells a candidate | |---|---|---| | Glassdoor, Indeed and similar review sites | Current and former staff | Recurring themes on management, pay, workload and direction | | Reddit and other forums | Anyone, often anonymously | Unfiltered stories about working at the firm | | LinkedIn | Your staff, your leavers and your leaders | How long people stay, who left recently, whether anyone posts about their work | | News and trade press | Journalists | Growth, awards, redundancies, disputes, leadership | | Companies House and other public records | Official | Accounts, directors, signs of instability | | Your careers, team and values pages | You | What you claim, and whether it matches everything else | | Your leaders' articles, posts and talks | Your leaders | Whether the firm is led by people who know their subject | Forums and LinkedIn matter more than many founders expect: a [Peec AI study of 30 million sources](https://searchengineland.com/ai-search-engines-cite-reddit-youtube-and-linkedin-most-study-473138), reported by Search Engine Land in March 2026, found Reddit, YouTube, LinkedIn, Wikipedia and Forbes were the most cited domains in AI answers. Because the assistant looks for agreement between sources, contradictions are what it notices most. A values page that promises balance, next to a run of reviews describing sixty-hour weeks, does not average out. It becomes the story. Correcting facts that are simply wrong is covered in [why ChatGPT describes a firm wrongly and how to fix it](/insights/ai-conflicting-information-online). ## Which signals do the most damage? Five signals reliably produce a discouraging summary: a repeated complaint in reviews, visible churn, recent redundancies, a leadership story that reads badly, and a gap between what you say and what employees say. ### A repeated theme in reviews One bad review is noise. Ten reviews over two years that all mention poor management are a pattern, and summaries are built from patterns. The theme matters more than the star rating. ### Visible churn LinkedIn shows how long people stay in each role and where leavers go. If a team turns over quickly, a candidate can see it, and an assistant asked about stability may mention it. A candidate reads that as a prediction about their own future. ### Recent redundancies A restructuring that felt well handled inside the firm reads, from outside, as instability, particularly if the press covered it and the firm never explained it. ### Leadership controversy Anything a founder has said or done that was reported badly can surface when a candidate asks about the leadership. Silence does not make it go away; a later, plain account of what happened gives the assistant a second source. ### The gap between claims and evidence "We invest in our people" with no training anyone has mentioned, or "flexible working" with reviews that say otherwise. The gap is the most informative thing the assistant finds, so the summary is likely to carry it. ## Can we ask our employees to leave reviews? Yes, as long as the request is voluntary, goes to everyone and comes with no reward and no pressure. What the review platforms forbid is paying for reviews, making them a condition of anything, or drafting them for staff; check each platform's current guidelines before you ask. An open invitation often helps, because satisfied people rarely think to write and unhappy leavers usually do. Ask the whole team rather than a hand-picked few. A sudden run of five-star reviews from the people you chose, after years of complaints, looks like what it is to candidates and to the assistants reading them. ## How do I change what ChatGPT says about my company as an employer? Fix the causes of the complaints, then give the assistants more true evidence to read: reviews from a wider share of your staff, leaders writing in public and a specific careers page. Check the answer monthly and give it time to move. You cannot manage your way out of it. The assistants read sources you do not own, and nobody can rewrite three years of reviews or change how long people stay. The underlying work is what makes a firm easier to run anyway: good managers, fair pay, visible progression and candour about the job. ### Step 1: Fix the causes first Read every review from the last three years, on every site, and list the themes. If management comes up, train, support or replace. If pay comes up, check it against the market and adjust or explain. If direction comes up, tell people where the firm is going, in writing, and repeat it. ### Step 2: Invite honest reviews from everyone Once the cause is being dealt with, send one invitation to the whole team to leave a candid review, with no reward and no follow-up pressure. Newer reviews that describe the firm as it is now give the assistant something current to set against the old pattern. ### Step 3: Put leaders in public as experts When an assistant assesses a firm, it looks at who leads it. Leaders who write plainly about their subject and are quoted in the trade press are read as credible, and their firms as stable. A workable rhythm for each leader is one useful LinkedIn post a week and one longer article a month. That matches what the assistants cite: in [Semrush's March 2026 LinkedIn study](https://www.semrush.com/blog/linkedin-ai-visibility-study/), articles made up 50% to 66% of LinkedIn citations, those of 500 to 2,000 words were cited most, and around three-quarters of the cited post authors had posted five or more times in four weeks. Individual members accounted for 59% of LinkedIn citations in ChatGPT search and Google AI Mode, more than company pages, and the median cited post had only 15 to 25 reactions, so a large following is not required. The same writing gets the firm named when buyers ask who to hire, so it pays twice. Our [Reputation That Wins](/reputation-that-wins) service runs this for founders, in their own voice, and [whether you can wait a year before acting on AI search](/insights/first-mover-advantage-ai) explains why starting early matters. ### Step 4: Make your own pages specific Replace adjectives with facts. Publish salary bands if you can. Describe how progression works, with an example, and what the job is like on a hard week. Candidates and assistants both trust specific, slightly unflattering detail over polish. Awards and accreditations help too, when they are real and named, as [whether awards and accreditations make ChatGPT more likely to recommend you](/insights/awards-accreditations-ai-trust) sets out. ### Step 5: Respond to reviews like a person A short, non-defensive reply to a critical review, from a named leader, shows both the candidate and the assistant that someone is listening. The review stays, and the pattern around it changes. ## How do we check what the assistants say about us as an employer? Run a fixed set of candidate questions each month in ChatGPT, Gemini, Claude, Copilot and Perplexity, each in a fresh chat, and record both the answer and the sources each assistant cites. Answers vary from run to run, so treat a theme as real only when it repeats. Five questions are enough to start: is [firm] a good place to work; what is the culture like at [firm]; what is the pay like for a [role] at [firm]; what should I know before interviewing at [firm]; and how does [firm] compare with [competitor] as an employer. Add them to the fixed list of 20 to 50 buyer questions we recommend tracking in [how to find out whether ChatGPT is recommending your firm](/insights/track-ai-recommending-your-business). | What to record | Why it matters | |---|---| | The exact question and the date | So next month's run is comparable | | The assistant, and whether it searched the web | Answers from live search change faster than answers from the model's memory | | The words used to describe you | Hedges and repeated themes show what the evidence says | | The sources cited | Shows which review page, thread or article is shaping the answer | | Themes mentioned: pay, management, stability, progression | Each one maps to a cause you can fix | The sources column is the useful one: if one old thread or news story keeps appearing, you know which evidence needs an answer. ## Does this affect the people we already employ? Yes. Current staff ask the assistants the same questions, about their pay and about whether to stay. When they leave, they update LinkedIn and perhaps Glassdoor, and the next candidate sees a slightly worse picture. The loop runs both ways: a firm people describe well keeps its people. For recruitment firms, where candidates are also part of the product, this cuts twice as deep. We cover being recommended to employers in [why ChatGPT recommends other recruitment agencies](/insights/ai-optimisation-recruitment-agencies), and what we do for recruitment firms is set out on [our recruitment page](/for/recruitment). ## What to do this month 1. Read every review from the last three years, on every site, and write down the recurring themes in one line each. 2. Take the most damaging theme to the leadership team with a plan to fix what causes it. 3. Run the five candidate questions above in ChatGPT, Gemini, Claude and Perplexity, in a fresh chat, and save each answer with the sources it cites. 4. Send one invitation to the whole team to leave an honest review, with no reward and no pressure. 5. Rewrite the careers page so that every sentence states something specific and true, and remove every claim you could not evidence. 6. Agree a posting rhythm with each leader: one useful post a week and one longer article a month. ## Frequently asked questions ### Does Glassdoor affect what AI says about my company? Yes, in our experience. When an assistant searches the web for an employer, review pages are among the first things it finds, and a theme repeated across Glassdoor, Indeed and Reddit is far more likely to reach the answer than any single review. Glassdoor is still one source among several, so a good rating will not outweigh a pattern of departures on LinkedIn or a critical news story. ### Can I get negative Glassdoor reviews removed? Not because they are negative. Review sites remove reviews that break their published guidelines, and you can report one that you believe does, but a fair, critical account of working at your firm will normally stay. The better route is a calm reply from a named leader, a fix for the cause, and newer reviews that describe the firm as it is now. ### Should we publish salary bands on our careers page? We think so, if you can. Candidates ask the assistants about pay, and if you say nothing, the answer comes from salary estimates on review and job sites, which may be out of date or wrong for your roles. A published band, even a wide one, gives the assistant an accurate figure and saves time for candidates who would never accept it. ### Our firm went through redundancies last year. Is that permanent damage? No, but it needs an account. If the only public record is a press story or a run of LinkedIn updates, the assistants will describe instability. A plain statement from the founder about what happened, why and what has changed gives them a second source and a more complete picture, and older coverage fades: [Muck Rack](https://www.globenewswire.com/news-release/2026/05/07/3290268/0/en/generative-pulse-earned-media-consistently-drives-ai-citations-holding-at-84.html) found citations of journalism drop sharply after six months. ### Is any of this a privacy problem for us or our staff? Mostly no, with one caveat. The assistants read public sources, such as reviews people chose to write, profiles people chose to publish and news that was reported. But the staff names, photos and profiles you publish on your own careers and team pages are personal data, and UK GDPR applies to how you handle them, as the [ICO's employment guidance](https://ico.org.uk/for-organisations/uk-gdpr-guidance-and-resources/employment/) sets out. Tell staff what you publish and why, keep it current, and take a profile down when someone leaves or asks. ### How long before the answer changes? Assistants that search the web live, such as ChatGPT search, Perplexity, Copilot and Google's AI features, can reflect new reviews, articles and pages within weeks. Answers drawn from a model's own memory change only when the model is retrained, which takes months. The underlying pattern in reviews takes longest of all, because it moves only as fast as the firm does. The candidate who did not reply on Thursday will never tell you why. The assistant will, if you ask it. If you want help turning your leaders' expertise into the public presence that candidates and buyers both trust, [Reputation That Wins](/reputation-that-wins) is built for that, and a free [Growth Review](/growth-review) is a good place to start. ## Sources - [OpenAI, A milestone in expanding access to AI (August 2026)](https://openai.com/index/expanding-access-to-ai-with-chatgpt-ads/) - [Semrush, LinkedIn AI visibility study (March 2026)](https://www.semrush.com/blog/linkedin-ai-visibility-study/) - [Muck Rack, Generative Pulse: earned media consistently drives AI citations (May 2026)](https://www.globenewswire.com/news-release/2026/05/07/3290268/0/en/generative-pulse-earned-media-consistently-drives-ai-citations-holding-at-84.html) - [Search Engine Land, AI search engines cite Reddit, YouTube and LinkedIn most: study (March 2026)](https://searchengineland.com/ai-search-engines-cite-reddit-youtube-and-linkedin-most-study-473138) - [ICO, Employment information (September 2026)](https://ico.org.uk/for-organisations/uk-gdpr-guidance-and-resources/employment/) --- Title: What does it cost us to leave our website unchanged for years? URL: https://fortitudemedia.ai/insights/hidden-cost-static-website Author: David Adams Published: 2026-03-31 Last reviewed: 2026-09-24 Topic: Websites that bring work in Short answer: More than the hosting bill suggests. An outdated website costs you enquiries from buyers who rule you out, pitches lost to more current competitors, longer sales calls and AI answers that describe the firm you used to be, because ChatGPT and Gemini build their picture of you partly from your site. You can size the cost in an afternoon. A named owner and a monthly routine of one update, one new page and one new piece of proof stops most of it. Your website went live a few years ago. It still loads, the hosting bill is small, and when someone suggests updating it the answer sounds entirely reasonable: "We'll update it later." The trouble is that the costs of an outdated website do not arrive as invoices. They arrive as enquiries that never happen, pitches you lose without knowing why, and AI answers that describe a firm you stopped being two years ago. Most of that damage happens before you know the buyer exists. In [6sense's 2025 Buyer Experience Report](https://6sense.com/newsroom/the-timeline-for-influencing-b2b-buyers-is-shrinking-insights-from-6senses-2025-buyer-experience-report/), a survey of more than 4,000 B2B buyers, buying groups had already ranked their preferred suppliers before first contact in 94% of cases, and bought from that first choice 77% of the time. Part of that ranking now happens inside ChatGPT, Gemini and Copilot, which build their answers partly from your website, and part on the website itself. This guide covers where the costs come from, how to estimate them, which pages to fix first, and how little work it takes to stop them. ## Key takeaways - An outdated website feels cheap because its real costs, lost enquiries, lost pitches and wrong AI answers, never appear on an invoice. - Putting updates off does not save money; it moves the cost out of the budget and into work you never see. - AI assistants describe your firm partly from your website, so out-of-date pages produce out-of-date answers that take weeks or months to correct. - Sales teams feel it first: they stop sending prospects to the site and spend the first call correcting what buyers read. - You can size the cost in an afternoon by date-checking pages, comparing the site with the firm and asking your sales team. - A named owner and a monthly routine of one update, one new page and one new piece of proof keeps most B2B websites current. ## Why does an outdated website feel so cheap? Because the visible costs are small and the invisible ones never appear in a report. Hosting and a maintenance contract are modest. The invisible costs are decisions that went against you: the buyer who chose someone else, the assistant that named a competitor, the candidate who did not apply. "Later" swaps a cost you can see for one you cannot, and a cost you cannot see feels like no cost at all. For example, say a managed IT firm launched its site three years ago and has since added a cyber security service, won two public sector contracts and hired a well-known technical director. None of that is on the website. A buyer comparing it with two rivals sees a smaller, narrower firm, and it loses that buyer before the conversation starts. ## How do I know if my website looks outdated to buyers? Open your website and find the most recent date anywhere on it. Then list three things about your firm that have changed since that date. If you can do it easily, buyers can see it too. The giveaways buyers notice are small: - A newest case study that is two or three years old. - A news or insights page whose latest post is long ago. - Team members who have left, or a senior hire who is missing. - Services described in the language you used before you changed direction. - A copyright line in the footer showing a year that has passed. Together they create a doubt, and a doubt is enough to move your name down the list. We cover the wider set of signals in [what makes a buyer trust a website enough to get in touch](/insights/psychology-of-trust-website-conversions). ## What are the hidden costs? - **Buyers quietly rule you out.** Old case studies and a stale team page suggest a firm less active than the one in the next tab. We cover what buyers expect in [what a B2B website should include](/insights/what-should-a-b2b-website-include). - **AI assistants describe an old version of your firm.** They may describe services you no longer offer, or leave you out of answers about the sector you now focus on. - **Competitors pull ahead.** Rivals who keep adding useful pages give search engines and AI assistants more reasons to name them. - **Recruitment suffers.** In our experience candidates check the site before an interview, and a neglected one suggests a firm that is not growing. See [what ChatGPT tells candidates about working at your firm](/insights/ai-search-recruitment-talent). - **The catch-up bill grows.** Every change put off makes the next one harder, until only a rebuild seems possible. ## What does an outdated website do to your sales team? It makes their job harder and their calls longer. A buyer who found a dated site arrives with questions the website should have answered, and sometimes with a wrong idea that takes half the meeting to correct. Listen for "ignore the website, it's out of date" or "I'll send you our latest credentials separately". A sales team that has built its own PDFs and decks to replace the website is a sure sign the site has stopped doing its job. Quality suffers too: when the site is vague, the wrong people enquire, because nothing helps them rule themselves out. Clear pages on who you work with, how an engagement starts and how long things take do some of that filtering for you. ## Does an outdated website hurt our Google rankings? Not directly. Google does not penalise a site for being old; the problem is relative, because competitors who keep adding useful pages give Google more reasons to show them. Cosmetic freshness does not help either. Google's guidance on [helpful, people-first content](https://developers.google.com/search/docs/fundamentals/creating-helpful-content) warns against changing dates when content has not substantially changed, and says adding or removing content just to seem fresh will not help rankings. AI assistants lean further towards current material. [Ahrefs found in July 2025](https://ahrefs.com/blog/do-ai-assistants-prefer-to-cite-fresh-content), across about 17 million citations, that pages cited by AI assistants were on average 25.7% fresher than Google's organic results, and in a [Seer Interactive study of July 2026](https://www.seerinteractive.com/insights/study-content-recencys-impact-on-ai-visibility-in-2026), 75% of cited pages had been updated in the previous 12 months (four consumer and finance industries, so directional for B2B). Ranking still helps without guaranteeing a mention: [Ahrefs found in March 2026](https://ahrefs.com/blog/ai-overview-citations-top-10) that only 38% of AI Overview citations came from top-ten pages, down from 76% in July 2025, because Google now runs several related searches for one question. We explain how old is too old in [does ChatGPT prefer newer content](/insights/content-freshness-recency-ai). ## How do we get ChatGPT to stop describing services we no longer offer? Correct the sources the assistant reads, starting with your own site, then give it time. Assistants that search the web live (ChatGPT search, Perplexity, Copilot, Google's AI features) can reflect a corrected page within weeks; answers from a model's own memory change only when it is retrained, which takes months. In practice: 1. Rewrite the services and about pages to describe what you sell today. 2. Redirect pages for dropped services to the closest current page, or say plainly you no longer offer them. 3. Make LinkedIn, directory listings and review profiles match the site. 4. Keep your structured data (the machine-readable facts about your firm) consistent with the pages. 5. Re-ask the same buyer questions monthly in a clean session, counting a change only when it repeats. When sources disagree, the old version can outlive the change. [ID-Pal](/work/id-pal), an identity verification firm that had bought NorthRow and folded it into id-pal.com, is our example. "The redirect from northrow.com was technically correct, but nothing on either site told Google or the AI engines that NorthRow and ID-Pal were the same company." The fix was "About four hours of structured data work declaring NorthRow part of ID-Pal on both domains. Live in week one, with no copy changes and no risk to the redirect." The result: "AI answers still recommending NorthRow as a rival: six in May, one in September 2026." More in [why ChatGPT describes your firm wrongly, and how to fix it](/insights/ai-conflicting-information-online). ## How can you tell what it is costing you? You cannot price decisions you never saw, but you can estimate the scale in an afternoon. ### Step 1: Date-check your pages List your main pages and note when each was last meaningfully updated. Case studies, team profiles, service pages and any news section matter most. ### Step 2: Compare the site with the firm today Write down your current services, sectors, offices, senior people and nameable clients, and mark every place the website is wrong or silent. Each gap is a question a buyer or an assistant cannot match you to. ### Step 3: Ask your sales team what they apologise for Ask which pages they avoid sending and what they say about the site. Their answers show where it costs them time and credibility. ### Step 4: Ask AI assistants about your firm Ask ChatGPT, Gemini, Copilot and Perplexity what your firm does and who it works for, then five questions a buyer would ask. Note anything outdated or missing. ### Step 5: Put a rough value on it Take the clients you win in a typical year and their average value. Then ask: if the website were accurate and current, would one more buyer a year get in touch? For most B2B firms that one client is worth far more than keeping the site up to date. You do not need an exact number to see which way the sum falls. ## What should we update first on an old website? Fix the facts before adding anything new: people, services, sectors, offices and contact details, then case studies, because buyers and AI assistants rely on them most. The table gives our rule of thumb for each type of page. | Page type | Maximum sensible age | How often to check | What goes wrong if it slips | |---|---|---|---| | Team and people | Update the same week anyone joins or leaves | Monthly | Departed staff on the site, senior hires missing | | Office addresses and contact details | Update the same day they change | Quarterly | Buyers ring or visit the wrong place | | Service pages | Rewrite within a month of any change in what you sell; review at least yearly | Quarterly | Assistants describe services you have dropped | | Case studies | Newest no more than about six months old | Quarterly | You look like a firm that did the work once | | Prices, fees and timescales | Update as soon as they change | Quarterly | Wrong expectations on the first call | | News and insights | A new piece at least monthly, or retire the section | Monthly | A section that stops in launch year | | Footer, copyright line and legal details | Current year and current details | Yearly | Small signs that nobody is looking | ## What does a sensible monthly routine look like? One update to an existing page, one new page and one new piece of proof each month keeps most B2B websites current. That takes hours, not weeks. ### One update to an existing page Pick the page that matters most and is most out of date, check it against what you sell today, and add the questions buyers asked this month. ### One new page An answer to a question a prospect asked, or a page for a sector you have started winning work in. The test: would a buyer comparing you with two rivals find it useful? ### One new piece of proof A case study, a client quote you have permission to use, a new accreditation or a speaking slot. Proof dates quickly, and fresh proof shows you are doing this work now. Each quarter, check the team page, contact details, broken links and anything with a date; once a year, check the structure still matches how you sell. ## How much does it cost to keep a website up to date each month? Mostly time: a named owner with an hour or two a week, and a director to sign off anything describing the firm's expertise. Outside help costs more or less depending on whether it covers only technical upkeep or also writing and new pages, so compare scope before price. Our guide to [the real cost of a B2B website](/insights/real-cost-of-a-website) sets out sourced UK market figures, and our own fees are on [what it costs](/what-it-costs). ## Why is the eventual catch-up so expensive? Because two years of small jobs do not stay small. By the time someone finds the budget, the site needs an audit, a rewrite and often a rebuild, done under pressure because a big pitch has exposed the problem, with the risks of moved pages and lost search positions. We set out when a rebuild makes sense in [whether to redesign your website or keep improving it](/insights/website-redesign-vs-continuous-iteration). When a rebuild is the right answer, years of material do not have to be lost. For [arum](/work/arum), a 28-year-old collections and recoveries firm, it meant "310 pages rebuilt across three regions, in weeks." "215 articles, research pieces and news items moved across, going back to 2014." "Every old address redirected, so years of rankings carried over." As Jamie Waller, CEO and Executive Chair, put it: "For most people, our website is the first time they meet arum. It now does us justice." In our experience the firms that avoid the cycle are not the ones with bigger budgets. They are the ones that made the website someone's job. ## How do you make the case to your board or partners? Present it as a monthly operating cost with an owner, a scope and a result they can check, rather than "we need a new website" every three years with a large number attached. Start with what you can measure: enquiries through the website last quarter, how many became proposals, and how recent wins found you. If you cannot answer, start recording it; our guide to [measuring the return from your website](/insights/measure-website-roi-b2b) shows how. Then put the choice plainly: a modest fixed monthly amount with a named owner and a quarterly review, or nothing now, an unknown loss of enquiries and a larger rebuild bill in two or three years. ## What to do this month 1. **Find the most recent date on your site.** List everything about the firm that has changed since then. 2. **Fix the facts first.** Correct services, sectors, offices and people before adding anything new, using the table above. 3. **Name an owner.** Give one person responsibility for the site and an hour or two a week in their diary. 4. **Check what the assistants say.** Put the same five buyer questions to ChatGPT, Gemini and Copilot, and note anything wrong or missing. 5. **Agree the monthly routine.** One update, one new page and one new piece of proof, with the first review in the diary for three months' time. ## Frequently asked questions ### How often should a B2B website be updated? Some part of it every month: at least one new page or piece of proof, and one existing page checked against what you sell today. Review services, team and contact details every quarter, and straight away after any change in the business. ### Does updating old pages help as much as publishing new ones? Often it helps more. Seer's 2026 study found that among cited pages showing both dates, 72% had been updated in the past year but only 42% had first been published in it. New pages still matter, because they answer questions you do not yet cover. ### Is a site that rarely changes ever good enough? Sometimes, if your firm has not changed, wins almost all its work through a small circle of relationships and never sends prospects to the site. Even then, check once a year that every fact is still true and that AI assistants describe you correctly. An accurate site that rarely changes is fine; an inaccurate one is a liability. ### Our website is two years out of date. Can we fix it without a rebuild? Usually, yes. If the site loads quickly, works on a phone and is easy to edit, fix the pages buyers see most first. A rebuild is the right answer only when the site is technically poor or its structure no longer matches how you sell. ### Is a blog enough to stop our website looking outdated? Not on its own. A fresh blog next to frozen service pages, case studies and team profiles leaves the most important pages out of date. Keep those current first. An outdated website rarely fails loudly. It simply lets work go elsewhere without telling you. If you want a site built to be kept current and read correctly by AI assistants, see how [Website That Sells](/website-that-sells) works, or book a free [Growth Review](/growth-review) for three fixes in writing. ## Sources - [6sense, 2025 Buyer Experience Report (November 2025)](https://6sense.com/newsroom/the-timeline-for-influencing-b2b-buyers-is-shrinking-insights-from-6senses-2025-buyer-experience-report/) - [Seer Interactive, Study: content recency's impact on AI visibility in 2026 (July 2026)](https://www.seerinteractive.com/insights/study-content-recencys-impact-on-ai-visibility-in-2026) - [Ahrefs, Do AI assistants prefer to cite fresh content? (July 2025)](https://ahrefs.com/blog/do-ai-assistants-prefer-to-cite-fresh-content) - [Ahrefs, AI Overview citations from top-ten pages (March 2026)](https://ahrefs.com/blog/ai-overview-citations-top-10) - [Google Search Central, Creating helpful, reliable, people-first content (December 2025)](https://developers.google.com/search/docs/fundamentals/creating-helpful-content) --- Title: What happens when buyers let an AI agent build the shortlist? URL: https://fortitudemedia.ai/insights/ai-agents-change-b2b-buying Author: Ashley Kenerson Published: 2026-03-31 Last reviewed: 2026-09-24 Topic: How buying is changing Short answer: An AI agent given a buying task searches the web, reads your site and what others say about you, checks the facts against the buyer's criteria and returns a shortlist with reasons, before anyone contacts a supplier. Relationships and persuasion only help once you are on that list. To make it, publish specific, evidenced and consistent facts: who you serve, prices or ranges, timescales, case studies with numbers and named experts, in plain HTML, with an easy way to book a call. Your best salesperson has a gift for the first call. She reads the room, finds the real problem, and by the end of the conversation the prospect trusts her. Now picture a procurement manager who never makes that call, because she asked an AI agent to research the market, compare the options against her requirements and come back with three names. Your firm was not one of them, and nobody in your business knows the exercise happened. How common is this? Using AI for supplier research is already close to universal. In [6sense's 2025 Buyer Experience Report](https://6sense.com/newsroom/the-timeline-for-influencing-b2b-buyers-is-shrinking-insights-from-6senses-2025-buyer-experience-report/), a survey of more than 4,000 buyers, 94% used large language models, mostly to summarise reviews or analyse data, and 94% of buying groups had ranked their preferred vendors before first contact. They bought from that first choice 77% of the time. Handing the whole research task to an agent is much newer, and no survey we trust yet measures how often it happens. But the tools now exist, and the firms that make those shortlists are the ones whose evidence is already public. This guide covers which agents can do the job today, how one builds a shortlist, what it reads, whether it can contact you, and seven steps a founder-led firm can take this year to be on the list. ## Key takeaways - An AI agent given a buying task searches, reads your site and what others say about you, applies the buyer's criteria and returns a shortlist with reasons, before any supplier is contacted. - Several agents can already do this, including ChatGPT agent, Perplexity Comet and Researcher in Microsoft 365 Copilot, though some features are paid or US only. - Relationships and persuasion still win deals, but only among the firms already on the shortlist. - An agent can only use what it can read: published prices or ranges, timescales, case studies with numbers, named experts and facts that agree across sources. - Agents are starting to fill in forms and call businesses, so a plain booking link and a simple form matter. - The work that gets you on an agent's shortlist is the same work that gets you named by ChatGPT today, so none of it is wasted. ## Which AI agents can research suppliers today? Several, though most sit on paid plans and some are available only in the US. An agent here means an assistant given a job rather than a question: it browses, clicks, compares and reports back. | Agent | What it can do | Since | Limits | |---|---|---|---| | [ChatGPT agent](https://openai.com/index/introducing-chatgpt-agent/) (OpenAI) | Navigates websites, clicks and filters results, and delivers editable spreadsheets or slides summarising what it found | 17 July 2025 | Launched on Pro, Plus and Team plans | | [ChatGPT Atlas](https://help.openai.com/en/articles/20001371-evolving-atlas-into-chatgpt-for-browser-based-agentic-work) (OpenAI) | A browser with ChatGPT's agent built in | Stopped working on 9 August 2026 | Its agent features moved into the ChatGPT desktop app and Chrome extension | | [Perplexity Comet](https://www.perplexity.ai/hub/blog/comet-is-now-available-to-everyone-worldwide) | A browser whose assistant can research and take actions in any tab | Free worldwide since 2 October 2025 | None stated by Perplexity | | [Researcher in Microsoft 365 Copilot](https://www.microsoft.com/en-us/microsoft-365/blog/2025/03/25/introducing-researcher-and-analyst-in-microsoft-365-copilot/) | Multi-step research across the web and the buyer's own email, files and meetings | Announced 25 March 2025, now generally available | Needs a Microsoft 365 Copilot licence | | [Chrome auto browse](https://blog.google/products-and-platforms/products/chrome/gemini-3-auto-browse/) (Google) | Handles multi-step tasks, fills in online forms and gets quotes from tradespeople | 28 January 2026 | US only, Google AI Pro and Ultra subscribers | | [Google Search business calling](https://blog.google/products/search/deep-search-business-calling-google-search/) | Calls local businesses to ask about pricing and availability | 16 July 2025 | US only, local services | Researcher matters most for B2B firms, because many buyers already have Copilot at work and it can combine a web search with the buyer's own notes and emails about the purchase. ## What does an AI agent actually do with a buying task? It turns the brief into criteria, searches, reads each candidate and what others say about it, drops firms that fail a hard requirement, and writes up a comparison. No agent maker publishes how it weighs sources, so read what follows as how a careful researcher would work. The difference from asking a question is delegation. Instead of "who are the good payroll providers for construction firms", the instruction becomes "find payroll providers that handle CIS deductions for construction firms with 100 to 500 staff, UK-based, with support hours that cover site start times, and give me three to compare". The agent then: 1. Turns the brief into criteria: some explicit (CIS, staff numbers, support hours), some inferred from the type of purchase. 2. Searches for candidates, using live search and what the model already knows. Firms that only appear for their own brand name often do not come up. 3. Reads each candidate's website, looking for the specific facts the criteria require. A page that says "flexible pricing tailored to your needs" gives it nothing to record. 4. Reads what others say: directories, reviews, trade press, partner listings and client case studies. 5. Checks whether the facts agree. A firm that describes itself differently on three pages, or whose reviews contradict its claims, is harder to recommend with confidence. 6. Writes a short comparison: this firm meets everything; this one is cheaper but lacks a capability; this one is strongest on support but smaller. The buyer reads the comparison. Some will contact all three. Some will ask follow-up questions. Few start again from a blank page. By the time your sales team hears about the opportunity, if it hears at all, the field is down to a handful. What that does to your website is covered in [what your website should do now visitors arrive pre-sold](/insights/death-cold-website-visit-ai-funnel). ## Why do relationships arrive too late? Because relationships, referrals and persuasion win deals among the firms already on the shortlist, and they cannot get you onto it. The 6sense figures show the shortlist was already largely settled before first contact, and an agent moves that moment earlier still. In a human-run process there are many ways to win from a weak start: a warm introduction, a good first call, persistence when the buyer goes quiet. None of those exist in the research phase of an agent-run process. The agent has either found the evidence it needs about you or it has not. Sales still matters. Once the buyer contacts the three firms, everything your team is good at comes back into play, with a warmer, better informed prospect. What has moved is the work of getting into the room: from your sales team to your published evidence. ## What does the agent read and weigh? It reads your website first, then everything else that mentions you, looking for facts it can match against the buyer's criteria and evidence that others confirm them. **The explicit criteria** come from the buyer: what you do, who for, where you are based and where you serve, what it costs or a range, accreditations, how long onboarding takes and how support works. If any is missing from your site, you fail that line of the comparison by default. **The implicit criteria** are where firms separate: - Results: outcomes for clients like the buyer, with numbers and context, or just logos and adjectives? - Expertise: are senior people named, and do they publish or get quoted on the buyer's problem? - Standing: do trade press, professional bodies, awards and independent reviews mention you? - Transparency: is what a buyer needs easy to find, or behind a form? - Consistency: do your website, directories, profiles and coverage tell the same story? - Currency: is your latest evidence from this year, or from 2022? **The sector criteria** vary: planning success for an architect, software and specialisms for an accountant, certifications and response times for an IT provider. The thread is the same: public proof of success with organisations and problems like the buyer's. | An agent can use | An agent cannot use | |---|---| | "We serve manufacturing firms with 50 to 500 staff in the Midlands" | "We help businesses of all sizes grow" | | "Reduced month-end close from nine days to four for a twelve-site logistics group" | "Trusted by hundreds of firms" | | "From £X a month, depending on the number of sites" | "Contact us for pricing" | | Technical detail on the page | A brochure PDF behind a form | | Named experts who write and are quoted | "Our team" | Consistency is where many established firms fall down. For [TCC Group](/work/tcc-group), a compliance firm with three brands, the finding was blunt: "Three brands, one group. AI could only see the pieces." Part of the cause was "machine-readable data that declared a name, a logo and nothing else." ## Can an agent contact us or book a call? Increasingly, yes, so make it easy. Chrome auto browse can already fill in online forms and get quotes from tradespeople for US subscribers, and since July 2025 Google Search has been able to call some local businesses in the US to ask about pricing and availability. Expect B2B enquiries to go the same way in time. - Publish a plain booking link to a named person's diary, readable as ordinary text on the page. - Keep the enquiry form short, with labelled fields and no hidden steps. - Check your spam protection. A CAPTCHA wall that stops every automated visitor also stops an agent acting for a real buyer, so prefer protection that filters spam after the form is sent. - Show a phone number and email address that someone answers, and state prices or ranges and lead times so a call or an agent gets a straight answer. ## How should we prepare? Seven steps Make everything an agent needs specific, evidenced, consistent and easy to reach, starting with the pages that say what you do and for whom. ### Step 1: Audit what an agent can find today Write the brief a procurement manager might give an agent to find a firm like yours, with three hard requirements, then read your own site looking only for those facts. Put the same brief to ChatGPT, Copilot, Gemini and Perplexity and note who is named and which sources are cited. ### Step 2: Publish the basics you hold back Price ranges and what drives them, onboarding timescales, who you are and are not a fit for, security and compliance information, integrations. An agent treats "not disclosed" as a gap, and a gap is a reason to prefer a competitor who filled it. ### Step 3: Turn results into case studies with numbers Ask your ten best clients what changed, in numbers. Anonymise where you must, but publish the sector, the size, the starting position, the work and the result, one page each. ### Step 4: Make your experts visible Name your directors and senior people, write proper biographies, and have them publish on the problems clients bring, on your site and in trade titles. This is the core of [Reputation That Wins](/reputation-that-wins). ### Step 5: Gather third-party evidence Independent reviews, trade press mentions, professional body listings, awards and speaking slots each tell the agent that someone other than you finds you credible. [Why ChatGPT sounds unsure about some firms](/insights/ai-cross-references-business-sources) explains how assistants cross-check. ### Step 6: Make the facts agree everywhere Check your name, address, services, sectors and claims on your website, Companies House, LinkedIn, Google Business Profile, directories and press. Agree the facts once and repeat them everywhere. ### Step 7: Make the site readable by agents Put the answer at the top of each page, one page per service and sector, key text in the page's HTML rather than loaded by JavaScript, and accurate structured data. ID-Pal's rebuilt industry pages show the pattern: "Each opens with the question a compliance lead types, names the regulator, and answers inside the first ninety words." The [technical checklist for an AI-ready website](/insights/technical-checklist-ai-ready-website) covers crawler access and the rest. ## What is agentic commerce, and will it affect B2B services? Agentic commerce means agents that complete purchases, not just research them. So far it has centred on consumer shopping and simple bookings, such as the quotes and appointments Chrome auto browse handles. For B2B services, expect research, shortlisting and first contact to be handed to agents well before contracts are, which is why the steps above matter now. Firms that prepare early also build the evidence that keeps them recommended, as [why first movers are hard to dislodge](/insights/first-mover-advantage-ai) explains. ## What to do this month 1. Write the three-requirement brief a buyer might give an agent, and read your site looking only for those three facts. 2. Pick your three biggest gaps. For most firms they are prices, case studies and named experts. 3. Write to your five best clients this week and ask for a case study with numbers. 4. Draft a page on what your work costs and how long it takes, with honest ranges, and get it live. 5. Test your enquiry form and booking link as a stranger would, and remove any step an agent could not complete. 6. Book thirty minutes with your sales lead to list every question buyers ask that your website does not answer. That list is next quarter's content plan. ## Frequently asked questions ### How soon will buyers delegate supplier research to agents? Some already do, and almost all use AI somewhere in the process. Fuller delegation, where the agent applies criteria and returns a ranked comparison, is arriving as the tools mature. We would not put a date on widespread use. The preparation is the same either way, and it improves how assistants describe you today. ### Will a human still make the final decision? For most B2B purchases of any size, yes. The agent researches, compares and recommends, and a person approves, often after speaking to the firms. The change is that the person is approving a shortlist you may not be on, for reasons you did not get to argue with. ### What does this do to our sales team? It moves their best conversations later and makes them better: prospects from an agent's shortlist already know what you do and how you compare, so the first call is about fit, edge cases and terms. Every question the agent could not answer is a page marketing should write. Researched outbound also reaches buyers before any agent is briefed. Our [Pipeline That Fills](/pipeline-that-fills) works this way, with every message written and approved by a person; for [Helm Club](/work/helm-club): "Around 2,000 well-matched founders approached a month across three LinkedIn accounts, each checked against their live profile first. The first message asks a question. No pitch, no link, and every reply gets a same-day answer from a person." ### How can I tell if an AI agent has visited my website? Check your server logs for the user agents ChatGPT-User, Claude-User and Perplexity-User, which fetch pages when a person asks, alongside the search crawlers OAI-SearchBot, Claude-SearchBot and PerplexityBot. Agents that run inside the buyer's own browser, such as Chrome auto browse or Comet, can look like an ordinary visit. Add "an AI assistant" as an option on your enquiry form and ask on first calls. ### Do AI agents check review sites like G2 or Trustpilot? They read what search surfaces, and review sites surface often in B2B answers. In [G2's April 2026 survey](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html) of 1,076 software buyers, 45% said citations of review sites were the most confidence-inspiring signal in an AI answer. Pick the one or two review sites your buyers use and keep them current. ### We are a small firm. Can we compete here? Yes. Agents reward specificity and evidence more than size. A small firm that says precisely who it serves and publishes real outcomes for that segment can beat a national firm whose website is written for everyone. ### Is publishing our pricing really necessary? Publishing how pricing works is, even if the final figure depends on scope. An agent that finds "from £X a month, depending on the number of sites" can use it; one that finds "contact us" has nothing to compare. It also filters out poor-fit enquiries. The buyer who hands the research to an agent will not call to explain why you were left off. The only way to know is to look at what the agent would read. For an outside view, [book a free Growth Review](/growth-review): 30 minutes with a founder, with your site, two competitors and what Google and ChatGPT say about all three checked beforehand, and three fixes in writing afterwards. ## Sources - [6sense, 2025 Buyer Experience Report findings (November 2025)](https://6sense.com/newsroom/the-timeline-for-influencing-b2b-buyers-is-shrinking-insights-from-6senses-2025-buyer-experience-report/) - [G2 via PR Newswire, Half of B2B software buyers now start their research with AI chatbots (April 2026)](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html) - [OpenAI, Introducing ChatGPT agent (July 2025)](https://openai.com/index/introducing-chatgpt-agent/) - [OpenAI Help Center, Evolving Atlas into ChatGPT for browser-based agentic work (July 2026)](https://help.openai.com/en/articles/20001371-evolving-atlas-into-chatgpt-for-browser-based-agentic-work) - [Perplexity, Comet is now available to everyone worldwide (October 2025)](https://www.perplexity.ai/hub/blog/comet-is-now-available-to-everyone-worldwide) - [Microsoft 365 Blog, Introducing Researcher and Analyst in Microsoft 365 Copilot (March 2025)](https://www.microsoft.com/en-us/microsoft-365/blog/2025/03/25/introducing-researcher-and-analyst-in-microsoft-365-copilot/) - [Google, Chrome auto browse with Gemini 3 (January 2026)](https://blog.google/products-and-platforms/products/chrome/gemini-3-auto-browse/) - [Google, Deep Search and business calling in Google Search (July 2025)](https://blog.google/products/search/deep-search-business-calling-google-search/) --- Title: What is AI share of voice, and how do we compare with competitors? URL: https://fortitudemedia.ai/insights/benchmarking-ai-visibility-competitors Author: Ashley Kenerson Published: 2026-03-31 Last reviewed: 2026-09-24 Topic: Costs, returns and the business case Short answer: AI share of voice is your firm's mentions divided by all firm mentions in AI answers to a fixed set of buyer questions. Ask 20 to 50 real buyer questions in ChatGPT, Google's AI features, Copilot, Gemini, Perplexity and Claude, record every firm named, and compare your share with three to five rivals. Count only mentions that repeat. Then list the sites that cite competitors but not you, and start with the questions no firm owns yet. Your sales director mentions it in passing. Two of the last five prospects had also looked at a firm you barely think of as a competitor, and both said ChatGPT had suggested it. You know exactly who you are up against in pitches. You have no idea who you are up against in AI answers. That blind spot is getting more expensive. In [G2's April 2026 survey of 1,076 software buyers](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html), 51% said they now start their research in an AI chatbot, and 69% had chosen a different vendor from the one they first planned because of what a chatbot told them. [6sense's 2025 Buyer Experience Report](https://6sense.com/newsroom/the-timeline-for-influencing-b2b-buyers-is-shrinking-insights-from-6senses-2025-buyer-experience-report/) found buying groups had ranked their preferred vendors before first contact in 94% of cases. The shortlist forms before anyone calls you, and the lost enquiry never shows up in a report. Checking your own name in ChatGPT once tells you very little. Being named in one answer in five sounds poor until you learn your nearest rival is named in none, and fine until you learn they are named in every one. AI share of voice is the number that settles it. Here is how to calculate it, find the websites that name your rivals and not you, and pick the gaps to win first. ## Key takeaways - AI share of voice is your firm's mentions divided by all firm mentions in the AI answers to a fixed set of buyer questions, set against the same figure for each competitor. - Benchmark against three to five rivals, including any firm the assistants name that you rarely meet in pitches. - Use a fixed list of 20 to 50 buyer questions, ask each one several times a month in a clean session, and count a mention only when it repeats. - Ahrefs Brand Radar and Semrush's AI Visibility Toolkit now report competitors' AI mentions and the sources cited for them. - The citation gap, the websites that cite your rivals but not you, becomes a target list for PR, listings and reviews. - Questions where no firm is named consistently are the quickest wins, because there is no incumbent answer to displace. ## What is AI share of voice? AI share of voice is the proportion of firm mentions in AI answers that go to you, for the questions your buyers ask. Count every firm named across the answers to a fixed set of questions, then divide your mentions by the total. The idea comes from advertising and PR. Here the conversation is the list of names an assistant gives when a buyer asks who to hire: usually a short list rather than ten links, though the length varies by assistant and question. Tools define the measure in slightly different ways, so check which one you are reading. | Measure | How it is calculated | What it tells you | |---|---|---| | Mention share (share of voice) | Your mentions divided by all firm mentions | How much of the shortlist conversation you hold | | Presence rate | Answers naming you divided by all answers | How often a buyer hears your name at all | | First-mention rate | Answers naming you first divided by all answers | How often you are the main recommendation | [Ahrefs defines its AI share of voice](https://ahrefs.com/brand-radar) as "the percentage of AI responses within your topic set that mention or cite your brand versus competitors", which is closer to a presence rate. Either works if you use the same definition every time, for every firm. Example: you ask 40 buyer questions and the answers name firms 100 times in total. Your firm is named 15 times, Competitor A 50 times, Competitor B 20 times and Competitor C 15 times. Your share of voice is 15% and Competitor A's is 50%. On its own, 15 mentions sounds like a start. Next to a rival holding half the conversation, it shows which firm the assistants treat as the default. ## How do we calculate our AI share of voice? Run a fixed set of buyer questions through the assistants your buyers use, record every firm named, and divide. How to choose the questions and run them cleanly is covered in our guide to [finding out whether ChatGPT recommends your firm](/insights/track-ai-recommending-your-business): 20 to 50 questions in your buyers' words, each asked several times a month in a clean session, with a mention counted only when it repeats. A clean session matters because you are testing what a stranger hears. In ChatGPT, a Temporary Chat set to unpersonalised [does not use memory or custom instructions](https://help.openai.com/en/articles/8914046-temporary-chat-faq). We would not rely on logged-out sessions alone, as they may not match what a signed-in UK buyer is shown. Include Microsoft Copilot alongside ChatGPT and Google: [Forrester found in January 2026](https://www.forrester.com/blogs/b2b_buyers_make_zero_click_buying_number_one/) that business buyers are four times as likely as consumers to use it. The competitor benchmark then adds four things to your normal tracking: 1. **Record every firm named**, with its order and role: the recommendation, an alternative, or a passing mention. 2. **Count mentions and work out each firm's share**, with presence and first-mention rates alongside. 3. **Split the results by assistant and question type.** A rival that leads only on shortlist questions ("who are the best...") has a different advantage from one that leads on the problem questions buyers ask earlier. 4. **Copy the cited sources into the same sheet.** ChatGPT search, Perplexity, Copilot and Google's AI features show the pages they drew on. They feed the citation gap below. ## Who should we benchmark against? Benchmark against the three to five rivals your sales team meets most often, plus the firms the assistants actually name for your buyers' questions. The two lists rarely match, and the differences are often the most useful part of the exercise. Run a handful of buyer questions before you settle the list. You will often find a firm you rarely meet in pitches named again and again, such as a national player or a small specialist with an unusually clear website. If the assistants name them, that is where buyers are meeting them. ## Which tools show competitor mentions in ChatGPT? Ahrefs Brand Radar and Semrush's AI Visibility Toolkit both report how often AI assistants mention you and your competitors, and which sources they cite. For a first benchmark, a spreadsheet does the same job and forces you to read the answers. Ahrefs says [Brand Radar](https://ahrefs.com/brand-radar) tracks mentions across AI Overviews, AI Mode, ChatGPT, Perplexity, Gemini, Copilot and Claude, lets you "benchmark your brand against competitors in AI search", reports citations, and accepts custom prompts for "the exact questions your buyers ask". [Semrush's guide to its AI Visibility Toolkit](https://www.semrush.com/blog/how-to-track-your-chatgpt-visibility/) (November 2025) shows how to compare your ChatGPT visibility with competitors and lists "domains that cite competitors but not you". | Option | Competitors' mentions | Cited sources | Your own buyer questions | Best for | |---|---|---|---|---| | Spreadsheet, by hand | Yes | Where the assistant shows them | Yes | A first benchmark | | Ahrefs Brand Radar | Yes | Yes | Yes, as custom prompts | Spotting rivals you missed across the market | | Semrush AI Visibility Toolkit | Yes | Yes, including domains citing rivals but not you | Yes, through prompt tracking | Regular tracking and the citation gap | | Search Console generative AI reports | No, your own site only | No | No | Your own impressions in AI Overviews and AI Mode | Vendor tools run questions from their own systems, so check a sample by hand before you trust a trend. ## What do the results tell us? Three things matter: your share against the leader's, the questions where nobody is named consistently, and the assistants where you are missing. - **The gap to the leader.** The firm with the largest share is the assistants' default. Take its places question by question, starting where your claim is stronger. - **The questions nobody owns.** Where no firm is named consistently, or the assistant names no one, there is no incumbent to displace. - **The assistants that leave you out.** Treat this as a question, not a diagnosis. An assistant searching the web live may have found a recent page, one answering from memory reflects older material, and some lean on different sources. Check what each cites for your rivals first. - **What is actually said.** A mention that describes you wrongly or uses an old name is a separate, usually quicker, fix. See [why ChatGPT describes a firm wrongly](/insights/ai-conflicting-information-online). [ID-Pal](/work/id-pal), the identity verification firm, shows why the comparison matters. It had bought NorthRow, a UK business verification platform, yet "AI could describe ID-Pal. It never recommended it." The case study records the before and after: | What was measured | From the ID-Pal case study | |---|---| | Being named for UK buyer questions | "UK buyer questions where AI names ID-Pal: none in May, two in August, four in September 2026." | | A former rival still recommended | "AI answers still recommending NorthRow as a rival: six in May, one in September 2026." | | Being described correctly | "6 of 6 correct answers to 'what does ID-Pal do?', every engine, every run." | It also showed where to work next: "Where a page has been rebuilt, ID-Pal appears. Where it has not yet, it does not, which tells us exactly what to build next." ## Why does ChatGPT recommend my competitor and not us? Usually because the assistant can find the competitor described consistently in several places it trusts, and cannot find the same for you. The reasons are mostly visible from outside. - **Coverage by others.** [Muck Rack's May 2026 analysis](https://www.globenewswire.com/news-release/2026/05/07/3290268/0/en/generative-pulse-earned-media-consistently-drives-ai-citations-holding-at-84.html) of more than 25 million links cited by ChatGPT, Claude and Gemini found earned media made up 84% of them and paid content 0.3%. - **Other sites that reference them.** An [SE Ranking study reported in November 2025](https://www.searchenginejournal.com/new-data-top-factors-influencing-chatgpt-citations/561954/) found referring domains were the strongest predictor of ChatGPT citations. - **Reviews and listings.** In [G2's survey](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html), 45% of software buyers said review-site citations were the most confidence-inspiring signal in an AI answer. - **Pages that answer buyer questions.** A clear guide to the exact problem beats a general services page. - **A consistent description** across website, LinkedIn, directories and press, so the assistant can name the firm with confidence. We explain how assistants weigh these sources in [why ChatGPT sounds unsure about a firm](/insights/ai-cross-references-business-sources) and [whether backlinks still matter for AI search](/insights/ai-evaluates-quality-referring-domains). ## How do we find which websites ChatGPT cites for our competitors? Export the cited URLs for every answer, reduce them to domains, and list the domains that cite your rivals but not you. The top ten become your PR and listings target list. We call this the citation gap. ### Step 1: Export the cited URLs for every answer Copy every source URL from your benchmark answers into a sheet, next to the question, the assistant and the firms named. Ahrefs and Semrush export the same data. ### Step 2: Reduce each URL to its domain and note who it names Open each page and note which firms it mentions: you, each rival, or none. ### Step 3: List the domains that cite rivals but not you Filter for domains that mention at least one competitor and never mention you, and count how many answers each appears in. Domains that recur across questions and assistants matter most. ### Step 4: Turn the top ten into a target list For each of the top ten, choose the route in: a directory entry, a review profile, a contributed article or comment in a trade title, an award entry, or a correction to an out-of-date profile. Give each an owner and a date. This is where our [Reputation That Wins](/reputation-that-wins) work usually starts. ## Where are the gaps we can win first? Start where there is no incumbent, where a rival's lead rests on a thin page, or where a sector has gone unclaimed. In our experience, four gaps come up again and again: questions with no clear answer; topics a competitor covers with a single page, which a more useful set of pages answering the follow-up questions can overtake; sectors nobody has claimed, such as "accountants for dental practices" when rivals are named only for "accountants for small businesses"; and the sources at the top of your citation gap. Comparison questions deserve a special look, because a competitor often writes the answer by default. [Helm Club](/work/helm-club), a founder membership community, took this route: "We built the structured data AI engines read, then a comparison page for every question a founder asks when weighing up peer groups: Helm against YPO, Vistage and the rest." The case study records: "Every comparison page we have published ranks first for its question on Google and the AI engines. The pages not yet built are the ones where a competitor still writes the answer." Give each gap a score out of three: how much buyers care, how strong your expertise is, and how hard it would be for a rival to follow. Work on the two or three highest first. The effect builds over time, as we explain in [how content, PR and a better website work together](/insights/compound-returns-content-pr-ai-together). ## What is a good share of voice in AI search? There is no published benchmark for a good AI share of voice, and a single figure would mislead, because it depends on how many firms compete in your market and how the measure is defined. Compare yourself with the leader on the questions that bring in revenue. In our view, three readings matter more than the headline: whether you lead on any of your ten most valuable questions, whether the gap to the leader is closing each quarter, and whether your first-mention rate is rising. ## How often should we run the comparison? Track your own fixed list several times a month, and run the full competitor benchmark with the citation gap every quarter, using the same questions. Our guide to [building a simple AI visibility dashboard](/insights/build-ai-visibility-dashboard) shows a layout. ## What to do this month 1. Ask your sales team which competitors come up most, add any firm the assistants name unexpectedly, and settle on three to five. 2. Fix a list of 20 to 50 buyer questions in your buyers' words, and choose one share-of-voice definition. 3. Run the list through ChatGPT, Google's AI Overviews and AI Mode, Copilot, Gemini, Perplexity and Claude in clean sessions, recording every firm named and every source cited. 4. Work out each firm's share of voice and presence rate, by assistant. 5. Build your citation-gap list and turn the top ten domains into named PR, listing and review targets. 6. Pick the two or three unowned questions you are best placed to win, and book the next full run for three months' time. ## Frequently asked questions ### How many competitors should we compare ourselves with? Three to five suits most firms. Fewer and you may miss the firm quietly taking your place; more and the tracking becomes a chore nobody keeps up. Include any firm the assistants name repeatedly, even if you rarely meet it in pitches. ### Why do we get different answers each time we ask? AI assistants write each answer fresh, and some search the web live, so the firms named vary between runs. That is why a single check tells you little. Asking a fixed set of questions several times and counting only repeat mentions gives a number you can trust. ### What if we are far behind the leading competitor? That is the normal starting point, and it is recoverable. Assistants that search the web live can reflect a new page or article within weeks, while answers drawn from a model's memory change only when it is retrained, which takes months. Start with the questions nobody owns. ### How do we know the benchmark reflects real buyers? Check it against what your sales team hears. If the firms leading the AI answers are the ones prospects mention as alternatives, the benchmark is picking up real behaviour. If not, your questions may be written in your language rather than your buyers'. ### Can a competitor pay to be named ahead of us? Not inside the answer. ChatGPT has shown labelled ads to UK users on its Free and Go plans since 11 August 2026, and OpenAI says the ads do not change the answer itself. Being named in the answer is still earned. To see where you stand before building your own spreadsheet, [book a free Growth Review](/growth-review): 30 minutes with David Adams, with your site, two competitors and what Google and ChatGPT say about all three checked beforehand, and three fixes in writing afterwards. ## Sources - [G2, The Answer Economy: half of B2B software buyers now start research with AI chatbots (April 2026)](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html) - [6sense, 2025 Buyer Experience Report (November 2025)](https://6sense.com/newsroom/the-timeline-for-influencing-b2b-buyers-is-shrinking-insights-from-6senses-2025-buyer-experience-report/) - [Muck Rack, What Is AI Reading? Earned media drives 84% of AI citations (May 2026)](https://www.globenewswire.com/news-release/2026/05/07/3290268/0/en/generative-pulse-earned-media-consistently-drives-ai-citations-holding-at-84.html) - [Search Engine Journal, New data: top factors influencing ChatGPT citations (November 2025)](https://www.searchenginejournal.com/new-data-top-factors-influencing-chatgpt-citations/561954/) - [Semrush, How to track your ChatGPT visibility (November 2025)](https://www.semrush.com/blog/how-to-track-your-chatgpt-visibility/) - [Ahrefs, Brand Radar (checked September 2026)](https://ahrefs.com/brand-radar) - [OpenAI Help Center, Temporary Chat FAQ (September 2026)](https://help.openai.com/en/articles/8914046-temporary-chat-faq) - [Forrester, B2B buyers make zero-click buying number one (January 2026)](https://www.forrester.com/blogs/b2b_buyers_make_zero_click_buying_number_one/) --- Title: What is E-E-A-T and why do AI assistants care about it so much? URL: https://fortitudemedia.ai/insights/eeat-why-ai-cares-more Author: David Adams Published: 2026-03-31 Last reviewed: 2026-09-24 Topic: Getting named by AI assistants Short answer: E-E-A-T stands for experience, expertise, authoritativeness and trustworthiness, the qualities Google's systems try to reward; Google says trust matters most and that E-E-A-T is not itself a ranking factor. AI assistants that search the web start from a search index, then quote specific, sourced passages, so the same qualities decide who gets cited. Show named, checkable authors, specific examples from real work, sources for every figure and honest limits, especially on financial and legal topics. Your marketing manager has just used the phrase "E-E-A-T" in a meeting and you nodded, because that seemed easier than asking. Then a prospect mentions they asked ChatGPT which firms to consider before ringing you, and you wonder whether the two things are connected. They are. E-E-A-T stands for experience, expertise, authoritativeness and trustworthiness. Google uses the idea to describe the content its systems try to reward, and [Google's own guidance](https://developers.google.com/search/docs/fundamentals/creating-helpful-content) is clear about which part matters most: "Of these aspects, trust is most important." It describes what both Google and AI assistants end up favouring. There is research behind that for AI answers too. In a [Princeton and IIT Delhi study presented at KDD 2024](https://arxiv.org/abs/2311.09735), adding quotations from relevant people raised a page's visibility in AI-generated answers by about 41%, adding statistics by about 34% and citing sources by about 30%, while keyword stuffing lowered it. It was tested on a research engine, so treat it as directional. The point for a founder is that the raw material of E-E-A-T is your firm's real experience, and only you have it. ## Key takeaways - E-E-A-T is Google's shorthand for experience, expertise, authoritativeness and trustworthiness, and Google says trust is the most important of the four. - Google says E-E-A-T is not itself a ranking factor, and its human raters' scores do not directly change rankings; its systems look for signals that line up with it. - AI assistants that search the web usually start from a search index, so pages Google and Bing trust are more likely to be read, and specific, sourced passages are more likely to be quoted. - Financial, legal and other "Your Money or Your Life" topics are held to a higher standard, which covers most fintech and professional services firms. - The clearest sign of experience on a page is specific detail: what you did, for whom, over what period, and what went wrong. - AI-written content is not banned, but pages produced at scale with little effort or originality breach Google's spam policies. - Named authors with credentials that can be checked elsewhere, including on LinkedIn, make every other signal more believable. ## What does E-E-A-T actually mean? E-E-A-T is four questions about a page, and each has a plain English version. - Experience: has the writer actually done this, rather than read about it? - Expertise: does the writer or the firm understand the subject properly? - Authoritativeness: do others recognise the firm as a serious voice on this subject? - Trustworthiness: can a reader rely on the page being accurate, honest and free of hidden motives? The four come from Google's Search Quality Rater Guidelines, the instructions for the people who check whether Google's results are any good. The [September 2025 edition](https://static.googleusercontent.com/media/guidelines.raterhub.com/en//searchqualityevaluatorguidelines.pdf) puts it bluntly: "Trust is the most important member of the E-E-A-T family because untrustworthy pages have low E-E-A-T no matter how Experienced, Expert, or Authoritative they may seem." ## Is E-E-A-T a ranking factor? No, not directly. Google says "E-E-A-T itself isn't a specific ranking factor", but that "using a mix of factors that can identify content with good E-E-A-T is useful". Its raters do not move pages either: "Rater data is not used directly in our ranking algorithms." It helps to separate what Google has said from what we infer from our own work. | What Google documents | What we infer from client work | |---|---| | Trust is the most important part of E-E-A-T | Pages with sourced figures, dated reviews and named authors are the ones assistants quote most readily | | E-E-A-T is not a single ranking factor; systems use signals that align with it | Links, mentions and reviews elsewhere are how those signals show up in practice | | Topics affecting financial stability, health or safety get more weight on E-E-A-T | Fintech, legal and financial advice pages need visible credentials more than most | | Scaled, low-value content breaches spam policy however it is made | Generic pages written to fill a keyword list rarely get quoted by any assistant | ## Why does E-E-A-T matter for AI assistants? Because assistants that search the web choose their sources from a search index first, and then quote the passages that answer the question best. Both stages favour the same qualities. Google says its AI features are rooted in its core ranking and quality systems, and ChatGPT search has drawn on Bing, so a page those systems distrust is less likely to be read at all. Once a page is read, the assistant needs something specific to lift. [Google's guide to its AI features](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide), updated in July 2026, says content offering first-hand expertise, which it calls non-commodity content, will influence AI search presence more than anything else, and contrasts it with commodity pages such as "7 Tips for First-Time Homebuyers". It is tempting to picture an assistant weighing every sentence for expertise. What we observe is simpler: pages that contain specific detail, sources and clear claims are the ones that get quoted, and generic pages are passed over. Authority off the page matters too. An [SE Ranking study of ChatGPT citations](https://www.searchenginejournal.com/new-data-top-factors-influencing-chatgpt-citations/561954/), reported in November 2025, found the number of referring domains was the strongest predictor of whether a page was cited. ## What does YMYL mean for a financial services or professional services website? YMYL stands for "Your Money or Your Life": topics that, in Google's words, "could significantly impact the health, financial stability, or safety of people, or the welfare or well-being of society". Google says its systems give even more weight to strong E-E-A-T on these topics, and the rater guidelines use examples such as retirement savings and tax. Many of the firms we work with sit inside that definition: fintech, wealth and investment firms, accountants, lawyers and compliance specialists. In practice that means: - Name a qualified author or reviewer on every advice page, with their credentials and a link to a profile that can be checked. - Name the regulator and the rule where one applies, and date the page so a reader knows when it was last checked. - State who the advice suits and when a reader should take professional advice instead. - Disclose any interest, such as a product you sell or a partner you refer to. ID-Pal, an identity verification firm we work with, rebuilt its sector pages this way: "Each opens with the question a compliance lead types, names the regulator, and answers inside the first ninety words." The [ID-Pal case study](/work/id-pal) shows what changed in AI answers afterwards. ## How do you show experience on a page? With specific detail that only comes from having done the work: a sequence of events, a named role, a problem that appears only in practice, and what you changed because of it. Compare two ways of writing about implementing a new finance system in a growing firm. The first: "A new finance system helps growing businesses gain visibility. Many firms struggle with the transition. A phased approach can be tailored to your needs. Common steps include planning, data migration and training." The second, written as an illustrative example: "Say a distribution business with three depots moves off spreadsheets. The migration itself takes six weeks, but the real problem shows up in month two, when the depot managers keep running their old spreadsheets alongside the new system because they do not trust the stock figures. The fix is not more training, it is a fortnight of reconciling both sets of numbers in front of them until the gap closes." The second reads as a report of what happens, not a prescription of what should. Whether you use a real client, with permission, or a clearly labelled example, that is the level of detail that signals experience. Founders often strip it out because they worry about giving the method away or about generalising from one client. The cost is a page that reads as though nobody in the firm has done the work. Put back timelines, roles, sectors and sizes of client, what did not work, and the edge cases you have hit. ## How do you show expertise, authority and trust? Expertise shows in explanation, authority shows in what others say about you, and trust shows in accuracy and honesty. Each has a few concrete signs. | Quality | What it looks like on the page | What undermines it | |---|---|---| | Expertise | Explains causes, not just symptoms; uses the right term for each concept and sticks to it; engages with other research; says where its knowledge stops | Generalities that could apply to any firm; claiming to cover every subject | | Authority | A body of work on one bounded subject, published steadily; credentials that check out elsewhere; coverage, awards and citations from others | Fifty articles on fifteen unrelated subjects; a burst of posts followed by silence | | Trust | Sources for every statistic; disclosed interests; "it depends" followed by what it depends on; corrections when you get something wrong | Unsourced figures; hidden commercial interests; simple answers to complicated questions | Two points need care. Precise terms matter because they are precise: "customer journey" and "sales cycle" are different things, so use each correctly rather than swapping them for variety. And sourcing matters because readers check. In [TrustRadius's 2026 survey](https://www.prnewswire.com/news-releases/trustradius-2026-b2b-buying-disconnect-report-reveals-ai-has-changed-how-buyers-research-but-not-what-they-trust-302825792.html), 94% of technology buyers who used AI in a purchase said they fact-checked its answers at least sometimes. Our guide to [what makes content expert rather than marketing](/insights/what-makes-expert-quality-content) goes deeper on the writing itself. ## Does AI-written content hurt E-E-A-T? Using AI to help write is not the problem; publishing pages with nothing of your own in them is. Google's [spam policies](https://developers.google.com/search/docs/essentials/spam-policies) define scaled content abuse as creating "large amounts of unoriginal content that provides little to no value to users, no matter how it's created", and give "using generative AI tools or other similar tools to generate many pages without adding value for users" as an example. The rater guidelines go the same way. The September 2025 edition defines generative AI and lists scaled content abuse and main content "created with little to no effort, little to no originality, and little to no added value for website visitors" among the signs of the lowest-quality pages. Google also suggests explaining how automation or AI was used where readers might wonder how a page was made. Our own approach is the one we recommend: AI can draft a structure or tidy the prose, but a person with the experience adds the examples, figures and judgement, and approves every page. Our guide to [whether ChatGPT and Google will quote content written with AI](/insights/thought-leadership-vs-keyword-stuffing) covers this in more detail. ## How do you build E-E-A-T into your firm's content? Work in order: audit, narrow the subject, make your people checkable, then build the evidence. ### Step 1: Audit what you already have Take your most important pages and ask five questions of each: does it include a specific example, cite sources, disclose any interest, acknowledge limits, and use the same terms as the rest of the site? Score each out of five. In our experience the gaps cluster around examples and limits. ### Step 2: Define the subject you want to own "Corporate finance" is too broad. "Management buyouts for owner-managed businesses" is a subject an assistant can recognise you for. Make it explicit in how the site is organised and how your people are described. ### Step 3: Make your people visible and checkable Every author needs a profile with their role, credentials and relevant experience, and links to where those can be checked. Take our own. [Ashley Kenerson's profile](/team/ashley-kenerson) says she spent thirteen years at Digitas, joining as an analyst in 2006 and leaving in 2019 as Head of Data, Analysis and Insights. [Ross Williams's profile](/team/ross-williams) says he bootstrapped Venntro Media Group, ran it for twenty-one years, sold it in 2024, and wrote Subscribe and Conquer, a playbook for subscription businesses. Each claim can be checked on LinkedIn, in the book and in the public record, and that is what makes it worth stating. ### Step 4: Set a citation standard Statistics need a named source and a date, quotations need attribution, and "best practice" claims need to say who recommends them. Remove any figure you cannot source. ### Step 5: Collect and publish evidence of experience Case studies with real outcomes and permission, named testimonials, your own research, talks your people have given. Keep a last-reviewed date on each page and update it when the facts change. ## How do you tell whether it is working? Measure the inputs and watch the outputs. On the input side, track the share of your key pages with a specific example, sources, a named author and stated limits. On the output side, run a fixed list of 20 to 50 buyer questions in ChatGPT, Google AI Mode, Copilot and Perplexity each month, in a clean session, and record whether you are named, how you are described and which pages are cited. Our guide to [tracking whether ChatGPT recommends your firm](/insights/track-ai-recommending-your-business) sets out the method. ## What to do this month 1. Score your ten most important pages on the five audit questions and note which signals are missing. 2. Write or rewrite a profile for everyone who publishes, with credentials that can be checked and a link to their LinkedIn. 3. If you work in finance, law or compliance, add a named reviewer, the relevant regulator and a review date to each advice page. 4. Take one generic page and rewrite it around a real or clearly labelled example, with a timeline, a complication and what changed. 5. Add a source and date to every statistic on those ten pages, and remove any figure you cannot source. ## Frequently asked questions ### Do author bios help our content get cited by ChatGPT? There is no published evidence that a bio on its own gets a page cited. What a bio does is make every other claim on the page believable to a reader and to Google, and it gives an assistant a checkable link between the page and a real, qualified person. Keep it short, specific and linked to profiles elsewhere. ### Does the founder's LinkedIn profile matter? Yes, more than many firms assume. A [Semrush study published in March 2026](https://www.semrush.com/blog/linkedin-ai-visibility-study/) found LinkedIn appeared in about 11% of AI answers, rising to 14.3% in ChatGPT search, with articles making up 50 to 66% of the LinkedIn pages cited. A founder profile that states credentials plainly and publishes practical articles is itself a source assistants read. It is why founder LinkedIn is part of [Reputation That Wins](/reputation-that-wins). ### Is E-E-A-T different for a B2B firm than for a consumer brand? Somewhat. For a B2B firm, verifiable credentials, professional affiliation and evidence of client work carry the most weight. A consumer site might build trust on "I have raised three children"; an accountancy practice builds it on "we have taken forty owner-managed businesses through a sale". ### How quickly can we build authority? Not quickly, and anyone who promises otherwise is selling links. Narrow your subject so depth is achievable, make your people's credentials visible now, publish steadily rather than in bursts, and earn mentions from credible sources over months. ### Should we delete old content that scores badly? Selectively. Rewrite pages that answer your top buyer questions. Retire or redirect low-value pages, especially any that contradict what you now say, because contradictions between old and new pages do more damage than a weak page alone. ### Can we be authoritative without real expertise? Not for long. Good writing can disguise a lack of understanding from a casual reader, but generic content lacks the specific detail, sources and limits that get quoted, and buyers check. The only reliable route is to write down what your people already know. The firms that do well here are the ones that write down what they already know. If you are not sure whether your site reads as experienced and trustworthy, [book a free Growth Review](/growth-review): 30 minutes with David Adams and three fixes in writing. ## Sources - [Google Search Central, Creating helpful, reliable, people-first content (December 2025)](https://developers.google.com/search/docs/fundamentals/creating-helpful-content) - [Google, Search Quality Rater Guidelines (September 2025)](https://static.googleusercontent.com/media/guidelines.raterhub.com/en//searchqualityevaluatorguidelines.pdf) - [Google Search Central, Spam policies for Google web search (August 2026)](https://developers.google.com/search/docs/essentials/spam-policies) - [Google Search Central, Optimizing your website for generative AI features on Google Search (July 2026)](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide) - [arXiv, Aggarwal and others, Princeton and IIT Delhi: source visibility in AI-generated answers, KDD 2024 (June 2024)](https://arxiv.org/abs/2311.09735) - [Search Engine Journal, New data: top factors influencing ChatGPT citations (November 2025)](https://www.searchenginejournal.com/new-data-top-factors-influencing-chatgpt-citations/561954/) - [TrustRadius via PR Newswire, 2026 B2B Buying Disconnect Report (July 2026)](https://www.prnewswire.com/news-releases/trustradius-2026-b2b-buying-disconnect-report-reveals-ai-has-changed-how-buyers-research-but-not-what-they-trust-302825792.html) - [Semrush, LinkedIn AI visibility study (March 2026)](https://www.semrush.com/blog/linkedin-ai-visibility-study/) --- Title: What is it costing us not to be recommended by ChatGPT? URL: https://fortitudemedia.ai/insights/cost-of-inaction-losing-right-now Author: David Adams Published: 2026-03-31 Last reviewed: 2026-09-24 Topic: Costs, returns and the business case Short answer: Estimate it from your own numbers: buyers in your market, the share who shortlist with an AI assistant, the gap in how often you are named, the share of named firms a buyer contacts, your win rate and client value. In our worked example the firm misses four clients and £60,000 of first-year fees a year, or £360,000 over three years as missed clients stack up. A cautious case still shows a recurring loss. Nobody sends you a letter when a buyer asks ChatGPT for three firms that do what you do, and you are not one of them. The buyer contacts the names on the list, picks one, and the work goes elsewhere. Your pipeline report shows nothing, because the enquiry never existed. You notice only indirectly: a competitor who seems to be winning more, prospects who arrive already talking about someone else, a quiet quarter nobody can explain. The habit is now common. In [G2's April 2026 survey of 1,076 software buyers](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html), 51% said they start their research in an AI chatbot, and 69% had chosen a different vendor from the one they first planned because of what a chatbot told them. [6sense's 2025 Buyer Experience Report](https://6sense.com/newsroom/the-timeline-for-influencing-b2b-buyers-is-shrinking-insights-from-6senses-2025-buyer-experience-report/) found buying groups had ranked their preferred vendors before first contact in 94% of cases, and bought from that first choice 77% of the time. If the shortlist is settled before anyone calls, missing it costs real money. That cost is easy to ignore because every other line in your management accounts has a number next to it and this one does not. The answer is not to guess a dramatic figure. It is to estimate a sensible one from numbers you already have. This guide shows how, with a worked example, a cautious case and a calculator you can copy into a spreadsheet. ## Key takeaways - Enquiries lost to AI shortlists never appear in your pipeline, so the cost has to be estimated rather than read from a report. - Six inputs are enough: market size, the share who shortlist with AI, the gap in how often you are named, the share of named firms a buyer contacts, your win rate and client value. - Do not treat every firm named in an answer as shortlisted; applying the share a buyer actually contacts keeps the estimate honest. - In our worked example the central case misses four clients and £60,000 of first-year fees a year, and £360,000 of fees over three years as missed clients stack up. - Show a cautious case beside the central one: a board will back a reasoned estimate with visible workings more readily than one confident number. - The biggest lever is usually the gap between how often you are named and how often you could be, which is the part you control. ## Why does this cost never show up in our reports? Because it happens before a buyer contacts you. A lost pitch leaves a record: a proposal, a debrief, a reason. A buyer who never heard of you leaves nothing. Search used to leave at least a trace, in rankings and traffic. When a buyer asks ChatGPT, Gemini, Copilot or Perplexity for a shortlist, there is no ranking report and often no click. Even on Google, [Pew Research Center found in July 2025](https://www.pewresearch.org/short-reads/2025/07/22/google-users-are-less-likely-to-click-on-links-when-an-ai-summary-appears-in-the-results/) that users clicked a traditional result on 8% of visits when an AI summary appeared, against 15% without one. So the signs are indirect: - Prospects mention competitors you rarely used to meet, and say an AI assistant suggested them. - The share of enquiries from people with no prior connection to you is falling. - Competitors' new clients say they "found them on ChatGPT", and you hear it second-hand. - Traffic to your advice pages is falling while your rankings hold. We explain why in [why traffic falls when rankings have not](/insights/zero-click-search-your-business). None of these gives you a number. For that, you need to build one. ## What percentage of B2B buyers use ChatGPT to find suppliers? About half of software buyers now start there, and almost all business buyers use AI somewhere in the process. Three recent surveys give the range, all from outside the UK services market: | Survey | Who was asked | Finding | |---|---|---| | [G2, April 2026](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html) | 1,076 B2B software buyers | 51% start research in an AI chatbot; 71% use one at some point | | [Forrester, January 2026](https://www.forrester.com/blogs/b2b_buyers_make_zero_click_buying_number_one/) | Business buyers | 94% use AI in their buying process | | [TrustRadius, July 2026](https://www.prnewswire.com/news-releases/trustradius-2026-b2b-buying-disconnect-report-reveals-ai-has-changed-how-buyers-research-but-not-what-they-trust-302825792.html) | 1,862 technology buyers | 63% used AI during their purchase; 94% of those checked what it said | These surveys cover software and technology buyers, not buyers of professional services, so treat them as a ceiling for your own estimate rather than a figure to copy. The number that matters is your own, which you get by asking every new enquiry. ## How do we put a number on it? Six inputs, most of which you already know. Where you have to estimate, estimate cautiously: a modest figure you can defend is worth more than a dramatic one nobody believes. The example below is illustrative throughout, so swap in your own numbers. ### Step 1: Estimate how many buyers choose a firm like yours each year Count the buyers in your target market who appoint a provider like you in a typical year. If you do not know, work backwards. For example, if you win around 25 new clients a year and think that is about one in twenty of the work available, the market is around 500 buyers. For the worked example, say you run a specialist employment law practice and 500 businesses in your region and sectors appoint a new employment law firm each year. ### Step 2: Estimate the share who build their shortlist with an AI assistant Use the surveys above as a ceiling, then find your own evidence. Ask every new enquiry, recent client and lost prospect whether they used ChatGPT or a similar assistant while researching, and add the question to your enquiry form. Our guide to [tracking leads that come through AI answers](/insights/attribution-ai-era-track-leads) explains how. For the example, say one in five of those buyers asks an AI assistant for a shortlist: 100 buyers a year, well below the survey figures for software. ### Step 3: Measure the gap in how often you are named Ask the assistants the questions your buyers ask and count how often you are named, then look at the most-named competitor to see what is achievable. Our guide to [measuring AI share of voice against competitors](/insights/benchmarking-ai-visibility-competitors) sets out the method. In the example, you are named in one answer in ten, so about 10 of the 100. The leading competitor is named in half. With a year of focused work, four in ten is a realistic target, or 40 answers. The gap is 30. ### Step 4: Apply the share of named firms a buyer actually contacts Being named in an answer is not the same as being on the shortlist. Some buyers contact every firm the assistant names; others contact one or two. Ask recent clients how many firms they spoke to. In the example, say buyers contact two in three of the firms named, so the 30 extra mentions become 20 real shortlists. ### Step 5: Apply your win rate Use your rate for competitive situations where the buyer speaks to several firms; your referral win rate is higher and would overstate the loss. In the example, you win one in five, so 20 shortlists means four clients a year. ### Step 6: Turn clients into money Use first-year fees and gross profit over the relationship, and show both. In the example, the average client pays £15,000 a year, gross margin is 50% and clients stay three years. Four missed clients is £60,000 a year in first-year fees, and four times £22,500, or £90,000, of gross profit over the life of those clients. ## What does the worked example tell us? The example firm is missing about four clients a year to competitors it may never meet, worth about £60,000 in first-year fees, and none of it shows in its reports. Your finance director will test that against cautious assumptions, so do it first. | Input (example) | Central case | Cautious case | |---|---|---| | Buyers choosing a firm each year | 500 | 500 | | Share who shortlist with AI | 1 in 5 (100) | 1 in 10 (50) | | Answers naming you today | 10 | 5 | | Answers achievable with a year's work | 40 | 15 | | Extra answers naming you | 30 | 10 | | Share of named firms the buyer contacts | 2 in 3 (20) | 1 in 2 (5) | | Win rate when contacted | 1 in 5 | 1 in 5 | | Clients missed each year | 4 | 1 | | First-year fees missed | £60,000 | £15,000 | | Fees missed over three years | £360,000 | £90,000 | The last row is the one to state plainly. Clients stay three years in the example, so a client missed in year one is still missing in years two and three. In the central case, the fees missed are £60,000 in the first year, £120,000 in the second and £180,000 in the third: £360,000 over three years, before the share of buyers using AI changes at all. Even the cautious case shows a real, recurring loss. And the biggest lever is Step 3, the gap between how often you are named and how often you could be, which is the part you control. ## Can we calculate it with our own numbers? Yes. Copy these nine inputs into cells B1 to B9 of a spreadsheet, using your own figures, and the three formulas below give you the answer. Run it twice: once with central inputs and once with cautious ones. | Cell | Input | Worked example | |---|---|---| | B1 | Buyers choosing a firm like yours each year | 500 | | B2 | Share who shortlist with an AI assistant | 0.2 | | B3 | Share of answers naming you today | 0.1 | | B4 | Share of answers achievable with a year's work | 0.4 | | B5 | Share of named firms a buyer contacts | 0.67 | | B6 | Win rate when contacted | 0.2 | | B7 | Average annual fee per client | 15,000 | | B8 | Gross margin | 0.5 | | B9 | Average client lifetime in years | 3 | - **Clients missed each year:** `=B1*B2*(B4-B3)*B5*B6` - **First-year fees missed:** `=B1*B2*(B4-B3)*B5*B6*B7` - **Lifetime gross profit of one year's missed clients:** `=B1*B2*(B4-B3)*B5*B6*B7*B8*B9` With the example inputs, the first formula returns just over four clients. Round down rather than up when you present it. ## What happens if a competitor gets recommended by ChatGPT first? They win work you never see, and each win gives them more case studies, reviews and press comment for the assistants to find. In our view that makes an early lead self-reinforcing, and catching up slower than starting early, because the pages, coverage and listings that get a firm named take months to build whenever you start. We make the fuller case, with the evidence, in [whether you can wait a year before acting](/insights/first-mover-advantage-ai). The flip side is what being named is worth. For [Helm Club](/work/helm-club), a founder membership community: "A channel that did not exist before now brings one in five new members: founders who ask ChatGPT, Gemini or Claude which peer group to join are told about Helm." ## What costs do not fit in the spreadsheet? Some damage is real but hard to price. Mention it to your board without putting a number on it. - **Price pressure.** A buyer who arrives through a competitor's recommendation has already been told who the expert is, so you are compared on price. - **Longer sales cycles.** Prospects who were not told about you need more meetings and more proof. - **Recruitment.** Candidates ask AI assistants about employers too. See [what ChatGPT tells candidates about your firm](/insights/ai-search-recruitment-talent). - **Due diligence.** If you plan a sale or new investment, "why do buyers not hear about you?" is an awkward question. ## How do we calculate the return on AI search for a professional services firm? Put the cost of inaction next to the cost of closing part of the gap. The comparison is what gets a decision made; the estimate alone rarely does. In the example, closing half the gap wins two extra clients a year: £30,000 in first-year fees and £45,000 of gross profit over their lifetime, with each following year's clients on top. Set that against a budget sized from the value of a client, using our guide to [what AI search optimisation costs in the UK](/insights/ai-search-optimisation-cost-uk), and present it with [making the business case to your board](/insights/business-case-ai-optimisation-board). Be open about the assumptions. Show the central and cautious cases side by side, say where each input came from, and commit to replacing estimates with real data as you collect it. ## What to do this month 1. Ask ChatGPT, Google, Copilot and Gemini ten of your buyers' main questions and count how often you and your leading competitor are named. 2. Start asking every new enquiry how they found you and whether they used an AI assistant, and record the answer in your CRM. 3. Ask your last ten clients how many firms they contacted before choosing you. 4. Pull your competitive win rate, average client fee, gross margin and client lifetime from last year's figures. 5. Fill in the calculator with central and cautious inputs. 6. Take the result, with a proposed budget beside it, to your next leadership meeting. ## Frequently asked questions ### How many leads are we losing because ChatGPT does not recommend us? Nobody can read that from a report, but you can estimate it. Multiply the buyers in your market by the share who shortlist with AI, the gap in how often you are named, and the share of named firms a buyer contacts: that gives the missed shortlists, or lost leads, before your win rate is applied. In the central example it is 20 a year. ### Is this figure reliable enough to show our board? It is an estimate, and you should present it as one. It is credible because every input comes from your own data or a stated assumption, and because a cautious case sits beside the central one. Boards distrust a single dramatic number with no workings. ### What if we are already named in some AI answers? Then the gap is smaller, but rarely zero. Firms are often named for their general category but missing from the specific questions that signal a ready buyer, naming a sector, a location or a problem. Those gaps are usually where the money is. ### Does this apply if most of our work comes through referrals? Yes, though the number will be smaller. Referred buyers may check the name they were given with an AI assistant before they call. If it describes you vaguely or suggests alternatives, some referrals leak away. Ask referred clients what they found when they looked you up. ### How quickly could we close the gap? Assistants that search the web live can reflect a new page or article within weeks; answers drawn from a model's memory change only when the model is retrained, which takes months. [ID-Pal](/work/id-pal) is a dated example: "UK buyer questions where AI names ID-Pal: none in May, two in August, four in September 2026." Closing a large gap to an entrenched competitor takes longer, so start with the questions where nobody is named yet. The first input you need is how often the assistants name you today. For an outside view, [book a free Growth Review](/growth-review): 30 minutes with David Adams, with your site, two competitors and what Google and ChatGPT say about all three checked beforehand, and three fixes in writing afterwards. ## Sources - [G2, The Answer Economy: half of B2B software buyers now start research with AI chatbots (April 2026)](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html) - [6sense, 2025 Buyer Experience Report (November 2025)](https://6sense.com/newsroom/the-timeline-for-influencing-b2b-buyers-is-shrinking-insights-from-6senses-2025-buyer-experience-report/) - [Forrester, B2B buyers make zero-click buying number one (January 2026)](https://www.forrester.com/blogs/b2b_buyers_make_zero_click_buying_number_one/) - [TrustRadius, 2026 B2B Buying Disconnect Report (July 2026)](https://www.prnewswire.com/news-releases/trustradius-2026-b2b-buying-disconnect-report-reveals-ai-has-changed-how-buyers-research-but-not-what-they-trust-302825792.html) - [Pew Research Center, Google users are less likely to click on links when an AI summary appears (July 2025)](https://www.pewresearch.org/short-reads/2025/07/22/google-users-are-less-likely-to-click-on-links-when-an-ai-summary-appears-in-the-results/) --- Title: What makes a buyer trust our website enough to get in touch? URL: https://fortitudemedia.ai/insights/psychology-of-trust-website-conversions Author: David Adams Published: 2026-03-31 Last reviewed: 2026-09-24 Topic: Websites that bring work in Short answer: Buyers trust a B2B website when it answers three questions fast: can this firm do the work, will it be reliable, and is it safe to deal with. The strongest signals are specific, named and dated proof, a plain statement of who you serve, real people, a contact page with a phone number, address and company details, and honest guidance on price, fit and risk. It is the same evidence AI assistants look for when deciding which firms to name. A buyer lands on your website with your name and two others on a list. Maybe a colleague gave them the names. Maybe ChatGPT did. Either way, they are not reading your site the way you wrote it. They are checking it. Within a minute or two they will decide whether you are worth thirty minutes of their week, and most of what decides it is small. They are looking for reasons to keep you on the list or cross you off. A specific case study from a firm like theirs keeps you on. A vague line about "delivering excellence" does not. A contact page with only a form makes them wonder. This matters more now. Buyers who arrive from an AI assistant have been told you are worth a look, and most of them check. In [TrustRadius's 2026 survey of 1,862 technology buyers](https://www.prnewswire.com/news-releases/trustradius-2026-b2b-buying-disconnect-report-reveals-ai-has-changed-how-buyers-research-but-not-what-they-trust-302825792.html), 63% had used AI during their purchase, 94% of those fact-checked its answers at least sometimes, and 74% used reviews. Much of that checking happens on your website, and the evidence that reassures a person is largely what ChatGPT, Gemini and Copilot look for when deciding whom to name. ## Key takeaways - Buyers judge a B2B website on three questions: can this firm do the work, will it be reliable, and is it safe to deal with. - Specific, named, dated proof from clients like the buyer does more for trust than any design choice. - The strongest proof sentence names the client, says what changed, gives the number and the date, and says how it was measured. - Say plainly who you serve, roughly what it costs and who you are not for, and give a contact page with a phone number, address and company details. - A guarantee builds trust only when it covers something the provider controls, such as meetings held; nobody can guarantee what an AI assistant will say. - The proof that reassures a buyer is the same material AI assistants look for, and review sites are part of it. ## What are the most important trust signals on a B2B website? Specific, named proof from clients like the buyer comes first, then a plain statement of who you serve, real people, visible contact and company details, and honest guidance on price, fit and risk. | Signal | What it proves to the buyer | Where it goes | What undermines it | |---|---|---|---| | Named, dated case studies with a number | You have done this work recently | Service and sector pages | "A leading client", no date, no figure | | Quotes with name, role and firm | A real person stands behind the claim | Next to the claim it supports | "Great to work with", initials only | | A plain statement of who you serve | You are the right fit | The first screen of the homepage | "Businesses of all sizes" | | Real people with photographs | Someone accountable will do the work | Team and service pages | Stock photos, founders only | | Price ranges, timescales, who you are not for | You are confident and honest | Service or pricing pages | "Contact us for pricing" | | Phone, address and company details | You are a real, established business | Contact page and footer | A form and nothing else | | Independent reviews and accreditations | Others confirm what you say | Review sites, linked from your pages | Badges with no link | | Clear terms on risk | Trying you is safe | Service and pricing pages | Promises you cannot control | ## What is a buyer actually checking when they land on your site? Three things, usually in this order: whether you can do the work, whether you will be reliable, and whether it is safe to deal with you. If your site leaves any of them open, the buyer moves on to the next name. - **Can you do the work?** Evidence that you have solved their kind of problem, for their kind of organisation, recently. - **Will you be reliable?** Who they will deal with, how long things take and how you communicate. - **Is it safe to deal with you?** Whether you are a real, established business that handles data properly and can be reached if something goes wrong. Buyers in regulated sectors think about this most. For example, say a facilities manager at a housing association is comparing three building services firms. One site shows a case study for another housing association, a named contracts manager, how a maintenance contract is mobilised and a phone number. The other two talk about quality and partnership. The first gets the call. ## Which proof persuades a B2B buyer most? Proof that is specific, attributed, recent and relevant. One detailed case study from a firm like theirs beats a wall of logos. In rough order of strength: - Case studies that name the client, or describe it precisely, with the problem, what you did and what changed. - Quotes from named people, with role and organisation, saying something specific. - Accreditations that matter to the buyer, such as ISO 27001 for an IT provider. - Awards and press coverage from publications your buyers read. - Client logos, used with permission, that your buyers recognise. Recency matters as much as quality: a case study from four years ago proves you could do the work once, one from last quarter that you are doing it now. Organise proof by sector or problem so buyers find their own. We explain why third-party proof carries so much weight with AI assistants in [why E-E-A-T matters so much to AI assistants](/insights/eeat-why-ai-cares-more). ### What does specific, attributed proof look like? Name the client, say what changed, give the number and the date, and say how it was measured. Compare a vague line with the way two of our own results are written. - Vague: "We helped a membership organisation grow through AI." Specific, for [Helm Club](/work/helm-club), a founder membership community: "1 in 5 new members this year came through an AI recommendation." - Vague: "We transformed a regtech firm's AI visibility." Specific, for [ID-Pal](/work/id-pal), an identity verification firm: "UK buyer questions where AI names ID-Pal: none in May, two in August, four in September 2026." With the method stated: "The same buyer questions run every month in ChatGPT, Gemini and Claude, in a fresh chat. A win only counts when it repeats, and losses are reported too." - Vague: "A pleasure to work with." Attributed: "They work like founders: they get on with it, and they tell you the truth." (Dan Brooksbank, Marketing Director, ID-Pal) A buyer can check a sentence built this way, and an assistant can lift it into an answer whole. In a [2024 study from Princeton and IIT Delhi](https://arxiv.org/abs/2311.09735), adding statistics to a source raised its visibility in AI-generated answers by about 34%, and adding quotations by about 41%. It was tested on a research engine, so treat the figures as a direction. ## Why does clarity build trust? Because vagueness looks like evasion. When a buyer cannot tell within a few seconds what you do, who you do it for and how to start, they assume you either do not know or are hiding something they will not like. Compare these pairs: - "We deliver end-to-end digital transformation solutions" against "We move mid-sized manufacturers from spreadsheets to a single ERP system in six to nine months." - "We work with businesses of all sizes" against "Most of our clients are UK firms with 50 to 500 staff." - "Get in touch" against "Book a 30-minute call with a partner." The second of each is more useful, more believable, and filters out firms that are not for you. A buyer should also reach your services, sectors, examples, people and contact details from any page, in a menu that uses their words. ## Does design really affect whether buyers trust you? Yes, mostly by removing reasons to doubt you. Nobody hires you for a beautiful site, but they may cross you off for a neglected one. The damage comes from simple things: obvious stock photography, broken links and "coming soon" pages, inconsistent fonts and layouts, pop-ups and chat windows that interrupt before the buyer has read anything, and slow pages. We cover speed in [whether speed and accessibility affect enquiries](/insights/website-accessibility-speed-matter), and when a theme used by thousands of firms works against you in [template or custom website](/insights/template-vs-custom-website-b2b). ## Why do real people on the site matter so much? Because B2B services are bought from people. Real photographs, names and a line on each person's experience make the firm feel accountable. In our experience the team page is one of the first a serious buyer opens and one of the most neglected. Show the people who will do the work, not only the founders. A short founder story helps too. For example, "We spent ten years inside large firms watching mid-sized clients get passed to junior staff, so we built a firm where partners do the work" tells the buyer why you exist, where "a leading provider of professional services" tells them nothing. ## Should a B2B firm put its prices on its website? Yes, at least as ranges. Buyers read "contact us for pricing" as "expensive, and we would rather not say". A starting price or typical range for each service, what moves it, what is included and a typical timescale answer the buyer's biggest question and filter out those who could never afford you. We publish all of ours on one page, [what it costs](/what-it-costs), for exactly that reason. Fit works the same way. "We specialise in regulated sectors, so we are rarely the right choice for retail" feels risky to write, but it makes the buyers you want more confident. ## Do guarantees increase enquiries for B2B services? In our view, yes, when the guarantee covers something the provider controls and is written plainly, because it lowers the risk of trying you; we know of no reliable study that measures the effect. What can fairly be guaranteed is effort, reporting and outcomes such as meetings held. Nobody can guarantee what ChatGPT or Gemini will say, so be wary of anyone guaranteeing a place in an AI answer. Our own example is the meetings guarantee in [Pipeline That Fills](/pipeline-that-fills). Month one is warm-up, the criteria for a qualified meeting are agreed with the client first, and if fewer than six qualified meetings are held across months two and three, month four's management fee is waived. Other ways to reduce risk are a short notice period, a fixed scope, and a small paid first step whose output the client keeps either way. Our [Paid Discovery](/discovery) works that way. For [TCC Group](/work/tcc-group) it meant "1 week to a plan TCC's own team built the new site from", and "TCC chose to build it with its own team, and the new tcc.group is live, built to the plan, with nothing about it depending on Fortitude." ## What should a B2B contact page include? The ordinary details that show you are a real, accountable business: - A phone number and an email address, not only a form. - A physical address. - Who will reply and how quickly, for example "a partner will reply within one working day". - A short form, with only the fields you need, and a link to your privacy notice. - Your company details. [GOV.UK guidance](https://www.gov.uk/running-a-limited-company/signs-stationery-and-promotional-material) (updated January 2025) says a limited company's website must show its registered number, its registered office address, where it is registered (England and Wales, Scotland or Northern Ireland) and the fact that it is a limited company, usually through its full name including "Limited" or "Ltd". Many firms put these in the footer of every page. Then keep the promise: a slow reply undoes the trust the website built. ## Do reviews matter for a B2B firm, and where should we collect them? Yes. Review sites are cited in B2B answers, and buyers weigh them. In [G2's April 2026 survey of 1,076 software buyers](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html), 45% said citations of review sites were the most confidence-inspiring signal in an AI answer. In [Profound's 2025 analysis of 680 million citations](https://www.tryprofound.com/blog/ai-platform-citation-patterns), G2 accounted for 1.1% of all ChatGPT citations, level with Forbes, and a [Peec AI study reported by Search Engine Land](https://searchengineland.com/ai-search-engines-cite-reddit-youtube-and-linkedin-most-study-473138) (March 2026) found G2 prominent for B2B questions in Perplexity. Both come from companies that sell AI visibility tools, and G2 surveyed software buyers, so read them as signs. | Your sector | Where to build reviews first | |---|---| | Technology and SaaS | G2 and Capterra, plus Google reviews | | Agencies and consultancies | Clutch, plus Google reviews | | Professional services | Google reviews and your professional body's directory | | Fintech and regtech | G2 for software products, plus Google reviews | | Recruitment | Google reviews and Trustpilot | To get your first five reviews: 1. Pick one platform your buyers use, and make the profile describe the firm exactly as your website does. 2. List ten clients who have had a good result in the last year. 3. Ask at a natural moment, such as the end of a project, in a personal note from the person who did the work, with a direct link. 4. Suggest what to cover, such as the problem and what changed, but never write the review for them or pay for it. 5. Reply to every review, and link the profile from your website. ## Do the same trust signals matter to AI assistants? Largely, yes. An assistant looks for things it can state with confidence: who you serve, what you have done and who says you are good at it. ID-Pal is the warning: "AI could describe ID-Pal. It never recommended it." After its industry pages were rebuilt: "Where a page has been rebuilt, ID-Pal appears. Where it has not yet, it does not, which tells us exactly what to build next." A buyer who arrives from an AI recommendation has already been partly persuaded, and your site's job is to confirm it; we cover that in [what your website should do now that visitors arrive pre-sold](/insights/death-cold-website-visit-ai-funnel). ## What to do this month 1. **Read your homepage as a stranger would.** Check that the first screen says who you serve, what you do and how to start. 2. **Rewrite your best result as one proof sentence.** Client, change, number, date and method, on the page for that service. 3. **Update the team page.** Current people, real photographs and a line on what each does for clients. 4. **Fix the contact page.** Add a phone number, an address, a reply time, a "what happens next" line and your company details. 5. **Publish one honest sentence about price or fit** on your main service page. 6. **Ask three clients for a review** on the platform your buyers use. ## Frequently asked questions ### How many case studies does a B2B website need? Fewer than most firms think. Three to five strong, relevant case studies covering your main services and sectors do more than twenty thin ones. Add one every quarter or two so the newest is never more than a few months old. ### What if our clients will not let us name them? Common in law, accountancy and financial services. Describe the client precisely, for example "a family-owned food manufacturer with three UK sites", be specific about the result, and ask for a quote attributed by role. Specific but anonymous beats vague and named. ### Do client logos help? Only if your buyers recognise the names and you have permission. Unknown logos do little, and logos used without permission do harm. Pair them with detailed case studies. ### Do Clutch reviews help a B2B agency get recommended by ChatGPT? They can help as one consistent source among several, though no study we trust isolates Clutch's effect. Review platforms appear among the sources assistants cite for B2B questions, and for agencies Clutch is the natural one. A profile with a few detailed, recent reviews that describes you as your website does gives an assistant one more source that agrees. It will not get you named on its own. ### How quickly should we respond to an enquiry from the website? Within one working day at the latest. A buyer who has just reached out is at peak interest and may have contacted two other firms. Say on the site how quickly you will reply, and make sure someone owns the inbox. Trust on a website is built from specific, honest, current evidence, and most firms already have that evidence in their inboxes and project files. If you want a website that puts it in front of buyers and AI assistants alike, see how [Website That Sells](/website-that-sells) works, or book a free [Growth Review](/growth-review) for three fixes in writing. ## Sources - [TrustRadius via PR Newswire, 2026 B2B Buying Disconnect Report (July 2026)](https://www.prnewswire.com/news-releases/trustradius-2026-b2b-buying-disconnect-report-reveals-ai-has-changed-how-buyers-research-but-not-what-they-trust-302825792.html) - [G2 via PR Newswire, Half of B2B software buyers now start their research with AI chatbots (April 2026)](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html) - [Profound, AI platform citation patterns (June 2025)](https://www.tryprofound.com/blog/ai-platform-citation-patterns) - [Search Engine Land, AI search engines cite Reddit, YouTube and LinkedIn most: study (March 2026)](https://searchengineland.com/ai-search-engines-cite-reddit-youtube-and-linkedin-most-study-473138) - [GOV.UK, Running a limited company: signs, stationery and promotional material (January 2025)](https://www.gov.uk/running-a-limited-company/signs-stationery-and-promotional-material) - [Aggarwal and others, Princeton and IIT Delhi, study of visibility in AI-generated answers (KDD 2024)](https://arxiv.org/abs/2311.09735) --- Title: What should my website do now that visitors arrive pre-sold by AI? URL: https://fortitudemedia.ai/insights/death-cold-website-visit-ai-funnel Author: Ross Williams Published: 2026-03-31 Last reviewed: 2026-09-24 Topic: How buying is changing Short answer: Your website's job changes from persuading strangers to confirming a recommendation. A visitor sent by ChatGPT, Gemini or Copilot already knows roughly what you do and is comparing you with the other firms the assistant named. So confirm the claim in the first screen, answer price, timescale and fit plainly, show a fair comparison, and make it easy to book a call with a person. Track these visitors with a GA4 channel group and by asking every enquiry. Look at your website analytics for the last year and you may see two things at once. Visits are flat or falling. Enquiries, if AI assistants are naming you, are holding up or getting better. Founders often read that as a contradiction. It is the funnel changing shape underneath you. Plenty of first visits to a B2B website were always warm: referrals, people who already knew your name. But a large share were cold. Someone clicked an advert or a search result, knew nothing about you, and spent ninety seconds deciding whether you were worth a second look. Now more buyers ask an assistant first. In [G2's April 2026 survey of 1,076 software buyers](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html), 51% said they start their research in an AI chatbot, up from 29% a year earlier. And when Google shows an AI summary, people click a traditional result far less: on 8% of visits against 15% without one, in a [Pew Research Center study of US users (July 2025)](https://www.pewresearch.org/short-reads/2025/07/22/google-users-are-less-likely-to-click-on-links-when-an-ai-summary-appears-in-the-results/). So the visitors who do arrive have often been sent. For [Helm Club](/work/helm-club), a peer group community for founders, "1 in 5 new members this year came through an AI recommendation." Those visitors land already believing you are credible. They are there to check, compare and find out how to get in touch, and a website built to persuade strangers handles them badly. ## Key takeaways - A visitor sent by an AI assistant has already been told you are credible, so your website's job is to confirm that quickly rather than argue for it. - Fewer visits alongside steady enquiries can be a sign of health, so report qualified conversations by source, not visits. - The pages that matter most are pricing, timescales, fit, proof and comparison, because those are the questions the assistant could not settle for the buyer. - AI-referred visitors arrive holding a shortlist, so a fair, factual "how we compare" page answers the question they came with. - GA4 now has an AI Assistant channel, but it misses Claude and Perplexity and counts AI Overviews as organic, so set up a custom channel group as well. - Researched outbound still matters, because it reaches the buyers no assistant has sent you yet. ## What actually changes in the funnel? The old funnel ran awareness, visit, form, conversation, close. The new one runs research, recommendation, visit, conversation, close, and by the time a person reaches your site an assistant has done the first two stages for them. That has three consequences. The top of the funnel is narrower, because the assistant filters before you see the buyer. The people who arrive are closer to a decision, because the assistant has already answered "does this firm do what I need" and "are they credible". And you no longer control the first impression: the assistant wrote it, from what it could find about you. The practical result is fewer visits and, if you are being named, better ones. That is why we tell founders to stop reporting visits at board meetings and start reporting qualified conversations. If your traffic is falling and you want to know whether AI answers are replacing clicks, start with [why traffic falls when rankings have not](/insights/zero-click-search-your-business). ## Who is the AI-referred visitor, and what do they want? They asked a specific question with constraints, such as sector, location, size or a must-have, and the assistant named you and a few others with reasons. They are comparing, not browsing. | | Visitor from a Google ad | Visitor sent by an AI assistant | |---|---|---| | What they know about you | Often nothing | Roughly what you do, and why you were named | | Who they are comparing you with | Unknown | The other firms the assistant named | | How close to buying | Anywhere from curious to ready | Usually closer: people rarely ask for a shortlist for fun | | What they came to do | Work out whether you are relevant | Confirm the claim, settle price, timescale and fit, compare, and find the next step | Everything on the site should serve one of those four jobs: confirm, answer, compare, contact. ## What should the website show them? Confirmation, then the specific answers the assistant could not give. The old job was persuasion; the new one is proof. A persuasion-first site has a big hero statement, a long feature list, testimonials chosen for warmth, "contact us for pricing" and forms on every page. A confirmation-first site states plainly who you serve and what you are known for, in the words an assistant would use, and then answers the visitor's remaining questions page by page: | Page | The question it answers for an AI-referred visitor | An example from a real site | |---|---|---| | Homepage, first screen | Is this the firm the assistant described? | Our own homepage says who we are for in one line: founder-led B2B firms | | Sector pages | Do they really work with firms like mine? | [ID-Pal](/work/id-pal): "Each opens with the question a compliance lead types, names the regulator, and answers inside the first ninety words." | | Service pages | Do they do the exact thing I need? | [arum](/work/arum): "Services organised the way a client buys them: five stages from advice to optimisation, plus three specialist services." | | Case studies | Have they done it before, with results? | arum: "Each of the 12 case studies leads with a number, and each of the nine sector pages names the regulation in play." | | Pricing | Roughly what will it cost, and what drives it? | Our [what it costs](/what-it-costs) page publishes every fee | | How it works | How fast could they start, and what happens first? | Our [Paid Discovery](/discovery) page states that it takes one week and what you receive | | How we compare | How do they stack up against the others on my list? | Our [comparison pages](/vs) set us against hiring a team, a lead generation agency, an SEO agency and doing nothing | | Team | Who would I actually deal with? | Named founders with their own pages, such as [Ashley Kenerson](/team/ashley-kenerson) | | Contact | How do I talk to a person? | Our [Growth Review](/growth-review) page books 30 minutes with David Adams directly | Add a short, plain page on who you are not the right fit for. It saves both sides time and makes the rest of the site more believable. The full picture of what a site needs is in [what a B2B website should include](/insights/what-should-a-b2b-website-include). For example, say a managed IT provider is named by Gemini for "IT support for a 150-person law firm in Leeds". The visitor wants three things: do they really work with law firms, what will it cost, and how quickly could they take over from the incumbent. A homepage about "empowering your digital transformation" answers none of them. A page called "IT support for law firms", with two case studies, a price band and a thirty-day onboarding plan, answers all three, and the visitor books a call. ## Should we build a "how we compare" page? Yes, if buyers compare you with named alternatives, and AI-referred buyers usually do. The assistant has handed them a shortlist, so a page that sets out the differences fairly answers the question they arrived with, and gives the assistants a clear source for the comparison next time. Keep it factual and fair: - Who each option is best for, including cases where a competitor is the better choice. - How each is priced: fixed fee, retainer, per outcome, and what is included. - Timescales: how long to start and how long to see results. - What the buyer has to do themselves in each case. - Contract terms, such as minimum periods and notice. - Dated facts with sources, reviewed at least twice a year, and nothing you could not defend to the competitor. Build one page per real question buyers ask, not one per phrasing. [Google's guide to its AI features](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide) warns that creating pages for every query variation, mainly to manipulate AI answers, breaks its scaled content abuse policy. Helm Club shows what the approach can do: "We built the structured data AI engines read, then a comparison page for every question a founder asks when weighing up peer groups: Helm against YPO, Vistage and the rest." And: "Every comparison page we have published ranks first for its question on Google and the AI engines. The pages not yet built are the ones where a competitor still writes the answer." ## Why is conversion simpler, and harder? Simpler, because you know why the visitor came, what they want and who they are comparing you with. The discipline comes down to two questions: is our answer clear, and is it easy to talk to us? Harder, because you cannot make up for a credibility gap on the page. The assistant did the credibility work before the visit, and if you were not credible to it, the visitor never arrives. You also cannot hide weaknesses. The visitor has other names in front of them, so vague pricing, old case studies or a slow-sounding start send them to the next firm, and those were people ready to buy. People who arrive convinced are checking for reasons to doubt; [what makes a buyer trust your website](/insights/psychology-of-trust-website-conversions) covers the signals that reassure them. Firms that close on persuasion alone, hide their prices or cannot say in one sentence why a buyer should choose them will find this harder. Firms with named experts, published prices and timescales, and a clear reason to choose them will find it easier. ## What does this mean for our sales team? Their week moves from finding and persuading towards validating, handling specifics and closing. A buyer sent by an assistant arrives with the assistant's reasoning and wants it confirmed, so first calls start further along. - Hire and train for domain knowledge, because the questions are now specific and technical. - Feed every question the website could not answer back to marketing as a page to write. - Report meetings held and deals closed by source, not calls made. - Keep researched outbound running. The assistants only send buyers who ask; outbound reaches the ones who have not asked yet, or who asked and were not told about you. Our [Pipeline That Fills](/pipeline-that-fills) does this with every message written and approved by a person. ## How do we know whether this is already happening? Four checks in your existing data will tell you, and the first takes an hour. ### Step 1: Set up GA4 to separate AI referrals GA4 added a default "AI Assistant" channel on 13 May 2026, covering ChatGPT, Gemini, DeepSeek, Copilot and Grok, but not Claude or Perplexity ([Nice Looking Data, July 2026](https://www.nicelookingdata.com/blog/ga4-ai-traffic-chatgpt-referrals)). Add your own custom channel group so you see them all: in Admin, go to Data display, then Channel groups, create a channel called "AI referrals" where the session source matches chatgpt.com, perplexity.ai, claude.ai, copilot.microsoft.com or gemini.google.com, and place it above the Referral channel. Since mid-2025 OpenAI has added a utm_source tag, showing as chatgpt.com or chatgpt, to a growing share of ChatGPT's outbound links, so match both values. Clicks from AI Overviews and AI Mode still count as Organic Search, so check Search Console's [generative AI performance reports](https://developers.google.com/search/blog/2026/06/gen-ai-performance-reports) for those. ### Step 2: Ask every enquiry Add "how did you hear about us" with an "AI assistant such as ChatGPT" option to every form, and ask on every first call. Analytics will always miss some of these visits. ### Step 3: Compare conversion by source Compare enquiry-to-meeting and meeting-to-deal rates for AI-referred leads against paid search, referrals and outbound. ### Step 4: Compare cost and deal quality by source Include your team's time, not just media spend, and look at average deal value, retention and how much negotiation each deal needed. If AI-referred enquiries are fewer but convert better, you are already in the new funnel. Our guide to [telling which leads came from ChatGPT](/insights/attribution-ai-era-track-leads) covers attribution in more depth. ## What to do this month 1. Set up the "AI referrals" channel group in GA4 and add the AI assistant option to every enquiry form. 2. Read your homepage as a visitor who has just been told you are a good fit. If the first screen does not confirm it, rewrite the headline and first paragraph in plain words. 3. Publish a pricing page with honest ranges and what drives them, and a "how it works" page with timescales. 4. List the firms an assistant names alongside you, and draft a fair "how we compare" page. 5. Sit with your sales lead and list every question buyers ask on first calls. Each one your website does not answer is a page to write this quarter. ## Frequently asked questions ### Do visitors from ChatGPT convert better than Google traffic? There is no reliable public figure for B2B firms yet, so measure your own. Google says clicks from results pages with AI Overviews are "higher quality", with people spending longer on the site ([Google Search Central, May 2025](https://developers.google.com/search/blog/2025/05/succeeding-in-ai-search)), but gives no number. Compare enquiry-to-meeting rates by source over at least a quarter before drawing conclusions. ### Why does ChatGPT traffic show as direct in GA4? Because the referrer is often lost. The ChatGPT apps on phones usually open links in the phone's browser, and that hand-off often drops the information about where the visitor came from; people also copy and paste links. Those visits land in Direct. The utm_source tag on ChatGPT's links recovers some of them, and asking on your forms and calls recovers more. ### What should a B2B pricing page include? Ranges rather than a single figure, what drives the price up or down, what is included and what is extra, the minimum engagement, how billing works, and typical timescales. If the final figure depends on scope, say which factors decide it. A page like this lets an assistant describe your pricing accurately and lets a visitor rule you in. ### Should our homepage change for visitors referred by AI? Yes, mainly the first screen. It should say who you serve and what you are known for in the words an assistant would use, then point to proof and a way to book a call. That does not harm a cold visitor, while heavy persuasion, forms everywhere and hidden prices lose the pre-sold one. ### Our traffic is down. Is that a problem? Only if enquiries and deals are down with it. If visits fall while conversations hold or rise, the mix is shifting from cold visitors to pre-qualified ones. Report qualified conversations and deals by source to your board instead of visits. If both are falling, that is a different problem and worth a proper look. ### Should we stop paying for Google Ads? Not immediately. Ads still bring traffic, and some of it converts. Measure cost per customer and deal quality by channel, and move budget over time towards the work that gets you named. Our guide to [whether being named by ChatGPT is cheaper than Google Ads](/insights/economics-ai-visibility-vs-advertising) sets out how to compare them. ### What if our service needs a lot of explaining? Then do the explaining in public, on your site and in your published writing, where an assistant can read it. An assistant is only informed by what is published. Firms with complex offers that write plainly about them stand out, because many competitors do not. The cold visit has not gone, but it is no longer the only kind that matters. Make sure your website confirms what the assistant said about you, and that a pre-sold buyer can reach a person in one click. [Website That Sells](/website-that-sells) starts with exactly that audit. ## Sources - [G2 via PR Newswire, Half of B2B software buyers now start their research with AI chatbots (April 2026)](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html) - [Pew Research Center, Google users are less likely to click on links when an AI summary appears (July 2025)](https://www.pewresearch.org/short-reads/2025/07/22/google-users-are-less-likely-to-click-on-links-when-an-ai-summary-appears-in-the-results/) - [Nice Looking Data, AI traffic in GA4: ChatGPT referrals and the AI Assistant channel (July 2026)](https://www.nicelookingdata.com/blog/ga4-ai-traffic-chatgpt-referrals) - [Google Search Central Blog, Generative AI performance reports in Search Console (June 2026)](https://developers.google.com/search/blog/2026/06/gen-ai-performance-reports) - [Google Search Central, Optimizing your website for generative AI features on Google Search (July 2026)](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide) - [Google Search Central Blog, Top ways to ensure your content performs well in Google's AI experiences on Search (May 2025)](https://developers.google.com/search/blog/2025/05/succeeding-in-ai-search) --- Title: Which accountant does ChatGPT recommend, and why is it not us? URL: https://fortitudemedia.ai/insights/ai-optimisation-accounting-practices Author: Ross Williams Published: 2026-03-31 Last reviewed: 2026-09-24 Topic: Sector guides Short answer: ChatGPT recommends practices it can match to a client's situation and check. Lead with a clear specialism by sector or client type, show named partners with ICAEW, ACCA, ICAS or CIOT credentials, and keep professional and software directory listings consistent. Then publish dated, plain-English explainers on the changes owners ask about, such as Making Tax Digital thresholds and Companies House identity checks, and update them whenever HMRC or Companies House guidance changes. The owner of a fast-growing online retailer is frustrated. Her accountant takes a week to reply, turnover has just passed the £90,000 VAT registration threshold, and she is not sure her stock accounting is right. A few years ago she would have asked other founders for a name. Now she asks ChatGPT: "Which accountants in Bristol specialise in ecommerce businesses and work with Xero?" The assistant names three firms with a line about each. The same thing is happening across the owner-managed businesses that make up most practices' client base. A software founder asks Perplexity which accountants in London specialise in SaaS companies and R&D tax relief. A landlord asks Copilot whether Making Tax Digital applies to him. That last question is not rare: HMRC said in February 2026 that more than 860,000 sole traders and landlords had to start digital reporting from 6 April 2026 ([HMRC, February 2026](https://www.gov.uk/government/news/act-now-864000-sole-traders-and-landlords-face-new-tax-rules-in-two-months)), and by August more than 570,000 had signed up and more than 436,000 had sent their first quarterly update ([HMRC, August 2026](https://www.gov.uk/government/news/436000-sole-traders-and-landlords-make-their-tax-digital)). Each is a likely question for an assistant, and a prospective client. Accountancy is still a referral business, but more referrals now pass through an assistant first, and the practice that is not named never hears about the enquiry. The assistants answer basic tax questions themselves; what they cannot supply is judgement for a specific kind of client, which is what the practices they recommend have written down. This guide is for the managing partner or practice owner who wants to be one of those names. ## Key takeaways - AI assistants recommend accountants they can match to a client's sector, size and situation, so a clear specialism matters more than a long services list. - Named partners with ICAEW, ACCA, ICAS or CIOT credentials, and accurate professional directory listings, give assistants independent records to check. - Dated, plain-English explainers on Making Tax Digital, Companies House identity checks, VAT and R&D relief make your practice the source assistants cite when owners ask what changed. - The FCA is due to take over anti-money laundering supervision of accountants once Parliament legislates, so state your current supervisor accurately and date the statement. - Syndicated tax newsletters shared by hundreds of practices tell an assistant nothing about yours; replace them over time with sector pages only you could write. - Publishing how your fees work and how switching accountant works removes the two biggest reasons owners hesitate to get in touch. ## What do business owners ask ChatGPT when they need an accountant? They describe a business, a problem and often a frustration with the current adviser. These are typical of the questions that decide who gets named. | Kind of question | Example | What the answer leans on | |---|---|---| | Finding | "Which accountants specialise in SaaS companies and R&D tax credits in London?" | Sector pages, partner profiles, directories | | Finding | "Best accountants for dental practices in the North West" | Sector pages, reviews, sector press | | Deciding | "Do I need to sign up for Making Tax Digital if I earn £40,000 from property?" | HMRC guidance, dated practice explainers | | Deciding | "How do I verify my identity at Companies House as a director?" | GOV.UK guidance, practices that are authorised agents | | Deciding | "Is it worth switching from a high street accountant to an online accountant?" | Practices that explain fees, service and switching plainly | | Checking | "Is [your practice] any good?" | Reviews, directories, press, your team pages | Most practices think only about the finding questions, yet the deciding questions are where owners judge who knows what they are talking about. ## Why does generic tax content no longer earn you anything? Because an assistant can already give the general answer, so a general article rarely gets quoted. Many practices also publish the same syndicated articles supplied by their website provider, word for word, alongside hundreds of other firms. To an assistant, that content says nothing about your practice. What assistants cannot produce is applied judgement: how the rules play out for a dental practice with associates, what a growing ecommerce business gets wrong about stock and VAT, when a landlord should incorporate and when not. A practice that writes that down, from real experience with that kind of client, gives the assistant something no one else has. We explain the wider pattern in [why ChatGPT never quotes most firms' blogs](/insights/why-ai-ignores-most-blog-posts). ## What does an AI assistant check before it names a practice? Broadly, five things: your specialism, your credentials, independent listings, what clients say, and whether it all agrees. - **Specialism.** The sectors and client types you know best, and the services, such as audit, tax advisory, outsourced finance director support or R&D claims. "Accountants for small businesses" matches nothing; "accountants for dental practices across the North West, from associates to multi-site groups" matches a real question. - **Credentials.** Name partners and senior staff with their qualifications and professional body: ICAEW, ACCA, ICAS or CIMA for accountants, CIOT or ATT for tax. Say whether the firm holds practising certificates and is a registered auditor, and who supervises it for anti-money laundering. - **Independent listings.** Professional and software partner directories, awards such as the British Accountancy Awards and coverage in AccountingWEB, Accountancy Age or your regional business press describe you from the outside. - **Reviews.** Recent, specific Google and Trustpilot reviews help most: "sorted out our VAT registration and explained it properly" says more than five stars. Reply without disclosing anything about a client's affairs. - **Consistency.** Many practices have merged or joined groups, leaving old names, offices and departed partners on directories for years. If an assistant finds you under three names, it struggles to connect the evidence. We explain how to fix that in [why ChatGPT describes a firm wrongly](/insights/ai-conflicting-information-online). ## Which practice directories matter, and what should each listing contain? The professional body directories and the software partner directories, because they are independent records an assistant can check against your website. Every listing should use your exact trading name and address and name the same specialisms your website does. | Directory | Who is listed | What your listing should contain | |---|---|---| | ICAEW [Find a Chartered Accountant](https://www.icaew.com/about-icaew/find-a-chartered-accountant) | ICAEW member firms | Trading name, offices, services and sectors matching your website | | ACCA [Find an ACCA accountancy firm](https://www.accaglobal.com/gb/en/member/find-an-accountant.html) | ACCA member firms | The same, plus the partners who hold practising certificates | | ICAS [Find a CA](https://www.icas.com/find-a-ca) | ICAS firms | The same, with Scottish offices named consistently | | Xero Advisor Directory | Xero partner practices | Sectors, specialisms and the apps you support, in plain words | | QuickBooks ProAdvisor directory | QuickBooks ProAdvisors | Certifications, sectors and services | | Sage's accountant partner listings | Sage partner practices | Software supported and client types | Check each one after any merger, office move or partner change. ## What should we say about our anti-money laundering supervisor? Name your current supervisor accurately and date the statement, because the supervisor is changing. The government decided in October 2025 that the FCA will take over anti-money laundering supervision of the legal and accountancy sectors from the professional bodies ([HM Treasury, October 2025](https://www.gov.uk/government/consultations/reforming-anti-money-laundering-and-counter-terrorism-financing-supervision)). It also said implementation "requires primary legislation" and that the start date "will be heavily dependent on the availability of parliamentary time", and it expects the professional bodies to continue "in their important wider regulatory and representative roles". So today, say who supervises you (for example ICAEW or ACCA), keep your professional body membership prominent, and update the page when the transfer happens. Tax advice is changing too. Tax advisers who deal with HMRC on behalf of clients must now register with HMRC and meet minimum standards, starting from May 2026 ([HMRC, November 2025](https://www.gov.uk/government/publications/mandatory-tax-adviser-registration-with-hmrc)), with registration windows running into 2027 ([HMRC, August 2026](https://www.gov.uk/government/news/second-registration-window-now-open-for-tax-advisers)). Once registered, say so. ## How do we build a sector page that shows we understand a client's world? A sector page is where your specialism becomes visible, and it is often the page an assistant draws on when an owner asks for an accountant in their industry. ID-Pal, which sells identity checks to accountants among others, shows the method. From the [ID-Pal case study](/work/id-pal): "Five industry pages rebuilt: fund administration, wealth management, legal, accountancy and estate agency. Each opens with the question a compliance lead types, names the regulator, and answers inside the first ninety words." And: "Where a page has been rebuilt, ID-Pal appears. Where it has not yet, it does not, which tells us exactly what to build next." ### Step 1: Open with the client's question, in their language For dentists that means NHS contracts, UDAs, associates and practice purchases. For SaaS it means R&D relief, recurring revenue, EMI options and investor reporting. An owner reading the first paragraph should feel you have sat in their office. ### Step 2: Set out the issues that matter most The five or six problems you see most with that kind of business, and how you deal with each, saying where the answer depends on circumstances. ### Step 3: Explain how you work with them The software you use, how often you report, the monthly or quarterly rhythm and who they will deal with. ### Step 4: Show evidence Named clients who agree to it, anonymised examples of the situation and outcome, and sector memberships. Only include figures you can stand behind. ### Step 5: Explain fees How you price for a client of that type and size, and what is included. For example, a practice with forty dental clients among four hundred will do more for its visibility with dentists through one proper dental page, written by the partner who looks after them, than through a hundred general articles. ## Which HMRC and Companies House changes should we be explaining now? The ones your clients are asking assistants about this year. Publish a short explainer soon after each change, link to the official guidance, put the date at the top and add an "updated" line when guidance changes, rather than writing a new post. | Change | What applies | Source | |---|---|---| | Making Tax Digital for Income Tax | Required from 6 April 2026 where qualifying income was over £50,000 in 2024 to 2025; from 6 April 2027 over £30,000 in 2025 to 2026; from 6 April 2028 over £20,000 in 2026 to 2027 | [HMRC, March 2026](https://www.gov.uk/guidance/check-if-youre-eligible-for-making-tax-digital-for-income-tax) | | Companies House identity verification | A legal requirement since 18 November 2025; existing directors give their personal code with the next confirmation statement, within a 12-month transition | [Companies House, July 2026](https://www.gov.uk/guidance/when-you-need-to-verify-your-identity-for-companies-house) | | VAT registration threshold | £90,000 of taxable turnover since 1 April 2024, up from £85,000 | [HMRC, March 2024](https://www.gov.uk/government/publications/vat-increasing-the-registration-and-deregistration-thresholds) | | R&D tax relief | A merged expenditure credit scheme at 20% for accounting periods beginning on or after 1 April 2024, plus enhanced support for loss-making SMEs whose R&D spend is at least 30% of total expenditure | [HMRC, January 2026](https://www.gov.uk/guidance/research-and-development-rd-tax-relief-the-merged-scheme-and-enhanced-rd-intensive-support) | Qualifying income for Making Tax Digital is self-employment and property income added together, before expenses, and excludes employment income and dividends ([HMRC, September 2026](https://www.gov.uk/guidance/work-out-your-qualifying-income-for-making-tax-digital-for-income-tax)). For example, a loss-making software company whose relevant R&D expenditure is at least 30% of its total expenditure may qualify for enhanced R&D intensive support, while one below that line claims under the merged scheme. A page that walks through that test in plain words is what a SaaS founder's assistant will quote. Keeping these pages current matters, as we explain in [how old is too old for content](/insights/content-freshness-recency-ai). ## How should we explain fees and switching? Answer both on your website, because not knowing the cost and fearing the hassle of switching are the two biggest reasons owners hesitate. Explain your packages plainly (bookkeeping, VAT returns, payroll, year-end accounts, tax returns and advisory time), say what is included and what costs extra, and publish typical starting points for your most common client types. Then describe switching: professional clearance, how you get the information you need from the previous accountant, how long it takes and what happens in the first month. Prospects then arrive having decided they can afford you. ## How do we tell whether it is working? Keep a fixed list of 20 to 50 questions covering your sectors, services and towns, run each several times a month in a clean session in ChatGPT, Google's AI Overviews and AI Mode, Copilot, Gemini, Perplexity and Claude, and count a win only when it repeats, because answers vary between runs. Record whether you are named, how you are described and which sources are cited, and add "AI assistant such as ChatGPT" to your enquiry form. Assistants that search the web live can reflect a corrected listing or a new sector page within weeks; answers drawn from a model's own memory change only when it is retrained, which takes months. Our guide to [checking whether ChatGPT recommends your firm](/insights/track-ai-recommending-your-business) sets out the method. ## What to do this month 1. Choose the one or two sectors where you have the most clients and experience, and write a proper sector page for each using the five steps above. 2. Update every partner's profile with qualifications, professional body, sector focus and a short biography in plain text. 3. Check your ICAEW, ACCA or ICAS listing and your Xero, QuickBooks and Sage listings, and correct old names, offices and people. 4. Publish or update dated explainers on Making Tax Digital thresholds and Companies House identity verification, linked to GOV.UK. 5. State your anti-money laundering supervisor and your HMRC tax adviser registration on your website, with the date. 6. Publish a fees and switching page for your most common client type. ## Frequently asked questions ### Do I need to sign up for Making Tax Digital if I earn £40,000 from property? It depends on the year, and a practice should answer this on its own site. Example: a landlord whose only qualifying income is £40,000 of rent a year was under the £50,000 test for 2024 to 2025, so did not have to join in April 2026. Because £40,000 is over the £30,000 test for 2025 to 2026, HMRC's guidance says Making Tax Digital applies from 6 April 2027. HMRC checks qualifying income from each year's Self Assessment return. ### How do I verify my identity at Companies House as a director? Through GOV.UK One Login or through an Authorised Corporate Service Provider, also called a Companies House authorised agent. New directors verify on appointment or within 14 days, depending on their situation; existing directors give their personal code with the company's next confirmation statement. If your practice is an authorised agent, say so on your website, because owners ask assistants who can do it for them. ### How much does an outsourced finance director cost per month in the UK? Owners ask this often, and assistants can only quote figures they can find. We do not quote market rates here, because they vary with days per month, sector and the seniority of the person. If you offer outsourced finance director support, publish how you price it, what a typical engagement includes and what changes the cost. ### Is it worth switching from a high street accountant to an online accountant? Assistants answer this by comparing price, software, responsiveness and depth of advice. Online accountants usually compete on price and software. A local or specialist practice competes on judgement and sector knowledge, so say plainly what a client gets from you that a fixed-price online service does not provide, and publish how switching works. ### We use syndicated articles from our website provider. Is that a problem? It is not harmful in itself, but it does little for you, because the same articles appear on many other practices' sites. Keep them if clients find them useful, and put your effort into content only you could write: sector pages, plain explanations of rule changes and guides built from questions your team answers every week. ### Our practice has joined a larger group. What should we do about our name? Decide how you want to be known, then make every listing match. If you keep your local brand, say clearly that you are part of the group. If you move to the group name, update your website, Google Business Profile, professional and software directories, and explain the change on your site so clients and assistants can connect the old name to the new one. If your website is mostly template content and a contact form, [Website That Sells](/website-that-sells) turns it into something assistants can read and quote. A free [Growth Review](/growth-review) is 30 minutes with David Adams, with three fixes in writing. ## Sources - [HMRC, Act now: 864,000 sole traders and landlords face new tax rules in two months (February 2026)](https://www.gov.uk/government/news/act-now-864000-sole-traders-and-landlords-face-new-tax-rules-in-two-months) - [HMRC, 436,000 sole traders and landlords make their tax digital (August 2026)](https://www.gov.uk/government/news/436000-sole-traders-and-landlords-make-their-tax-digital) - [HMRC, Check if you're eligible for Making Tax Digital for Income Tax (March 2026)](https://www.gov.uk/guidance/check-if-youre-eligible-for-making-tax-digital-for-income-tax) - [Companies House, When you need to verify your identity for Companies House (July 2026)](https://www.gov.uk/guidance/when-you-need-to-verify-your-identity-for-companies-house) - [HM Treasury, Reforming anti-money laundering and counter-terrorism financing supervision: consultation response (October 2025)](https://www.gov.uk/government/consultations/reforming-anti-money-laundering-and-counter-terrorism-financing-supervision) - [HMRC, Mandatory tax adviser registration with HMRC (November 2025)](https://www.gov.uk/government/publications/mandatory-tax-adviser-registration-with-hmrc) - [HMRC, VAT: increasing the registration and deregistration thresholds (March 2024)](https://www.gov.uk/government/publications/vat-increasing-the-registration-and-deregistration-thresholds) - [HMRC, R&D tax relief: the merged scheme and enhanced R&D intensive support (January 2026)](https://www.gov.uk/guidance/research-and-development-rd-tax-relief-the-merged-scheme-and-enhanced-rd-intensive-support) --- Title: Which AI assistants do buyers actually use, and how do they differ? URL: https://fortitudemedia.ai/insights/ai-search-landscape-chatgpt-perplexity-google-claude Author: Ashley Kenerson Published: 2026-03-31 Last reviewed: 2026-09-24 Topic: How buying is changing Short answer: ChatGPT and Google's AI answers (AI Overviews and AI Mode) reach the most buyers, followed by Microsoft Copilot, which many buyers have at work, and Gemini, then Perplexity and Claude. They differ in which search index they use (Google's, Bing's or their own), whether they show sources, and how fast they reflect a change. All of them reward the same foundation: a website crawlers can read, the same facts on independent sources, reviews and named experts. Check them all with one fixed list of buyer questions, then fix what they have in common. A prospect tells you they "asked AI" for a shortlist. Which one they asked matters. The answer a buyer gets from ChatGPT can differ from the one Microsoft Copilot gives inside their company's Microsoft 365, and both can differ from the AI Overview at the top of Google. One might name you, one a competitor, and one nobody. The shift is big enough to plan for. In [G2's April 2026 survey of 1,076 software buyers](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html), 51% said they now start their research in an AI chatbot, up from 29% a year earlier. ChatGPT alone passed [one billion weekly users in August 2026](https://openai.com/index/expanding-access-to-ai-with-chatgpt-ads/), according to OpenAI. You cannot court six assistants separately. This guide sets out who uses which, how they differ, and the one body of work that helps with all of them, voice assistants included. ## Key takeaways - ChatGPT and Google's AI answers reach the most people, Microsoft Copilot is what many buyers have at work, and Gemini, Perplexity and Claude complete the list. - The assistants differ in the search index they draw on, whether they show their sources, and how quickly they reflect a change to your site. - Ranking on Google helps with AI Overviews but is no longer a gate: in March 2026 only 38% of AI Overview citations came from top-ten pages. - Copilot builds its web answers from Bing, so a firm that has ignored Bing can be named by ChatGPT and missed by Copilot. - Which assistant names you and which does not points to the specific gap to fix. - One foundation works for all of them, voice included: a website crawlers can read, the same facts on independent sources, reviews, and named experts. ## Which AI assistant do B2B buyers use most? ChatGPT has the largest audience of any standalone assistant, and Google's AI answers reach the most people overall because they sit inside Google Search. Behind them come Microsoft Copilot, which many buyers have at work, and Gemini, then Perplexity and Claude. No company publishes UK-only figures, and we know of no UK survey of which assistant B2B buyers use to shortlist. The best guide is each company's own announcements, which we check every quarter. | Assistant | Latest published figure | Who said it, and when | |---|---|---| | ChatGPT | More than 1 billion weekly users | OpenAI, August 2026 | | Google AI Overviews | More than 2.5 billion monthly users | [Alphabet, June 2026](https://blog.google/alphabet/investor-presentation-june-2026/) | | Google AI Mode | More than 1 billion monthly users | Alphabet, June 2026 | | Gemini app | More than 900 million monthly users | Alphabet, June 2026 | | Microsoft 365 Copilot | More than 30 million paid business seats | Microsoft, July 2026 | | Claude | No user figure published; 9.3% of visits to AI chatbot websites | Similarweb, September 2026 | | Perplexity | No 2026 user figure published; 0.9% of visits to AI chatbot websites | Similarweb, September 2026 | The units differ, so the rows cannot be compared directly, and [Similarweb](https://x.com/Similarweb/status/2096878021378466096) counts website visits only, leaving out apps and AI built into other products. Business buyers also behave differently from consumers: [Forrester found in January 2026](https://www.forrester.com/blogs/b2b_buyers_make_zero_click_buying_number_one/) that they were twice as likely to use ChatGPT and four times as likely to use Microsoft Copilot. So work in this order: ChatGPT and Google, then Copilot and Gemini, then Perplexity and Claude. The wider change in how buyers choose is in [how AI is changing B2B supplier choice](/insights/ai-search-change-your-industry). ## How do the assistants differ? They differ in where their web results come from, whether they show their sources, and how fast a change to your site can appear. Every assistant mixes what its model learned in training with what it fetches when the question is asked. Answers from a live search can reflect a new page or article within weeks; answers from the model's memory change only when it is retrained, which takes months. | Assistant | Where its web results come from | Shows its sources? | Search crawler to allow | How fast a change can show | |---|---|---|---|---| | ChatGPT | Third-party search providers (reported to include Bing) and OpenAI's own crawler | Yes, when it searches | OAI-SearchBot | Weeks; months for the model's memory | | Google AI Overviews and AI Mode | Google's index | Yes | Googlebot | Weeks, once Google recrawls the page | | Gemini | Google's index | Often, when it searches | Googlebot | Weeks; months for the model's memory | | Microsoft Copilot | Bing | Yes | Bingbot | Weeks, once Bing recrawls the page | | Claude | Anthropic's web search | Yes, when it searches | Claude-SearchBot | Weeks; months for the model's memory | | Perplexity | Its own index, searched for every question | Always, numbered | PerplexityBot | Weeks | Each company documents the crawler behind its search answers: [OpenAI says](https://help.openai.com/en/articles/9237897-chatgpt-search) a site must allow OAI-SearchBot to appear in ChatGPT search, and [Anthropic](https://support.claude.com/en/articles/8896518-does-anthropic-crawl-data-from-the-web-and-how-can-site-owners-block-the-crawler) and [Perplexity](https://docs.perplexity.ai/guides/bots) name Claude-SearchBot and PerplexityBot. Blocking a training crawler such as GPTBot does not stop you being cited; blocking a search crawler does. Cloudflare [changed its default to block AI crawlers in July 2025](https://blog.cloudflare.com/content-independence-day-no-ai-crawl-without-compensation/), so check that setting too. The detail is in [how AI crawlers read your website](/insights/ai-crawlers-vs-google-spiders). ## Which search engine does ChatGPT use for web results? OpenAI says ChatGPT search uses third-party search providers, alongside its own crawler and content from partners, without naming the providers. Microsoft's Bing was [reported as one of them when ChatGPT search launched](https://searchengineland.com/chatgpt-search-officially-launches-447919) in October 2024. So look at how your firm appears in Bing as well as Google: submit your site in Bing Webmaster Tools and claim your Bing Places listing. The same work helps with Copilot. ## How do Google AI Overviews and AI Mode choose their sources? Both build answers from Google's index using what Google calls query fan-out: several related searches for one question, with results drawn from all of them. That is why ranking helps but no longer decides. [Ahrefs found in March 2026](https://ahrefs.com/blog/ai-overview-citations-top-10) that only 38% of AI Overview citations came from pages ranking in the top ten, down from 76% in July 2025. AI Overviews are the summary at the top of an ordinary results page, shown in the UK since August 2024. AI Mode is a separate conversational search mode, [live in the UK since 28 July 2025](https://blog.google/around-the-globe/google-europe/united-kingdom/ai-mode-search-uk/), where a buyer keeps asking follow-up questions. For both, a page that answers the first question and the likely follow-ups gives Google more reasons to draw on it than a page that ranks for one phrase. We compare the two in [what Google AI Mode is](/insights/end-of-ten-blue-links), and explain why cited pages still get fewer visits in [why traffic falls when rankings hold](/insights/zero-click-search-your-business). ## Does Microsoft Copilot recommend suppliers differently from ChatGPT? It can, because Copilot works from Bing and sits inside the tools many buyers use all day. Microsoft says that when web search is on, Copilot sends a query to the Bing search service and uses the results to compose its answer ([Microsoft Learn, updated August 2026](https://learn.microsoft.com/en-us/copilot/microsoft-365/manage-public-web-access)). Microsoft 365 Copilot passed 30 million paid business seats in July 2026, according to [Microsoft's earnings release](https://www.microsoft.com/en-us/investor/earnings/fy-2026-q4/press-release-webcast), so a buyer writing a board paper can ask for a shortlist without going near Google. The extra work is looking after Bing: allow Bingbot, check that Bing has indexed your service pages, and make your Bing Places listing match your website. In our experience, a firm that only watches Google can be named by ChatGPT and still missed by Copilot. ## How do Gemini and Claude differ? Gemini draws on Google's index and Google's understanding of entities (the firms, people and places it recognises as distinct things), so it rewards the same housekeeping as Google Search. Claude uses its own search crawler and cites a different mix of sources. For Gemini, a complete Google Business Profile, consistent name and address details, and structured data (a short block of code that labels your firm, services and people for software) help it read your facts correctly; Google says its AI features need no special markup, so treat it as accuracy, not a ranking trick. Gemini keeps reaching buyers in new places: Google began [moving phone users from Google Assistant to Gemini](https://blog.google/products/gemini/google-assistant-gemini-mobile/) in March 2025, and [Gemini in Chrome reached UK desktop users](https://blog.google/products-and-platforms/products/chrome/were-expanding-gemini-in-chrome-to-users-in-the-uk/) in July 2026. The assistants favour different sources. In [Muck Rack's May 2026 analysis](https://www.globenewswire.com/news-release/2026/05/07/3290268/0/en/generative-pulse-earned-media-consistently-drives-ai-citations-holding-at-84.html) of more than 25 million links cited by ChatGPT, Claude and Gemini, ChatGPT's most cited domain was Wikipedia, Gemini's was Reddit and Claude's was PubMed Central, and earned media (coverage written by others) made up 84% of cited links. For every assistant, "award-winning" without naming the award gives it nothing to repeat; specific, checkable statements confirmed by other sources give it something to say. ## Is Perplexity worth optimising for in B2B? Not as a separate project, but it is worth checking every month. Its audience is small, 0.9% of visits to AI chatbot websites in Similarweb's September 2026 figures, but it searches the web for every question and numbers its sources, which makes it the quickest place to see why a competitor was named. A [Peec AI study reported by Search Engine Land](https://searchengineland.com/ai-search-engines-cite-reddit-youtube-and-linkedin-most-study-473138) in March 2026 found G2 prominent in Perplexity's answers to B2B questions, so a gap in your reviews often shows there first. ## Why do ChatGPT and Gemini recommend different companies? Because they read different things: different indexes, different favourite sources, memory in one case and a live search in another, and answers that vary from run to run. The disagreement tells you where the gap is, so ask each question several times and count a result only when it repeats. In our experience these patterns hold more often than not: | What you see | What it usually means | What to fix first | |---|---|---| | Perplexity names you, ChatGPT does not | Your site is clear, but the wider web is thin or out of date | Independent coverage, directories and reviews | | ChatGPT and Gemini name you, Copilot does not | Bing has not indexed you well | Bing Webmaster Tools and Bing Places | | Google's AI answers name you, the others do not | Few independent sources repeat your facts | Coverage and listings away from your own site | | The assistants describe you but never recommend you | No evidence you suit the buyer's question | A page answering that question for each sector | | Nobody names you | Crawlers blocked, or no page answers the question | robots.txt, then pages built around buyer questions | ID-Pal, an identity verification firm we work with, fitted the fourth row: "AI could describe ID-Pal. It never recommended it." After pages were rebuilt around the questions compliance leads ask, the assistants moved one at a time: "Gemini now names ID-Pal for UK wealth managers. Claude names it for UK fund administrators, on a page rebuilt in August, and links NorthRow to ID-Pal in the same answer." The detail is in [the ID-Pal case study](/work/id-pal). ## Will Siri, Alexa or Gemini recommend our firm to a buyer? They may, because voice assistants are being rebuilt on the same kind of models. [Apple and Google said in January 2026](https://blog.google/company-news/inside-google/company-announcements/joint-statement-google-apple/) that Apple's next foundation models will be based on Google's Gemini and will help power a more personalised Siri. Amazon opened [early access to Alexa+ in the UK on 19 March 2026](https://www.aboutamazon.co.uk/news/devices/introducing-alexa-in-the-uk-the-next-generation-of-alexa); it routes each request to a model [including Amazon Nova and Anthropic's Claude](https://www.aboutamazon.com/news/devices/new-alexa-tech-generative-artificial-intelligence). A spoken answer has room for one name, perhaps two. If an assistant would put you third on a written shortlist, you may not be mentioned aloud at all, so voice favours the firm that is plainly the answer to a narrow question. In our experience few buyers of B2B services shortlist by voice yet; local and professional services, where assistants read business listings and reviews, are further ahead. Test it by asking Siri, Alexa+ and Gemini your clients' two main questions aloud. The preparation is the same as for text, plus a specialism that fits in one spoken sentence. ## How do we prepare for all of them at once? Build one foundation. Every assistant is checking what you do, for whom, and whether the rest of the web agrees; how they cross-check is in [why ChatGPT sounds unsure about a firm](/insights/ai-cross-references-business-sources). ### Step 1: Let the crawlers in Allow Googlebot, Bingbot, OAI-SearchBot, Claude-SearchBot and PerplexityBot in robots.txt, and keep key text in the page's HTML: the main AI crawlers [do not run JavaScript](https://vercel.com/blog/the-rise-of-the-ai-crawler). ### Step 2: Write your specialism in one sentence "A full-service agency helping businesses with all their digital needs" cannot be spoken as a recommendation. For example, "Demand generation for industrial manufacturers with sales teams of ten to fifty" can. Put your sentence at the top of your homepage and in every listing. ### Step 3: Claim the listings each assistant reads Complete your Google Business Profile (Google and Gemini), Bing Places (Copilot) and Apple Business Connect (Siri and Apple Maps), even with no walk-in clients, plus LinkedIn and the directories in your field, with identical details. ### Step 4: Earn reviews where your buyers look Reviews matter for B2B firms. In G2's April 2026 survey, 45% of software buyers said citations of review sites were the most confidence-inspiring signal in an AI answer. Ask satisfied clients for reviews on Google and on G2, Clutch or Trustpilot, and keep them recent. ### Step 5: Answer buyer questions on your own pages Build pages around the questions buyers ask, with the answer near the top and evidence beneath: named clients, figures, named authors. See [building content around customer questions](/insights/content-around-customer-questions). ## What to do this month 1. Write a fixed list of 20 buyer questions in a buyer's words, including the follow-ups they ask after the first answer. 2. Check robots.txt and any Cloudflare setting for the search crawlers above, and confirm in Bing Webmaster Tools that your service pages are indexed. 3. Put each question to ChatGPT, Google (AI Overview and AI Mode), Copilot, Gemini, Claude and Perplexity in a clean session, several times, recording who is named and which sources are cited. 4. Use the pattern table to find your most common gap, and fix it first. 5. Claim or complete Google Business Profile, Bing Places and Apple Business Connect so they match your website, with your one-sentence specialism at the top. 6. Ask Siri, Alexa+ and Gemini the same two questions aloud, then repeat the whole check in four weeks. ## Frequently asked questions ### Do we need a separate strategy for each assistant? No. Check each one, because a buyer might use any of them, but the foundation is shared. Once your website and the sources that mention you tell the same story, the assistants tend to converge. The only platform-specific work most firms need is Bing for Copilot and Google Business Profile for Gemini. ### Should we pay for advertising inside the assistants? ChatGPT has shown labelled ads to UK users on its Free and Go plans since 11 August 2026 ([OpenAI](https://openai.com/index/testing-ads-in-chatgpt/)); paid plans such as Plus, Business and Enterprise show none. OpenAI says the ads do not change the answer itself, so being named in the answer is still earned. An ad puts your name beside a conversation, not in the recommendation. We compare the two in [whether being named by ChatGPT is cheaper than Google Ads](/insights/economics-ai-visibility-vs-advertising). ### Does our Google ranking still matter? Yes. Google AI Overviews, AI Mode and Gemini start from Google's index, so ranking helps, but only 38% of AI Overview citations came from top-ten pages in Ahrefs' March 2026 data, because Google runs several related searches for one question. Copilot starts from Bing, and ChatGPT search has drawn on Bing. Keep search healthy and cover the follow-up questions too. ### How quickly do changes show up? Assistants that search the web live (ChatGPT search, Perplexity, Copilot and Google's AI features) can reflect a new page or article within weeks. Answers from a model's own memory change only when it is retrained, which takes months. ID-Pal shows the pace: "UK buyer questions where AI names ID-Pal: none in May, two in August, four in September 2026." ### Are there other assistants we should watch? Yes. Similarweb's September 2026 figures show DeepSeek at 3.4% and Grok at 2.4% of visits to AI chatbot websites, and assistants are being built into browsers and office software. Because they all read the same web, a firm the main assistants name consistently is well placed for the next one. If buyers in your sector let agents build shortlists, read [what happens when an AI agent builds the shortlist](/insights/ai-agents-change-b2b-buying). To see which assistants name you today, and why the others do not, [book a free Growth Review](/growth-review): 30 minutes with David Adams and three fixes in writing. Where the gap is what other sites say about you, [Reputation That Wins](/reputation-that-wins) is built to close it. ## Sources - [G2, New G2 research: half of B2B software buyers now start their research with AI chatbots (April 2026)](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html) - [OpenAI, A milestone in expanding access to AI (August 2026)](https://openai.com/index/expanding-access-to-ai-with-chatgpt-ads/) - [Alphabet, investor presentation (June 2026)](https://blog.google/alphabet/investor-presentation-june-2026/) - [Microsoft, FY26 fourth quarter earnings press release (July 2026)](https://www.microsoft.com/en-us/investor/earnings/fy-2026-q4/press-release-webcast) - [Forrester, B2B buyers make zero-click buying number one (January 2026)](https://www.forrester.com/blogs/b2b_buyers_make_zero_click_buying_number_one/) - [Similarweb, AI chatbot website traffic share (September 2026)](https://x.com/Similarweb/status/2096878021378466096) - [Ahrefs, AI Overview citations from top-ten pages (March 2026)](https://ahrefs.com/blog/ai-overview-citations-top-10) - [Muck Rack, What is AI reading? Earned media drives AI citations (May 2026)](https://www.globenewswire.com/news-release/2026/05/07/3290268/0/en/generative-pulse-earned-media-consistently-drives-ai-citations-holding-at-84.html) --- Title: Which kinds of PR actually win work for a B2B firm like ours? URL: https://fortitudemedia.ai/insights/pr-for-b2b-what-works Author: Ross Williams Published: 2026-03-31 Last reviewed: 2026-09-24 Topic: Press, reviews and being trusted Short answer: B2B PR works when it reaches the few hundred people who might buy from you, in the trade titles, events, LinkedIn feeds, review sites and AI answers they already trust. Expert comment, original data, bylined articles, speaking, client stories and reviews do that. Press releases about yourself, reach figures and paid placements rarely win work. Judge PR by coverage in target titles, enquiries that mention it and what AI assistants say about you. You have probably paid for PR before. The agency sent a monthly report listing pieces of "coverage", a combined readership in the millions and a figure for what the coverage would have cost as advertising. Your sales director could not point to a single deal that came from any of it. When the contract ended, nobody noticed. That experience is common, and it usually has the same cause. Most PR methods were built for consumer brands, where reach is the point. A B2B firm with £1m to £20m+ of revenue does not need millions of readers. It needs the few hundred people who could buy from it to trust it when they come to choose. Those people read a handful of trade titles, go to a few events, ask their peers and, more and more, ask an AI assistant: in [G2's April 2026 survey of 1,076 software buyers](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html), 51% said they now start their research in an AI chatbot, up from 29% a year earlier. This guide sets out which kinds of PR do that job for founder-led B2B firms, which waste money, what it costs to buy, how to choose an agency and how to tell whether it is working. ## Key takeaways - B2B PR exists to make your firm the trusted name when a small group of buyers is choosing, not to reach a large audience. - Eight methods work: expert comment, original data, bylined articles, speaking, client stories, independent recognition, the founder's LinkedIn and reviews. - Press releases about yourself, reach figures, paid placements and sponsorship without a speaking slot rarely win work, however good the report looks. - AI assistants cite what others say about you far more than what you pay for: earned media made up 84% of AI citations in Muck Rack's May 2026 study, paid content 0.3%. - Coverage alone does not get you recommended; your own pages must also answer the questions buyers ask. - Choose an agency on who your buyers are and what it will report, and be wary of anyone guaranteeing coverage or a place in ChatGPT's answers. - Judge PR by coverage in target titles, enquiries that mention it and what AI assistants say about you, not by reach. ## What is PR for in a B2B firm? It is for making your firm the obvious, trusted choice when the small number of people who could buy from you are doing their research. Everything else is secondary. B2B buying is slow, involves several people and carries real risk for the buyer. A finance director choosing an audit firm or a managing partner choosing an IT provider will be judged on the decision. So they look for evidence before they make contact: who is talked about in the trade press, who spoke at the conference, who their peers mention and who ChatGPT or Copilot names when asked. Thought leadership reaches the people behind the scenes too: in the [2025 Edelman and LinkedIn B2B Thought Leadership Impact Report](https://www.edelman.com/expertise/Business-Marketing/2025-b2b-thought-leadership-report), a survey of US executives, 64% of hidden decision-makers said they trust an organisation's thought leadership more than its marketing materials. Good B2B PR puts your firm into those places, described accurately, often enough that you are on the shortlist when it is drawn up. ## Which kinds of PR win work? Eight methods work for founder-led B2B firms. Most should pick two or three and do them well. | Method | What it needs from you | What it gives buyers and AI assistants | |---|---|---| | Expert comment in the trade press | A founder who replies within the hour with a clear view | Independent mentions, in context, on trusted titles | | Original data and research | A question your buyers argue about, and your records or a survey | A figure others quote with your name attached | | Bylined articles | A founder's view, written up well | Your thinking in a place buyers trust, tied to a named person | | Speaking | A founder willing to take a panel seat or short talk | A small, exactly right audience, plus material for later | | Client stories | Clients willing to be named and quoted | A buyer believes another buyer more readily than you | | Independent recognition | Award entries, certifications, professional registers | A quick, checkable reason to trust you | | The founder's LinkedIn | Regular posts and articles that share practical knowledge | A source assistants cite often | | Reviews | A habit of asking happy clients, on the platforms buyers use | The evidence buyers and assistants trust most | Expert comment is usually the cheapest route in and builds fastest, because every news story needs someone to explain it. We cover it in [how to get trade journalists to call us for comment](/insights/building-relationships-journalists-editors), and research in [how to turn our own data into coverage in the trade press](/insights/leverage-data-research-press-coverage). The founder's LinkedIn belongs on this list because AI assistants cite it. A [Semrush study published in March 2026](https://www.semrush.com/blog/linkedin-ai-visibility-study/) found LinkedIn in about 11% of AI answers, with articles making up 50% to 66% of LinkedIn citations and many cited posts having only 15 to 25 reactions. Useful, specific posts count for more than popularity. Reviews belong here because buyers weigh them. In the same [G2 survey](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html), 45% of software buyers said citations of review sites were the most confidence-inspiring signal in an AI answer. For B2B firms that means G2, Clutch, Trustpilot and Google reviews, whichever your buyers use. For example, a hypothetical commercial insurance broker specialising in logistics might choose expert comment and research. Its founder becomes the person the transport trade press calls about insurance costs, and each year it publishes a short, anonymised study of claims trends. Logistics buyers researching brokers keep meeting the same name in the titles they already read. ## What does not work, and why do firms keep paying for it? These methods persist because they are easy to report on, not because they win work. - **Press releases about yourself, sent to everyone.** New hires and anniversaries sent to thousands of journalists rarely get covered, because they are not news outside your firm. - **Reach and advertising value figures.** A total readership or an estimate of what the space would have cost as advertising looks impressive and means little. One piece in the title your buyers read beats a hundred on sites they never visit. We compare the two approaches in [digital PR or traditional PR for a B2B firm](/insights/digital-pr-vs-traditional-pr). - **Paid placements and link packages.** Buyers do not read them. [Google's spam policies](https://developers.google.com/search/docs/essentials/spam-policies) treat paid articles with links that pass ranking credit as link spam, and in [Muck Rack's study](https://www.globenewswire.com/news-release/2026/05/07/3290268/0/en/generative-pulse-earned-media-consistently-drives-ai-citations-holding-at-84.html) paid and advertorial content made up 0.3% of links AI tools cited. - **Sponsorship without a speaking slot.** A logo on a lanyard rarely changes what buyers think. Negotiate a talk or a hosted roundtable, or keep the money. - **Publishing only on your own channels.** Your blog reaches people who already follow you. Independent coverage builds trust with people who do not know you yet. ## Can PR help my firm get recommended by ChatGPT? Yes, and it is one of the strongest inputs. Assistants cite what others say about you far more than what you pay for: in [Muck Rack's May 2026 analysis](https://www.globenewswire.com/news-release/2026/05/07/3290268/0/en/generative-pulse-earned-media-consistently-drives-ai-citations-holding-at-84.html) of more than 25 million links cited by ChatGPT, Claude and Gemini, earned media made up 84% and journalism 27%. Why that is so is covered in [whether ChatGPT trusts press coverage more than our website](/insights/earned-owned-paid-authority-signals). Coverage alone is not enough. Assistants also need a page of yours that answers the buyer's question. When we started work with ID-Pal: "AI could describe ID-Pal. It never recommended it." After the pages were rebuilt: "Where a page has been rebuilt, ID-Pal appears. Where it has not yet, it does not, which tells us exactly what to build next." ([ID-Pal case study](/work/id-pal)). Coverage makes an assistant sure of you; pages give it a reason to recommend you. You can now pay for space next to the answer. ChatGPT has shown labelled ads to UK users on its Free and Go plans since 11 August 2026, and [OpenAI says](https://openai.com/index/testing-ads-in-chatgpt/) the ads do not change the answer itself. Being named in the answer is still earned. ## How much does a B2B PR agency cost, and what do you get at each level? Prices depend on scope, seniority and how much founder time the agency saves, so ask each provider for a written quote against the same brief. What differs most is what you get: | Provider | Who does the work | What it usually delivers | Best for | |---|---|---|---| | Freelancer | One experienced person | Pitching, drafting and requests for comment in a sector they know | A firm with one clear market and a founder who is available | | Boutique B2B agency | A small senior team | A plan, steady pitching, research support and monthly reporting | Most founder-led B2B firms | | Larger agency | An account team, often with junior staff doing the day-to-day work | Wider media reach, events, crisis support | Firms with several markets or national ambitions | Whoever you use, commit to a steady year rather than a short burst. Our own prices are on [what it costs](/what-it-costs). ## How do we choose a B2B PR agency? Start with who your buyers are, and judge the agency on what it will report. Ten questions to ask: 1. Who do you think our buyers are, and what do they read? 2. Which titles, events and people would you target first, and why? 3. Who will do the work each week, and how senior are they? 4. What do you need from our founder, in time and responses? 5. How will you find requests for comment in our sector? 6. What will you report each month, and how will you show enquiries and AI answers? 7. Can we see an anonymised monthly report? 8. What is the minimum term and the notice period? 9. Who owns the media list, drafts and research if we leave? 10. Which clients in similar sectors can we speak to? Red flags: reports built on reach or advertising value, link packages or paid placements sold as coverage, and guarantees of coverage in titles the agency does not control. Be wary of anyone guaranteeing a place in ChatGPT's answers; nobody can guarantee what an AI assistant says. A provider can guarantee effort, activity and reporting. A good first-month report shows the target list and why, every request answered and pitch sent, replies received, coverage won with links, and a baseline of what AI assistants say about you now. ## How do we put a PR plan together? Keep it to one page and run it for a year. ### Step 1: Define who you need to reach Describe the people who buy from you by role, sector, size and region. Ask your best clients what they read, which events they attend and whose opinions they trust. ### Step 2: Choose three to five topics to be known for Pick the subjects where your firm has real expertise and your buyers have real problems: a regulation, a recurring cost, a technical change. ### Step 3: Pick two or three methods Choose the ones that suit your people. A founder who enjoys speaking should speak. A firm sitting on useful data should publish it. Most start with expert comment. ### Step 4: Set a steady rhythm Decide what happens every month: requests answered, journalists contacted, articles due, events pitched. Steady and modest beats ambitious and occasional. ### Step 5: Agree how you will judge it Set the measures before the first report arrives, so nobody is tempted to count reach. ## What results should we expect in the first six months? Quotes, replies from journalists and perhaps a published article in the first quarter, then journalists coming back and the first requests that arrive unprompted. Assistants that search the web live can reflect new coverage within weeks; answers from a model's own memory change only when it is retrained, which takes months. Enquiries usually follow later. The quarter-by-quarter picture is in [how long PR takes to work and what a year should look like](/insights/compounding-effect-consistent-pr-12-months). ## How do we tell whether PR is working? Measure what buyers would notice, not what an agency can count. - **Coverage in the titles that matter.** Count pieces in your target titles only, and note whether you were quoted, featured or merely listed. - **What buyers say.** Ask "How did you hear about us?" on your enquiry form and in first meetings. - **What AI assistants say.** Put a fixed list of 20 to 50 buyer questions to the main assistants several times a month, in a clean session, and count a win only when it repeats. See [how to find out whether ChatGPT is recommending our firm](/insights/track-ai-recommending-your-business). - **What your sales team notices.** Prospects who already know your founder's views and ask fewer questions about credibility. ## What to do this month 1. Ask five of your best clients what they read, which events they attend and whose opinion they trust in your field. 2. Write down the three to five topics your firm should be known for, and your founder's view on each. 3. Stop any press release distribution, paid placement or sponsorship with no clear link to your buyers. 4. Ask three happy clients for a review on the platform your buyers use. 5. Choose two methods from the table above and set a monthly rhythm you can keep for a year. ## Frequently asked questions ### Is B2B PR worth it for a small specialist firm? Often more so. In a specialist market there may be only a few trade titles and a few hundred buyers, so a modest amount of consistent coverage can make you the recognised name. Journalists covering niche sectors also have fewer sources, so a founder who is helpful and available gets quoted more easily. ### Should we use an agency or do it ourselves? Usually a mix. The relationships and expertise have to come from your founder and senior people, because journalists want the expert. An outside team adds the list, the preparation, the pitching, help with writing and the discipline of doing it every month when everyone is busy. ### Do press releases still have any use? Yes, as a way of giving journalists the facts. For real news, such as an acquisition, a new study or a client win the client is happy to announce, a clear release sent personally to the relevant journalists is useful. Routine internal news sent to large distribution lists is not. ### How is PR different from content marketing? Content marketing is what you publish on your own channels. PR is what others publish about you, or publish from you after choosing it: quotes, features, bylined articles and talks. Both matter, but only PR provides the independent endorsement buyers and AI assistants treat as evidence. ### Can PR help if we have had negative coverage in the past? Yes, over time. Deal with the issue, then build a steady record of accurate coverage so the old piece is no longer the main thing people find. AI assistants summarise what the weight of trusted sources says, so a year of good coverage changes the balance. If you want PR measured by who knows and trusts you, and by what AI assistants say about you, that is what [Reputation That Wins](/reputation-that-wins) is built to do. If you would rather start with an outside view, the free [Growth Review](/growth-review) gives you three fixes in writing. ## Sources - [G2, The Answer Economy: half of B2B software buyers now start research with AI chatbots (April 2026)](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html) - [Muck Rack, What is AI reading? Earned media drives 84% of AI citations (May 2026)](https://www.globenewswire.com/news-release/2026/05/07/3290268/0/en/generative-pulse-earned-media-consistently-drives-ai-citations-holding-at-84.html) - [Semrush, LinkedIn AI visibility study (March 2026)](https://www.semrush.com/blog/linkedin-ai-visibility-study/) - [Edelman and LinkedIn, 2025 B2B Thought Leadership Impact Report (June 2025)](https://www.edelman.com/expertise/Business-Marketing/2025-b2b-thought-leadership-report) - [Google Search Central, Spam policies for Google web search (August 2026)](https://developers.google.com/search/docs/essentials/spam-policies) - [OpenAI, Testing ads in ChatGPT (August 2026)](https://openai.com/index/testing-ads-in-chatgpt/) --- Title: Which sources does ChatGPT cite when a B2B buyer asks who to hire? URL: https://fortitudemedia.ai/insights/ai-optimisation-b2b-vs-b2c Author: David Adams Published: 2026-03-31 Last reviewed: 2026-09-24 Topic: Getting named by AI assistants Short answer: Mostly sources other than your own website. Large citation studies put Wikipedia, Reddit, LinkedIn, YouTube and Forbes at the top, with G2 and Gartner prominent on business questions, and Muck Rack found in May 2026 that 84% of links cited by AI tools were earned media. Your own site still supplies the facts an assistant repeats, including who you suit. Run your buyers' questions, log the sources cited, and earn a place in the three that recur most. A finance director asks ChatGPT which firms to talk to about an audit. A founder asks Copilot for three outbound agencies that work with software companies. Each time, the assistant replies with names, and under the names sits a short list of sources. Those sources decide who is named far more than anything a firm says about itself. This matters because the first shortlist is now often drawn up this way. In G2's April 2026 survey of 1,076 software buyers, [51% said they now start their research in an AI chatbot](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html), up from 29% a year earlier, and 45% said citations of review sites were the most confidence-inspiring signal in an AI answer. That survey covers software buyers, but the habit is wider: [Forrester found in January 2026](https://www.forrester.com/blogs/b2b_buyers_make_zero_click_buying_number_one/) that 94% of business buyers use AI somewhere in their buying process. This guide sets out which sources the assistants cite when a business buyer asks for a supplier, what the published studies show, how Copilot differs from ChatGPT, and what a founder-led firm can do to appear in the places that count. For the wider question of getting recommended, read our guide on [how to get ChatGPT to recommend your firm](/insights/what-is-llm-optimisation). ## Key takeaways - When a buyer asks an assistant who to hire, the answer rests mostly on what other sites say about the firms it names, not on the firms' own claims. - Large citation studies put Wikipedia, Reddit, LinkedIn, YouTube and Forbes at the top overall, with G2 and Gartner prominent on business questions. - Review sites count for B2B firms: G2 and Capterra for software, Clutch for agencies and consultancies, and Google reviews and Trustpilot as general proof. - Earned coverage dominates: Muck Rack found that 84% of links cited by AI tools in May 2026 were earned media, and 0.3% were paid. - ChatGPT and Google come first for most firms, but Copilot matters in B2B because it is the assistant many buyers have at work. - Your own website still supplies the facts an assistant repeats, so state plainly who you suit and who you do not. - The only reliable way to know the source mix in your market is to run your buyers' questions yourself and log what gets cited. ## Which websites does ChatGPT cite most for B2B supplier recommendations? The most cited sources are reference, community and business media sites: Wikipedia, Reddit, LinkedIn, YouTube and Forbes, with review and analyst sites such as G2 and Gartner prominent on business questions. Your own website is one source among many, and usually not the one that decides whether you are named. [Profound's analysis of 680 million citations](https://www.tryprofound.com/blog/ai-platform-citation-patterns) (August 2024 to June 2025) found Wikipedia was ChatGPT's most cited source at 7.8% of its citations, followed by Reddit (1.8%), Forbes (1.1%) and G2 (1.1%). [A Peec AI study of 30 million sources](https://searchengineland.com/ai-search-engines-cite-reddit-youtube-and-linkedin-most-study-473138), reported by Search Engine Land in March 2026, ranked Reddit, YouTube, LinkedIn, Wikipedia and Forbes as the top five and found Perplexity emphasised Reddit, LinkedIn and G2 for B2B questions. [Muck Rack's May 2026 study](https://www.globenewswire.com/news-release/2026/05/07/3290268/0/en/generative-pulse-earned-media-consistently-drives-ai-citations-holding-at-84.html) of more than 25 million cited links found 84% were earned media, against 0.3% paid. Treat the shares as direction, not a league table. The studies are mostly American, several are run by firms that sell AI tracking tools, and the mix moves from month to month. | Source type | Examples | What the studies show | What it means for a founder-led firm | |---|---|---|---| | Reference sites | Wikipedia | ChatGPT's most cited source, 7.8% of its citations (Profound) | Few firms qualify for an entry; make sure what is written about your sector is accurate | | Business and trade press | Forbes, trade titles, business pages of national papers | Forbes in the top five (Peec AI); earned media 84% of cited links (Muck Rack) | Comment, contributed articles and research in the titles your buyers read | | Review and analyst sites | G2, Gartner, Capterra, Clutch | G2 1.1% of ChatGPT citations, Gartner 1.0% of Perplexity's (Profound) | A complete profile and recent reviews on the platform your buyers use | | Professional network | LinkedIn | Top five overall (Peec AI); 1.3% of AI Overview citations (Profound) | Founder articles and a company page that describe you in the same words as your site | | Communities and video | Reddit, YouTube, Quora | Reddit first overall (Peec AI) | Matters most in software and technical markets; less for most advisory firms, in our experience | | Your own site | Service pages, case studies, fit statements | Supplies the facts the assistant repeats | Say who you are for, what you do and what happened for named clients | ## What percentage of B2B buyers use ChatGPT to research suppliers? Most use an AI assistant at some stage, and around half now start there. The figures vary with who was asked. | Survey | Who was asked | What it found | |---|---|---| | G2, April 2026 | 1,076 software buyers | 51% start research in an AI chatbot; 71% use one at some point | | Forrester, January 2026 | Business buyers | 94% use AI in their buying process; 61% use private AI tools supplied by their employer | | TrustRadius, July 2026 | 1,862 technology buyers | 63% used AI in their purchase; 94% of those fact-checked it at least sometimes; 74% use reviews | The TrustRadius finding explains why sources matter twice: the answer starts the shortlist, then the buyer checks the sources behind it. None of these surveys is UK-only, and G2 and TrustRadius surveyed software and technology buyers, not buyers of professional services. Our guide to [how AI is changing the way B2B buyers choose suppliers](/insights/ai-search-change-your-industry) covers the wider shift. ## Why do supplier answers lean on other people's sources? Because the buyer is spending the company's money and has to defend the choice, and the assistant can only back a name with something the buyer can check. What you say about yourself is a claim; what a client, a reviewer or a journalist says is evidence. In our experience, supplier questions usually get several named firms with a condition attached to each: for mid-sized firms, this one; for a regulated sector, that one. Each condition comes from somewhere, usually a review profile, a listing, an article or a case study. If nothing outside your website says what you are good at, the assistant has no condition to attach to your name. ID-Pal, an identity verification firm we work with, shows the gap between being known and being recommended: "Across 36 recorded AI answers to the questions ID-Pal's buyers actually ask, ID-Pal was named three times, and only for its brand or its Salesforce integration. AI knew who it was, and had no reason to recommend it." ## Do G2 or Clutch reviews help my company get recommended by ChatGPT? Yes, if they are on the platform your buyers use and they are recent. G2 was one of ChatGPT's four most cited sources in Profound's data, Peec AI found Perplexity leaned on G2 for B2B questions, and in G2's own survey 45% of software buyers said review-site citations were the most confidence-inspiring signal in an AI answer. That last figure comes from a review site's own research, so weigh it accordingly. Star ratings do matter for a B2B firm, as long as they sit on a platform the assistants read and your buyers recognise. Capterra, Clutch and Trustpilot were not in Profound's published top-ten lists, so the table below reflects who uses each platform rather than a citation share. | Platform | Suits | When it matters | What to do | |---|---|---|---| | G2 | Software and SaaS firms | When buyers compare tools in a category | Claim the profile, match its description to your site, ask for reviews after a milestone | | Capterra | Software sold to smaller businesses | When a small firm is choosing a tool | Keep the listing consistent with G2 | | Clutch | Agencies and consultancies | When a buyer asks for an agency or a consultancy | Collect a few detailed reviews that name the service and the outcome | | Trustpilot | Firms with many clients | When the assistant checks general trust | Reply to reviews; never buy or filter them | | Google reviews | Any firm with an office | When the question includes a place, and in Google's AI features | Keep your Business Profile accurate | For most founder-led firms, one platform done properly beats five done thinly. ## Does Microsoft Copilot recommend suppliers differently from ChatGPT? Copilot draws on Microsoft's Bing index, so if Bing has not indexed your pages, Copilot cannot cite them. ChatGPT search has also drawn on Bing, so the two often overlap, while Google's AI features draw on Google's index. Copilot matters more in B2B than its public profile suggests. Forrester found business buyers four times as likely as consumers to use Copilot, and 61% of buyers use private AI tools supplied by their employer. [Microsoft 365 Copilot passed 30 million paid seats](https://www.microsoft.com/en-us/investor/earnings/fy-2026-q4/press-release-webcast) in July 2026. In many companies it is the assistant IT has approved. We have not found a large published study of which sites Copilot cites, so run your own questions in it. | Assistant | Where its web answers come from | What studies show it cites | Why it matters for B2B | |---|---|---|---| | ChatGPT | ChatGPT search, which has drawn on Bing | Wikipedia, Reddit, Forbes, G2 (Profound) | Largest audience: more than a billion weekly users (OpenAI, August 2026) | | Microsoft Copilot | Bing | No large published study found | The assistant many buyers have at work (Forrester, Microsoft) | | Google AI Overviews and AI Mode | Google's index | Reddit, YouTube, Quora, LinkedIn, Gartner (Profound) | Reaches everyone who still starts with a search | | Perplexity | Its own web search | Reddit, YouTube, Gartner (Profound); G2 on B2B questions (Peec AI) | Smaller audience, used for research | Whichever assistant it is, it can only cite pages its crawler can read. Our guide to [how AI crawlers read your website](/insights/ai-crawlers-vs-google-spiders) explains what to check. ## Which assistants should a B2B firm check first? Check ChatGPT and Google (AI Overviews and AI Mode) first, then Microsoft Copilot and Gemini, then Perplexity and Claude. ChatGPT has [more than a billion weekly users](https://openai.com/index/expanding-access-to-ai-with-chatgpt-ads/) (OpenAI, August 2026), Google's AI features sit in front of every search, and Copilot is what many buyers have at work. Some advice puts Perplexity and Claude first for business research, but we have seen no evidence that business buyers favour them. Both are worth checking, after the assistants most buyers use. For how the five differ in detail, see [which AI assistants buyers actually use](/insights/ai-search-landscape-chatgpt-perplexity-google-claude). ## How do you find the sources that matter in your market? Run the questions your buyers ask, in each assistant, and log every source cited. A fixed list of 20 to 50 questions, each asked several times a month in a clean session, shows the pattern; answers vary from run to run, so count a source only when it keeps appearing. ### Step 1: Write down the questions buyers actually ask Take them from sales calls and enquiry forms, not keyword tools. "Which audit firms work with owner-managed businesses in Leeds?" is a real question; "audit services" is not. ### Step 2: Run each question in a clean session Use a temporary or logged-out chat in ChatGPT, Copilot, Google AI Mode, Gemini and Perplexity, so your own history does not shape the answer. ### Step 3: Log every firm named and every source cited Keep one row per question, per assistant, per run: the date, the firms named, each link shown and its type (review site, trade press, directory, forum or a firm's own site). ### Step 4: Work on the three sources that recur most After a month, the sources that keep naming your competitors and not you are your target list. Earn a place in the top three, then run the same questions again next month. ## What should your own website say so assistants repeat it? Say who you work with, what you do and what happened for named clients, in words an assistant can lift. Third-party sources decide whether an assistant trusts you; your own pages decide whether it describes you correctly. The pages that get used are comparisons with the alternatives, case studies with outcomes, what working with you involves, a page for each sector you serve, and an honest account of where you do not fit. Our guide to [what an article that ChatGPT quotes looks like](/insights/anatomy-of-article-cited-by-ai) shows the structure. ID-Pal's rebuild shows why sector pages matter: "Five industry pages rebuilt: fund administration, wealth management, legal, accountancy and estate agency. Each opens with the question a compliance lead types, names the regulator, and answers inside the first ninety words." The result: "UK buyer questions where AI names ID-Pal: none in May, two in August, four in September 2026." The [ID-Pal case study](/work/id-pal) has the detail. ### Worked example: a fit statement an assistant can pass on The most useful sentences on a B2B website say who you suit and who you do not. They give the assistant a condition it can repeat: "if you are this kind of firm, consider them". Example: here is ours, written the way we would want an assistant to repeat it. "Fortitude Media works with founder-led B2B firms with £1m to £20m+ revenue, where the founder still decides, in professional services, technology and SaaS, fintech, recruitment, and agencies and consultancies. The buyer is a CEO or an owner, and the purchase is considered." And the other half: "If you sell to consumers, or buying decisions have passed from the founder to a marketing department, another kind of firm will suit you better." Compare that with "we help businesses of all sizes grow", which an assistant cannot match to any buyer's question. The specific version can be matched, filters out poor fits before they call, and costs nothing to publish. ## What to do this month 1. Write a two-sentence statement of who you work with and who you do not, and put it on your homepage, your main service page and your LinkedIn company page. 2. Claim and complete your profile on the one review platform your buyers use (G2 or Capterra for software, Clutch for agencies and consultancies, Google Business Profile for everyone), then ask three recent clients for a review. 3. List 20 buyer questions and run them in ChatGPT, Copilot and Google AI Mode in a clean session, logging every source cited. 4. From that log, pick the two trade titles or directories that appear most and plan one contribution or listing in each. 5. Check in Bing Webmaster Tools that Bing has indexed your key pages, because Copilot depends on it. 6. Rewrite your best client result as a case study: sector, size, timeline, what was done and what changed, with the client's permission. ## Frequently asked questions ### Does ChatGPT cite our website when it recommends us? Sometimes, usually for facts about what you do and who you serve. The decision to name you rests more on reviews, coverage, directories and LinkedIn. A clear website makes sure the assistant describes you correctly once it mentions you. ### Do star ratings matter for a B2B firm? Yes, on the platforms assistants read and buyers recognise, such as G2 for software and Clutch for agencies. A few detailed, recent reviews that name the service and the result are worth more than many one-line ones. ### Should a smaller B2B firm chase analyst recognition such as Gartner? For most founder-led firms, no. Gartner appears in Perplexity and AI Overview citations, but in TrustRadius's 2026 survey only 13% of technology buyers consulted analyst reports. Case studies, reviews and recognition from your professional body do more for the money. ### Can we pay to appear in the sources ChatGPT cites? Paying for placement rarely helps. Muck Rack found paid and advertorial content made up 0.3% of links cited by AI tools in May 2026, against 84% for earned media. ### Does being honest about trade-offs put business buyers off? No. Business buyers, and the assistants they use, treat stated limits as a sign of credibility. A page that says where you do not fit is more likely to be cited than one that claims to suit everyone. ### How long before new reviews or coverage change what ChatGPT says? Assistants that search the web live (ChatGPT search, Perplexity, Copilot, Google's AI features) can reflect a new article or review within weeks. Answers drawn from a model's own memory change only when the model is retrained, which takes months. If you want to know which sources your buyers' assistants cite today and where your firm is missing, [book a free Growth Review](/growth-review): 30 minutes with David Adams and three fixes in writing. Earning a place in the reviews, coverage and listings that assistants rely on is the work of [Reputation That Wins](/reputation-that-wins). ## Sources - [Profound, AI platform citation patterns: how ChatGPT, Google AI Overviews and Perplexity source information (June 2025)](https://www.tryprofound.com/blog/ai-platform-citation-patterns) - [Search Engine Land, AI search engines cite Reddit, YouTube and LinkedIn most: study (March 2026)](https://searchengineland.com/ai-search-engines-cite-reddit-youtube-and-linkedin-most-study-473138) - [Muck Rack via GlobeNewswire, Generative Pulse: earned media consistently drives AI citations (May 2026)](https://www.globenewswire.com/news-release/2026/05/07/3290268/0/en/generative-pulse-earned-media-consistently-drives-ai-citations-holding-at-84.html) - [G2 via PR Newswire, Half of B2B software buyers now start their research with AI chatbots (April 2026)](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html) - [Forrester, B2B buyers make zero-click buying number one (January 2026)](https://www.forrester.com/blogs/b2b_buyers_make_zero_click_buying_number_one/) - [TrustRadius via PR Newswire, 2026 B2B Buying Disconnect Report (July 2026)](https://www.prnewswire.com/news-releases/trustradius-2026-b2b-buying-disconnect-report-reveals-ai-has-changed-how-buyers-research-but-not-what-they-trust-302825792.html) - [Microsoft, FY26 fourth quarter earnings press release (July 2026)](https://www.microsoft.com/en-us/investor/earnings/fy-2026-q4/press-release-webcast) - [OpenAI, A milestone in expanding access to AI (August 2026)](https://openai.com/index/expanding-access-to-ai-with-chatgpt-ads/) --- Title: Why does ChatGPT describe my firm wrongly, and how do I fix it? URL: https://fortitudemedia.ai/insights/ai-conflicting-information-online Author: David Adams Published: 2026-03-31 Last reviewed: 2026-09-24 Topic: Getting named by AI assistants Short answer: Usually because the facts about your firm disagree across the web: your website, LinkedIn, Companies House, directories, listings and old press each say something slightly different about what you do, for whom and since when. ChatGPT, Gemini and Claude compare those sources and, when they conflict, repeat the wrong or older version. Fix it with one master fact sheet, update every source in a set order over four weeks, then report remaining errors through each assistant's feedback button. Live search answers can change within weeks; answers from a model's memory take months. Ask ChatGPT what your firm does and you may get an answer that is two years out of date and slightly wrong. It names a service you stopped offering, gets the founding year wrong, and calls you "a marketing agency" when you have spent three years repositioning as a specialist consultancy. Ask Gemini and you get a different wrong answer. Neither is lying. Both are doing their best with what the web says about you, and the web disagrees with itself. That first description matters more each year. In [G2's April 2026 survey of 1,076 software buyers](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html), 51% said they now start their research in an AI chatbot, up from 29% a year earlier. G2 surveyed software buyers, so the figure will differ for services, but for many buyers the assistant's summary is now the first account of your firm they read. In our experience it is one of the most common problems founder-led firms have with AI search, and one of the most fixable. AI assistants try to understand your firm as a single thing, an entity in their terms, and describe it in a sentence. When their sources contradict each other, they pick the wrong version, repeat an old one or leave you out. ## Key takeaways - AI assistants build one picture of your firm from every source they find, so a wrong fact anywhere can surface in the answer a buyer reads. - Errors about what you do and for whom do the most damage; differences of tone or formatting matter least. - A master fact sheet, with every public source as a column, shows every conflict on one screen. - Updating sources in a fixed order over four weeks, website and LinkedIn first, stops the old version creeping back. - Feedback buttons, and OpenAI's privacy route for false statements about named founders, are worth using, but neither is a quick or guaranteed fix. - Live search answers can reflect corrections within weeks; answers from a model's memory change only when it is retrained, which takes months. ## Why do AI assistants get facts about my firm wrong? Because they compare many sources, and those sources disagree. Your website, LinkedIn, Companies House, directories, reviews and old press each hold a slightly different version of what you do, where and since when, and the assistant has to choose. That causes four problems. **Misattribution:** an assistant may attach a claim to the wrong version of your firm, or describe a dropped service as current. **Lost confidence:** it qualifies what it says, a problem covered in [why ChatGPT sounds unsure about your firm](/insights/ai-cross-references-business-sources). **Exclusion:** asked for the best firms in a category, it may leave out one whose category is unclear, in our reading of the answers we record. **Exposure:** a negative review has no clear story to set against it. ## Which sources do AI assistants compare? Almost anything public that names your firm: your website, Companies House, LinkedIn, your Google, Bing and Apple listings, Wikidata, Crunchbase, business data sites, directories, press coverage, client case studies and speaker biographies. The table in Step 1 lists who can edit each. Bing matters more than most firms assume, because Copilot draws on it and ChatGPT search has too. How assistants weigh your own sources against independent ones is covered in [how AI cross-checks what the web says about you](/insights/ai-cross-references-business-sources); this guide is about correcting the wrong facts. ## Why do ChatGPT and Gemini give different answers about my firm? Because they read different sources and remember different snapshots of the web. Gemini and AI Overviews start from Google's index and Copilot from Bing's; in a small [Seer Interactive test published in February 2025](https://www.seerinteractive.com/insights/87-percent-of-searchgpt-citations-match-bings-top-results), 87% of ChatGPT search citations matched Bing's top results. A firm with a stale Bing listing can be described correctly by Gemini and wrongly by Copilot on the same day. Each assistant also has favourite sources. [Muck Rack's May 2026 analysis](https://www.globenewswire.com/news-release/2026/05/07/3290268/0/en/generative-pulse-earned-media-consistently-drives-ai-citations-holding-at-84.html) of more than 25 million cited links found ChatGPT's most cited domain was Wikipedia and Gemini's was Reddit. One model's memory may also predate your rebrand while another's does not, and answers vary from run to run, so ask each question several times in a fresh chat and note what repeats. Our guide to [which AI assistants buyers use](/insights/ai-search-landscape-chatgpt-perplexity-google-claude) goes further. ## How do assistants decide which version to believe? No assistant publishes its rules, so what follows is what we observe in the answers we record. Assistants compare meaning, not words: "employment law for manufacturers" and "HR consultancy" read as two different firms. They cross-check claims: four offices on your site and two on LinkedIn may become a range, or nothing at all. They also seem to favour your website over LinkedIn, LinkedIn over directories, and recent statements over old ones. The pattern breaks easily: a detailed recent trade article can beat a thin, stale website. It is plausible that a neutral write-up saying you are slow to respond could outweigh a homepage promising a one-hour response; we have not tested that directly, so treat it as a risk to manage. ## Which errors matter most? In our experience the priorities fall into three tiers, and positioning comes first because recommendations depend on category. | Tier | What the sources disagree on | What the assistant tends to do | When to fix | |---|---|---|---| | Serious | What you do, for whom, your category, founding date and history | Puts you in the wrong category, or none | Week one | | Moderate | Numbers of offices, staff or clients, accreditations, price signals | Repeats the wrong figure or qualifies the claim | Weeks two and three | | Minor | Tone, how your name is written ("Acme Ltd", "ACME"), which founder is quoted | Rarely changes the answer | Week four, then ongoing | Fix minor conflicts too, because they suggest a firm that is not paying attention. ## Why does ChatGPT say we offer a service we stopped years ago? Usually because something public still says you do, or the model learned it before you stopped: an old service page still live or indexed, a directory profile written years ago, an old press release, or a training snapshot that predates the change. Fix what is live first. Redirect or rewrite the old page, update every directory profile, and add one plain sentence to your about page, for example: "We stopped offering payroll in 2023 to focus on advisory work." Assistants that search the web then find a current source that explains the change; answers from memory catch up when the model is retrained. ## How to fix wrong facts: the master fact sheet and four-week plan One organised person can do most of this in four weeks. ### Step 1: Build a master fact sheet Make a spreadsheet. Down the left: what you do in one sentence, who you serve, sectors, locations, founding date, leadership, size, accreditations and former names. Across the top, one column per public source. Fill in what each says today and the conflicts appear at once. Mark the facts that must mean the same thing everywhere: what you do, for whom, where, since when and who leads the firm. These are the columns most firms need: | Source (one column each) | Who can edit it | How to correct it | |---|---|---| | Your website | You | Make it the fullest statement of who you are | | LinkedIn company page | Your page admins | Edit it; founders update their own profiles | | Companies House | The company, through its filings | File changes to office or directors; update nature of business at the next confirmation statement | | Google Business Profile | The verified owner; anyone can suggest edits | Claim, verify and edit | | Google knowledge panel | Google, automatically; verified representatives can suggest edits | Claim it, then "Suggest edits" (see below) | | Bing Places for Business | The business, once it claims the listing | Claim it and match it to the fact sheet | | Apple Business Connect (now part of Apple Business) | The business, once it claims its place on Apple Maps | Claim the location and update it | | Wikidata | Anyone, with or without an account | Edit with references to public sources; other editors will check | | Crunchbase | Mostly its community of contributors | Submit corrections as a contributor | | Business data sites (Endole, Dun & Bradstreet) | The site, often from Companies House filings | Fix Companies House first, then ask for a refresh | | Directories and review sites | The owner; most let a business claim its listing | Claim it, or ask for a correction | | Press coverage | The publication | Ask the editor | ### Step 2: Trace each conflict to its cause For example, the website says founded 2015 because that was the rebrand, LinkedIn says 2014 because that was incorporation, and a directory copied the website years ago. Knowing the cause tells you what to change. ### Step 3: Decide the truth and write it on one page Choose one version of each fact and commit. Write it on one page, with your one-sentence description, three differentiators and former names, and check every profile and biography against it from now on. ### Step 4: Update every source in a fixed order Week one: website and LinkedIn company page. Week two: Google Business Profile, Bing Places, Apple Business Connect, Companies House and the knowledge panel. Week three: directories, review sites, business data sites, Crunchbase, Wikidata and staff LinkedIn profiles. Week four: email signatures, proposals and pitch decks, so nobody reintroduces the old version. ### Step 5: Explain changes rather than hide them If you have repositioned, say so on your site: "We began in 2015 as a general marketing agency and since 2023 have focused on demand generation for industrial firms." That explains the conflict an assistant will still find in old coverage. ### Step 6: Publish the update and label the facts A short LinkedIn post and a dated page on your site give assistants that search the web a current source. Structured data, code that states your name, former names, address and official profiles for software to read, helps search engines read those facts correctly. Google says its AI features need no special markup and OpenAI and Anthropic have not said they read it, so keep it accurate and do not treat it as a ranking trick; see [schema markup for AI assistants](/insights/structured-data-schema-markup-ai). ## How do I update my company's Google knowledge panel? You claim it, then suggest edits. Google says knowledge panel information "generates automatically based on public information on the web", so you cannot write it, but you can ask for changes. Search Google for your firm. If the panel shows "Claim this knowledge panel", click it and sign in to one of the official sites or profiles Google lists, such as YouTube or Search Console, as set out in [Google's verification guide](https://support.google.com/knowledgepanel/answer/7534902?hl=en). Once verified, search for your firm while signed in, click "Suggest edits", choose the wrong item, describe the correction and add supporting links. [Google says](https://support.google.com/knowledgepanel/answer/7534842?hl=en) it reviews feedback from verified users within a few days, but may not make a change it cannot confirm. Not every panel can be claimed yet. ## How do I report a wrong answer directly? Use each assistant's feedback controls and, for false statements about a named person, the data protection route. Do both alongside fixing the sources. | Where the error appears | How to report it | What happens | |---|---|---| | ChatGPT | Thumbs-down button under the answer | [OpenAI asks users](https://help.openai.com/en/articles/6783457-what-is-chatgpt) to flag incorrect answers this way; it is feedback, not an edit | | Gemini | "Bad response", or "Report legal issue" from the "More" menu | Google says feedback helps make Gemini more accurate | | Google AI Overviews | Thumbs down, then "Report a problem" | Feedback goes to Google Search | | A false statement about a named founder in ChatGPT | privacy.openai.com or dsar@openai.com | OpenAI considers it under data protection law | Data protection law (UK GDPR) protects information about people. If ChatGPT says something false about you or a co-founder by name, [OpenAI's Europe privacy policy](https://openai.com/policies/eu-privacy-policy/), which covers UK users and was updated in August 2026, says you can request a correction or removal, and that OpenAI "will consider your request based on applicable law and the technical capabilities of our models". The [ICO says](https://ico.org.uk/for-the-public/your-right-to-get-your-data-corrected/) an organisation normally has one month to respond to a correction request, and that if you are still unhappy after complaining to it, you can complain to the ICO. Wrong facts about the company itself are not personal data, so fix the sources. ## Described correctly, still not recommended Fixing wrong facts makes you describable. It does not, on its own, get you recommended, and [ID-Pal](/work/id-pal) shows the gap. ID-Pal provides identity verification for fund administrators, wealth managers, law firms and accountants, and had bought NorthRow, a UK business verification platform, and folded it into id-pal.com. The assistants got the basics right: "6 of 6 correct answers to 'what does ID-Pal do?', every engine, every run." But "AI could describe ID-Pal. It never recommended it." There was a conflict too, of exactly the kind this guide describes: "The redirect from northrow.com was technically correct, but nothing on either site told Google or the AI engines that NorthRow and ID-Pal were the same company." The fix had two parts. For the conflict: "About four hours of structured data work declaring NorthRow part of ID-Pal on both domains. Live in week one, with no copy changes and no risk to the redirect." For recommendation: "Five industry pages rebuilt: fund administration, wealth management, legal, accountancy and estate agency. Each opens with the question a compliance lead types, names the regulator, and answers inside the first ninety words." | Measure | Result | |---|---| | UK buyer questions where AI names ID-Pal | "none in May, two in August, four in September 2026" | | AI answers still recommending NorthRow as a rival | "six in May, one in September 2026" | If you are described correctly and still not named, read [why ranking first on Google does not guarantee a mention](/insights/google-rankings-wont-protect-you). ## How do we keep it consistent? Give one person ownership and check against the fact sheet every quarter. Conflicts come back when marketing updates the website, sales updates LinkedIn and nobody updates the directories. Define the triggers that force an update everywhere at once: new leadership, positioning or offices, a new accreditation, a dropped service. The trust signals assistants look for are set out in [what E-E-A-T is and why AI cares about it](/insights/eeat-why-ai-cares-more). ## What to do this month 1. Build the master fact sheet: ten facts down the side, every source in the table above across the top. 2. Ask ChatGPT, Gemini, Claude, Copilot and Perplexity "what does [your firm] do?" three times each in a fresh chat, and log every error. 3. Decide the truth for each conflict and write the one-page source of truth. 4. Update your website and LinkedIn this week, claim Google Business Profile, Bing Places and Apple Business Connect next week, then work through the rest in order. 5. Report remaining errors with each assistant's feedback button, and claim your knowledge panel if Google offers it. 6. Put a quarterly check in the diary with a named owner. ## Frequently asked questions ### How do I correct wrong information ChatGPT says about my company? Fix the sources it reads, then report the answer with the thumbs-down button. Correct your website, LinkedIn and listings against a master fact sheet and ask directories and publications to correct theirs. There is no form that edits what ChatGPT says about a company. ### Can I complain to OpenAI if ChatGPT says something false about my business? You can report it with the thumbs-down button. If the false statement is about a named person, such as a founder, UK data protection law gives that person the right to ask for a correction through privacy.openai.com or dsar@openai.com, and the ICO handles complaints about the response. Facts about the company itself are not personal data, so fix the sources. If a statement is seriously damaging, take legal advice. ### How long until the assistants describe us correctly? Assistants that search the web live (ChatGPT search, Perplexity, Copilot, Google's AI features) can reflect corrections within weeks. Answers from a model's own memory change only when it is retrained, which takes months. ID-Pal is a dated example. AI answers still recommending NorthRow as a rival: "six in May, one in September 2026." In our experience descriptions usually improve within a quarter of fixing the serious conflicts, but nobody can guarantee what an assistant will say. ### Will conflicting information hurt our Google rankings? Probably not much in the classic results, which rank pages. Google's AI Overviews and AI Mode describe your firm as a whole, so inconsistency can cost you there. Ranking helps you get cited but does not guarantee it, as we explain in [whether you still need SEO now buyers ask ChatGPT](/insights/seo-vs-llm-optimisation). ### Should we delete old information or update it? Update where you can. Deleting leaves a gap that older copies elsewhere fill with the wrong version. Old, accurate history is fine; old, contradictory positioning is not. ### A trade publication described us wrongly. Can we fix that? Ask for a correction; most publications will fix factual errors. One wrong article against consistent official sources is a minor problem; one that matches an old version of your own site is a serious one, so make your own site the stronger source. Our [technical checklist for an AI-ready website](/insights/technical-checklist-ai-ready-website) covers what else it needs. Assistants can only describe the firm the web shows them, so give them one firm to describe. To see how you are described today, [book a free Growth Review](/growth-review): 30 minutes with David Adams and three fixes in writing. Making press, listings and reviews tell one story is the work of [Reputation That Wins](/reputation-that-wins). ## Sources - [G2, New G2 research: half of B2B software buyers now start their research with AI chatbots (April 2026)](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html) - [Seer Interactive, 87% of SearchGPT citations match Bing's top results (February 2025)](https://www.seerinteractive.com/insights/87-percent-of-searchgpt-citations-match-bings-top-results) - [Muck Rack, Generative Pulse: earned media consistently drives AI citations (May 2026)](https://www.globenewswire.com/news-release/2026/05/07/3290268/0/en/generative-pulse-earned-media-consistently-drives-ai-citations-holding-at-84.html) - [Google Knowledge Panel Help, Get verified on Google (accessed September 2026)](https://support.google.com/knowledgepanel/answer/7534902?hl=en) - [Google Knowledge Panel Help, Submit feedback on content about you (accessed September 2026)](https://support.google.com/knowledgepanel/answer/7534842?hl=en) - [OpenAI Help Center, What is ChatGPT? (accessed September 2026)](https://help.openai.com/en/articles/6783457-what-is-chatgpt) - [OpenAI, Europe privacy policy (August 2026)](https://openai.com/policies/eu-privacy-policy/) - [ICO, Your right to get your data corrected (accessed September 2026)](https://ico.org.uk/for-the-public/your-right-to-get-your-data-corrected/) --- Title: Why does ChatGPT ignore our website when we have hundreds of pages? URL: https://fortitudemedia.ai/insights/ai-penalises-thin-content Author: David Adams Published: 2026-03-31 Last reviewed: 2026-09-24 Topic: Getting named by AI assistants Short answer: First rule out access: AI crawlers blocked in robots.txt or by your CDN, key text that only appears after JavaScript runs, pages missing from Bing, or no mentions on other sites. If assistants can read you, the usual cause is thin content: definitions, feature lists, padding and near-duplicate pages that give an assistant nothing worth quoting, even when they rank on Google. Check which pages AI already uses, score your twenty most important pages, then merge duplicates, deepen what matters and delete the rest. You have paid for content for years. There are three hundred pages on the site, a blog fed every fortnight, and service pages for every variation of what you do. Then someone asks ChatGPT or Perplexity who to talk to about your specialism, and the answer names two competitors with a fraction of your output. Rule out the technical causes first. If the assistants can read you and still do not name you, the cause in our experience is usually that most of those pages are thin. Not short, necessarily, but empty: they define the subject, say it is important, list some features and stop. The research points the same way. In a [Princeton and IIT Delhi study presented at KDD 2024](https://arxiv.org/abs/2311.09735), adding statistics to a page raised its visibility in AI-generated answers by about 34% and adding quotations from relevant people by about 41%, while keyword stuffing lowered it by about 8%. It was tested on a research engine, so treat it as directional. Google's own [guide to its AI features](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide), last updated in July 2026, makes the same distinction: content with "unique expert or experienced takes" counts for more than commodity content that "could originate from anyone". ## Key takeaways - Before blaming your content, check that AI crawlers can reach your pages, key text is in the HTML, Bing has indexed you and other sites mention you. - Thin content is any page that gives a reader nothing they could not have guessed: definitions, feature lists, padding, keyword repetition and near-duplicates. - A page can keep ranking on Google and still be left out of AI answers, because an assistant needs a passage worth quoting. - Near-duplicate pages for different segments, towns or keywords blur which page is yours and, at scale, can breach Google's scaled content abuse policy. - Check which pages AI assistants already use before you delete anything. - Start with your twenty most important pages, then merge, deepen or delete instead of rewriting everything. ## Is thin content the reason, or is it something else? Check four other causes first, because each makes content quality irrelevant until it is fixed. An assistant cannot use a page it cannot reach or read, or a firm nobody else mentions. | Possible cause | How to check | What to do | |---|---|---| | AI crawlers are blocked | Read your robots.txt for GPTBot, OAI-SearchBot, ClaudeBot and PerplexityBot, and check your Cloudflare or CDN bot settings; Cloudflare has blocked AI crawlers by default on new domains since July 2025 | Allow the search crawlers you want; ChatGPT search needs OAI-SearchBot | | Key text only appears after JavaScript runs | View the page source, not the rendered page, and search for your main heading and first paragraph | Serve the text in the HTML; GPTBot, ClaudeBot and PerplexityBot do not run JavaScript | | Bing does not have your pages | Search Bing for your firm and your key pages | Submit your sitemap in Bing Webmaster Tools; Copilot draws on Bing and ChatGPT search has too | | Nobody else mentions you | Look for your firm in trade press, directories and other firms' sites | Earn coverage and listings; in an [SE Ranking study reported in November 2025](https://www.searchenginejournal.com/new-data-top-factors-influencing-chatgpt-citations/561954/), referring domains were the strongest predictor of ChatGPT citations | Our guide to [how AI crawlers read your website](/insights/ai-crawlers-vs-google-spiders) covers the first two checks in detail, and [whether backlinks still matter for AI search](/insights/ai-evaluates-quality-referring-domains) covers the fourth. ## What do AI assistants count as thin content? Any page that offers little a reader could not have guessed. Years of writing to satisfy search signals such as keywords and length left many B2B sites with long pages that say nothing, and an assistant needs a passage worth quoting. ### Definitions with no depth Thin: "Cash flow forecasting is the process of estimating the money coming into and going out of a business over a period. It helps businesses plan." Substantive: "Cash flow forecasting estimates the money moving in and out of the business week by week, usually over thirteen weeks. Unlike a budget, which sets targets for a year, a forecast is a working document you update every Monday. The point is to see the shortfall in week nine while you still have eight weeks to act on it: chase debtors, delay a purchase, arrange an overdraft. For most owner-managed firms the thirteen-week horizon is right because it covers a quarter's VAT and a full debtor cycle." The second explains how it works, how it differs from a budget, why it matters and when it applies: a definition, a distinction, a use and a reason, all quotable. ### Five other kinds of thin page - **Lists with no analysis:** "Five ways to improve cash collection: invoice promptly, chase early, check credit." True and useless until each point says why firms fail at it. - **Padding:** "Cash flow is important for B2B firms. Many B2B firms rely on cash flow." The same idea twice. - **Keyword repetition:** "Our cash flow software provides cash flow solutions for cash flow management." - **Boilerplate:** "Cash flow is a complex subject. Working with an adviser can help." Swap the subject for any other and the page reads the same. That is the test. - **Near-duplicates:** "cash flow forecasting", "guide to cash flow forecasting" and "best cash flow forecasting approach", the same ground with the words rearranged. The most common form of thinness on B2B sites, and the least noticed. Feature lists, comparison tables with no verdict and case studies with no outcomes fail the same test; [why ChatGPT never quotes our blog](/insights/why-ai-ignores-most-blog-posts) covers those. ## How does an assistant decide a page is thin? Nobody outside the AI companies can see the exact process, but the pages we see quoted share five qualities, the same ones a sceptical buyer looks for. **Density:** "spending varies and depends on your needs" says nothing; a range, what drives it and a named source says something. **No repetition:** each sentence adds a new idea. **Specificity:** "you should use LinkedIn" is generic; "for firms selling to IT directors, LinkedIn advertising is worth testing, with a landing page built for that one audience" is specific. **Support:** "the return is typically eight to one", on its own, is a red flag. **Coherence:** the page sets out an argument and works through it. How assistants judge whether the writer knows the subject is covered in [what E-E-A-T is and why AI cares](/insights/eeat-why-ai-cares-more). ## Why can a page rank on Google and still be left out of AI answers? Because ranking and being quoted are related but different tests. Ranking helps: Google's AI features, Copilot and ChatGPT search all start from a search index. But [Ahrefs found in March 2026](https://ahrefs.com/blog/ai-overview-citations-top-10) that only 38% of AI Overview citations came from pages ranking in the top ten, down from 76% in July 2025, because Google now runs several related searches for one question (query fan-out). What we see in the answers we record is exclusion more than demotion: a thin page that ranks well is simply not quoted, while a more specific page further down is. We cannot see inside the assistants, so treat that as observation. For example, a broker ranking second for "commercial property insurance broker" with generic pages may be named less often than a competitor ranked fifth with detailed pages on underinsurance and insuring listed buildings. [Why ranking first on Google does not guarantee a mention](/insights/google-rankings-wont-protect-you) covers the wider problem. ## Why do duplicate pages do more damage than you think? Because they blur which page is yours, and at scale they can break Google's rules. Google's scaled content abuse policy, [introduced in March 2024](https://developers.google.com/search/blog/2024/03/core-update-spam-policies), says: "Scaled content abuse is when many pages are generated for the primary purpose of manipulating search rankings and not helping users." It applies "no matter how it's created". Google's [AI features guide](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide) adds that creating separate content for every variation of how people search, "primarily to manipulate rankings or generative AI responses", violates that policy, and its [spam policies](https://developers.google.com/search/docs/essentials/spam-policies) list "pages targeted at specific regions or cities that funnel users to one page" as doorway abuse. Mass location and segment pages with the same body are the obvious risk. Duplication creeps in for sensible-sounding reasons: one page per segment, per town, per channel or per keyword variation, each with the same body. An assistant reading them cannot tell which page is the real one. When you update one and not the others, the site contradicts itself, which is worse than being thin; [why ChatGPT describes firms wrongly](/insights/ai-conflicting-information-online) shows the result. Near-duplicates need not be word for word: rearranging a sentence does not make it new. Google's spam systems can act against a whole site; whether AI assistants mark down a whole domain in the same way is unproven. The practical cost is plainer: four pages saying one thing give an assistant one idea to quote. ## How do I find out which of my pages ChatGPT has cited? Look in four places before you delete anything. Pages that assistants send people to, or fetch to answer questions, are the last ones to cut. | Where to look | What it shows | Limits | |---|---|---| | GA4 | Since 13 May 2026, a default "AI Assistant" channel for visits from ChatGPT, Gemini, DeepSeek, Copilot and Grok, which you can break down by landing page ([Nice Looking Data, July 2026](https://www.nicelookingdata.com/blog/ga4-ai-traffic-chatgpt-referrals)) | Claude and Perplexity are not included, so filter session source for claude.ai and perplexity.ai; clicks from AI Overviews and AI Mode count as Organic Search | | Server logs | Hits from ChatGPT-User, Claude-User and Perplexity-User, which fetch a page when someone asks a question | Shows pages read, not whether they were cited | | Search Console | A [generative AI performance report](https://developers.google.com/search/blog/2026/06/gen-ai-performance-reports) with impressions and pages for AI Overviews and AI Mode, available to all sites since 31 August 2026 | Google's AI features only | | Your own prompt checks | Which of your pages are cited when you ask 20 to 50 buyer questions, each several times in a clean session | Answers vary from run to run, so count a page only when it repeats | Our guide to [finding out whether ChatGPT recommends your firm](/insights/track-ai-recommending-your-business) sets up the prompt checks. ## How to audit your content for thinness ### Step 1: Score your twenty most important pages Choose them by traffic, AI use and commercial importance, and score each with the method in our [guide to auditing content for AI assistants](/insights/audit-existing-content-ai-readiness), so every page is judged the same way. For a quick first pass, ask: would I learn something, is it specific, and does every paragraph add something? ### Step 2: Find the duplicates Search Console's page indexing report shows pages Google treats as duplicates, and tools such as Screaming Frog or Copyscape find near-duplicates. Look especially for the blog post, white paper and case study on one subject, and "beginner's" and "expert" guides that cover the same ground. ### Step 3: Compare with the pages assistants cite Ask ChatGPT and Perplexity the questions your buyers ask, read the pages they cite, and compare their depth with yours on the same subject. Where you are thinner, you have found the gap. ### Step 4: Mark every page merge, deepen, keep or delete Record the decision next to the score and the AI use from the previous section. That list is your work plan. ## How to fix thin content without rewriting everything Merge the duplicates, deepen the pages that matter, delete the rest, in that order. **Merge duplicates.** Combine three segment pages into one with a section per segment, or make them different in substance. "Software for small firms: five options under a set budget" and "software for enterprise: multi-region requirements" have different decision criteria, which makes them segmentation. **Deepen the pages that matter.** Give each point on a thin "five ways to improve" page a proper paragraph or two: why firms neglect it, a real or labelled example, how long it takes, what changes when it is done. Where one thin page covers a big subject, replace it with a main guide and three or four deeper pages under it, as set out in [organising content so AI assistants see you as the expert](/insights/topic-clusters-ai-understands). **Delete what cannot be saved.** Pages that do not rank, do not convert, are not used by assistants and duplicate what competitors do better should go, with the address redirected to the best remaining page. Keep old pages only if they are still accurate. ## What does a rebuild look like in practice? A large site is not the problem when each page has a distinct job. [arum](/work/arum), which advises banks, governments and major brands on collections technology, had "more than 500 projects in 23 countries, from offices in London, Atlanta and Sydney" to describe. The rebuild: "310 pages rebuilt across three regions, in weeks." "215 articles, research pieces and news items moved across, going back to 2014." | Principle | What the arum rebuild did | |---|---| | Give each page one job, organised around how buyers buy | "Services organised the way a client buys them: five stages from advice to optimisation, plus three specialist services." | | Deepen the pages buyers use to decide | "Each of the 12 case studies leads with a number, and each of the nine sector pages names the regulation in play." | | Say each fact once, and keep what ranks | "Proper structure and markup on every page, a plain summary AI assistants can read, facts stored once so two pages never disagree, and every old address redirected so years of rankings carry over." | As arum's CEO, Jamie Waller, put it: "For most people, our website is the first time they meet arum. It now does us justice." ## How do we stop thin content coming back? Apply a short checklist before anything goes live: the page answers a specific question a buyer asks, gives at least three usable points, includes a concrete example or sourced figure, differs clearly from what competitors have published, and adds something in every section. Length is not on the list. Concise content is specific in few words; thin content is generic at any length, as [whether ChatGPT prefers long articles or short ones](/insights/long-form-vs-short-form-ai-prefers) explains. Review the library once a year, merging what has drifted together. ## What to do this month 1. Run the four technical checks: robots.txt and CDN bot settings, page source, Bing, and mentions on other sites. 2. Pull AI referrals by landing page from GA4 and ChatGPT-User hits from your server logs. 3. Score your twenty most important pages with the audit method and mark each merge, deepen, keep or delete. 4. Run a duplicate check and merge or differentiate each group, starting with the subjects that matter most commercially. 5. Add real depth to the five thin pages with the most traffic: examples, reasons and figures with sources. 6. Give the pre-publication checklist to everyone who writes for you. ## Frequently asked questions ### Does having too many pages hurt my website in Google and ChatGPT? Page count is not the problem; empty and duplicate pages are. Sites that break Google's spam policies "may rank lower in results or not appear in results at all". AI assistants have no published rule on page count, but a site where most pages repeat each other gives them little to quote. ### Should I delete old blog posts that get no traffic? Check first whether assistants use them, in GA4 and your server logs. A post with no traffic, no AI visits, no links and a better page on the same subject should be merged or deleted, with the address redirected. A post that is accurate and says something no other page does can stay. ### Are location pages for every town bad for SEO and AI search? Pages that swap a town name into the same text are what Google's doorway abuse policy describes. A location page earns its place when it says something true only of that place, such as the team based there. For most B2B firms, one page per real office is enough. ### Can AI-generated pages get my site penalised by Google? They can if they are produced at scale and add nothing: Google's scaled content abuse policy applies "no matter how it's created". Use AI to draft a structure or tidy prose, then add what only you have: your examples, figures and judgement. ### Will merging duplicate pages hurt our Google traffic? Sometimes, briefly, because three addresses become one. In our experience the stronger single page usually recovers within a few months. Redirect every old address to the merged page. ### How long before fixing thin content changes what AI assistants say about us? Assistants that search the web live (ChatGPT search, Perplexity, Copilot, Google's AI features) can reflect a rebuilt page within weeks; answers from a model's memory change only when it is retrained, which takes months. ID-Pal is a dated example. Its case study reports: "UK buyer questions where AI names ID-Pal: none in May, two in August, four in September 2026." And: "Where a page has been rebuilt, ID-Pal appears. Where it has not yet, it does not, which tells us exactly what to build next." Read the [ID-Pal case study](/work/id-pal). Most firms have far more expertise than their website shows, and the fix is bringing it to the surface. To see how your firm is described today, [book a free Growth Review](/growth-review): 30 minutes with David Adams and three fixes in writing. If the site itself needs rebuilding, that is [Website That Sells](/website-that-sells). ## Sources - [arXiv, Aggarwal and others, Princeton and IIT Delhi: source visibility in AI-generated answers, KDD 2024 (June 2024)](https://arxiv.org/abs/2311.09735) - [Google Search Central, Guide to optimizing for generative AI features on Google Search (July 2026)](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide) - [Google Search Central, Spam policies for Google web search (August 2026)](https://developers.google.com/search/docs/essentials/spam-policies) - [Google Search Central Blog, What web creators should know about our March 2024 core update and new spam policies (March 2024)](https://developers.google.com/search/blog/2024/03/core-update-spam-policies) - [Ahrefs, AI Overview citations and top-ten rankings (March 2026)](https://ahrefs.com/blog/ai-overview-citations-top-10) - [Search Engine Journal, New data: top factors influencing ChatGPT citations (November 2025)](https://www.searchenginejournal.com/new-data-top-factors-influencing-chatgpt-citations/561954/) - [Nice Looking Data, AI traffic in GA4: how ChatGPT, Perplexity, Gemini and AI Mode referrals show up (July 2026)](https://www.nicelookingdata.com/blog/ga4-ai-traffic-chatgpt-referrals) - [Google Search Central Blog, Generative AI performance reports in Search Console (June 2026)](https://developers.google.com/search/blog/2026/06/gen-ai-performance-reports) --- Title: Why does ChatGPT name other debt collection agencies and not us? URL: https://fortitudemedia.ai/insights/ai-optimisation-debt-collection-credit-management Author: Ross Williams Published: 2026-03-31 Last reviewed: 2026-09-24 Topic: Sector guides Short answer: ChatGPT recommends debt collection agencies it can verify and trust: a clear specialism, checkable FCA authorisation and CSA membership, and plain guidance for creditors. It also reads what debtors say about you. Publish separate commercial and consumer pages naming the rules that apply, dated guides to the Late Payment Act and the coming reforms, country pages for overseas recovery, and a fair, clear page for the people you contact, then make every listing match. For example, a finance director at an engineering business has a customer who owes £80,000 and has stopped answering the phone. She asks ChatGPT whether to use a debt collection agency or a solicitor, then which B2B agencies are best for manufacturers. She gets a sensible answer and four names. Your firm recovers commercial debt for manufacturers every week. It is not one of the four. At the same time, someone your team wrote to last week types your firm's name into ChatGPT and asks "is this a real company, and do I have to pay them". The answer draws on reviews from other people you have contacted, on your website and on the FCA Register. If that answer is muddled or hostile, it is not only the debtor who sees it. Creditors ask the same assistants about you before they sign. arum shows what a collections firm's website can carry. A collections and recoveries firm with 28 years behind it, arum advises banks, governments and major brands on collections technology, with "more than 500 projects in 23 countries, from offices in London, Atlanta and Sydney". We rebuilt its site: "310 pages rebuilt across three regions, in weeks", with "Proper structure and markup on every page, a plain summary AI assistants can read, facts stored once so two pages never disagree, and every old address redirected so years of rankings carry over." Jamie Waller, its CEO, says: "For most people, our website is the first time they meet arum. It now does us justice." ([arum case study](/work/arum)). The government says late payment "causes the closure of 38 businesses per day" ([Department for Business and Trade, July 2026](https://www.gov.uk/government/consultations/late-payments-tackling-poor-payment-practices)), and finance directors now research recovery options with an assistant. This guide is for the founder or managing director of a UK collections, recoveries or credit management firm. ## Key takeaways - Finance directors and credit managers now ask ChatGPT who to use for debt recovery, and most agencies give it too little to go on. - AI assistants name agencies they can verify: a clear specialism, FCA authorisation and CSA membership where relevant, and guidance that shows how you work. - Commercial and consumer collections run under different rules, so publish separate, clearly labelled pages for each, because finance directors, utilities and councils ask different questions. - The late payment reforms confirmed in March 2026 (60-day maximum payment terms, non-waivable statutory interest and stronger Small Business Commissioner powers) are not yet law, so date anything you write about them. - The people you collect from ask assistants about you too, and their reviews shape the answer your prospective clients read. - A named specialism, such as commercial debt for manufacturers or utility arrears, gets you named far more often than "full-service collections". ## What do creditors and debtors ask ChatGPT about collections? Commercial creditors ask how to get paid and who can help; consumer creditors ask who will protect their brand; debtors ask whether they have to pay and who you are. These are typical of the questions that decide who gets named. | Who asks | Typical question | What the answer leans on | |---|---|---| | Finance director, commercial debt | "Should I use a debt collection agency or send a letter before action myself?" | Plain guides comparing the routes | | Credit manager | "What statutory interest can I charge on a late invoice in 2026?" | GOV.UK guidance, dated agency explainers | | Finance director, overseas debtor | "How do I recover an unpaid invoice from a company in Ireland or Spain?" | Agencies with named partner networks and country pages | | Utility, council or lender | "Which agencies handle vulnerable customers well?" | FCA and CSA standing, published treatment policies, debtor reviews | | Anyone checking | "Which debt collection agencies are CSA members and FCA authorised?" | The CSA member directory, the FCA Register, your website | | Debtor | "Is this debt collector legitimate, and can they come to my house?" | Your page for the people you contact, reviews, the FCA Register | Most creditor questions are about a decision, not an agency. An assistant answers the question and then names firms it has evidence for, so the firm with the clearest answer to "agency or solicitor" is the obvious one to mention. Law firms compete for the same questions, as our [guide for law firms](/insights/ai-optimisation-law-firms) shows. ## How does ChatGPT decide which agencies to name? It names the agencies it can verify and describe with confidence. Four things decide it: clear standing (FCA authorisation where required, CSA membership, CICM qualifications among your people and a company record that matches your trading name); a stated specialism; evidence of how you work at each stage; and what others say, including the tone of what debtors write. Assistants are cautious with anything involving money owed. If they cannot tell whether you are authorised, what you do or how you treat people, they hedge, and a firm they hedge about is easy to leave off a list. Firms that trade under several names often have an FCA Register entry, a Companies House record and a website that each tell a different story. We explain how to fix that in [why ChatGPT describes firms wrongly](/insights/ai-conflicting-information-online). ## Commercial collections: what should a B2B agency publish? Plain, dated guides to the decisions a finance director faces: which route to take, what they can add to the debt and what it will cost, because the rules are about to change. ### What can a creditor add to a late invoice today? Under the Late Payment of Commercial Debts (Interest) Act, a business can charge statutory interest of "8% plus the Bank of England base rate for business to business transactions" ([GOV.UK, September 2026](https://www.gov.uk/late-commercial-payments-interest-debt-recovery/charging-interest-commercial-debt)), plus a fixed sum for recovery costs of £40 on debts under £1,000, £70 on debts from £1,000 to just under £10,000 and £100 on £10,000 or more, and reasonable further recovery costs ([GOV.UK, September 2026](https://www.gov.uk/late-commercial-payments-interest-debt-recovery/claim-debt-recovery-costs)). ### What will the late payment reforms change? The government's response to its late payment consultation, published in March 2026, confirmed it will legislate for a 60-day maximum payment term with limited exemptions, statutory interest at 8% above base rate that parties can no longer contract out of, and powers for the Small Business Commissioner to investigate, fine and settle disputes outside court ([Department for Business and Trade, March 2026](https://www.gov.uk/government/consultations/late-payments-tackling-poor-payment-practices/outcome/late-payment-consultation-time-to-pay-up-government-response-web-version)). A proposed cut to 45 days was not taken forward. As of GOV.UK's July 2026 update, the government intends to legislate "as soon as Parliamentary time allows", so none of this is law yet. Say so on your page, and review it when a bill is published. ### Agency, letter before action, court claim or solicitor? | Route | Best suited to | How the creditor pays | What decides the time | |---|---|---|---| | Letter before action from the creditor | Undisputed debts where the customer is still trading and talking | Staff time; statutory interest and fixed recovery costs can be added | How quickly the debtor replies | | Collection agency | Undisputed debts, higher volumes, debtors who need tracing or chasing by phone | Usually commission on what is recovered, sometimes fixed fees | Whether the debtor engages or the case needs legal action | | County court claim (England and Wales) | Debts with clear evidence, where the debtor may dispute | A court fee set by the government, rising with the claim value | Whether the claim is defended; disputed claims of £10,000 or less go to free court mediation first ([GOV.UK, November 2024](https://www.gov.uk/make-court-claim-for-money)) | | Solicitor | Larger or disputed debts, insolvency risk, enforcement | Legal fees, some recoverable depending on the claim | Complexity, defences and enforcement | Scotland and Northern Ireland have their own court processes, so say which jurisdictions you handle. ### How do we answer overseas recovery questions? When a finance director asks how to recover a debt from a company in Germany, Ireland or Spain, the assistant explains the local process and names firms that show a partner network. FENCA, the Federation of European National Collection Associations, has 25 national member associations, and the UK's Credit Services Association is one of them ([FENCA, September 2026](https://www.fenca.eu/)); in the United States, [ACA International](https://www.acainternational.org/) plays a similar role. The UK has also been a party to the Hague Judgments Convention 2019 since 1 July 2025 ([HCCH, September 2026](https://www.hcch.net/en/instruments/conventions/status-table/?cid=137)), which can make a UK judgment easier to enforce in EU countries. Publish a page for each country you recover in, naming your partner, the process, the languages you work in and your memberships. ## Consumer collections: what changes when the debt is regulated? The rules, the audience and the tone. Collecting debts due under credit agreements is an FCA-regulated activity ([Regulated Activities Order, article 39F](https://www.legislation.gov.uk/uksi/2001/544/article/39F)), and those agreements include credit to sole traders and to partnerships of two or three partners, not only consumers ([article 60L](https://www.legislation.gov.uk/uksi/2001/544/article/60L)). Collecting a company's unpaid invoices from another company is not regulated in the same way. Say which kinds of debt you collect, and show your FCA reference number wherever it applies. Regulated firms also work under the FCA's Consumer Duty, in force since 31 July 2023 for products on sale ([FCA, July 2022](https://www.fca.org.uk/publications/policy-statements/ps22-9-new-consumer-duty)). Where the debtor is an individual, including a sole trader, the Pre-Action Protocol for Debt Claims applies, and if the debtor does not reply to the letter of claim "within 30 days", the creditor may start court proceedings ([Ministry of Justice](https://www.justice.gov.uk/courts/procedure-rules/civil/pdf/protocols/debt-pap.pdf)). The protocol "does not apply to business-to-business debts unless the debtor is a sole trader". Utilities, councils and lenders buying consumer collections ask about vulnerability, complaints and brand protection before price. Publish your approach to customers in vulnerable circumstances, how you handle disputes and how complaints can go to the Financial Ombudsman Service, written for the creditor's compliance team. ## Why do the people you collect from shape your reputation? Because the assistants read what they write. Reviews, forum posts and complaints are public, and when anyone asks about your firm, creditor or debtor, the assistant draws on them. You cannot stop people disliking being chased for money, but you can make the full picture visible. Start with a page for the people you contact. It should explain, calmly and in plain words, who you are and who you act for; how to check you are genuine, including your FCA reference number; how to pay, set up an arrangement or dispute a debt; where to get free, independent advice from MoneyHelper, StepChange, National Debtline and Citizens Advice; and how to complain. That page answers the debtor, gives the assistant a source to quote when someone asks whether you are legitimate, and shows prospective clients you will protect their brand. Reply to reviews politely and factually, and ask satisfied creditor clients to review you on Google or Trustpilot, so the record includes the people who hire you. ## Which outside sources matter most for a collections firm? The assistants trust independent sources over your own claims, so check these first. | Source | Why it matters | What to do | |---|---|---| | FCA Register | Confirms permissions and trading names | Make names, permissions and address match your website | | [Credit Services Association](https://www.csa-uk.com/) | "The sole national trade association in the UK" for debt collection and purchase, with "over 250 member companies" and a member directory searchable by service and location | Keep your listing accurate and reference its Code of Practice | | [Chartered Institute of Credit Management](https://www.cicm.com/) | Qualifications, Credit Management magazine and the British Credit Awards | Show CICM qualifications on named team profiles | | FENCA and ACA International | Evidence of an overseas network | Name your partners and memberships | | Companies House and the ICO register | Tie your company and data protection registration to your trading name | Check both match | | Trade press | Credit Connect, CICM's Credit Management magazine and your clients' sector titles | Offer comment and anonymised data | Published data helps most of all. Anonymised, aggregated findings, such as how payment behaviour differs by sector, get quoted by trade press and cited by assistants. Our article on [publishing your own research](/insights/original-research-data-ai-trust) explains how to do that without exposing client information. ## Why does a named specialism win more work? Because most creditor questions are specialist, and an agency clearly known for one or two kinds of debt will be named for those far more often than a generalist is named for anything. A website promising "debt recovery solutions for every sector" reads as a generalist; rebuilt around, say, manufacturing and construction, with named specialists, the same firm becomes a candidate whenever a finance director in those sectors asks. arum's site follows this pattern: "Each of the 12 case studies leads with a number, and each of the nine sector pages names the regulation in play." The same applies to utility arrears, council tax, telecoms and lending, where a deep guide written by the people who do the work will be quoted in a way a general page never is. ## How do we tell whether it is working? Keep a fixed list of 20 to 50 questions a prospective client would ask, plus the questions debtors ask about your firm by name, and run each several times a month in a clean session in ChatGPT, Google's AI Overviews and AI Mode, Copilot, Gemini, Perplexity and Claude. Count a win only when it repeats, because answers vary between runs, and note whether the description of your conduct is fair. Assistants that search the web live can reflect a corrected listing within weeks; answers drawn from a model's own memory change only when it is retrained, which takes months. Our guide to [checking whether ChatGPT recommends your firm](/insights/track-ai-recommending-your-business) sets out the method. ## What to do this month 1. Check that your FCA Register entry, CSA listing, Companies House record, Google Business Profile and website agree on names, permissions and address. 2. Split your services into clearly labelled commercial and consumer sections, each naming the rules that apply. 3. Write or update the page for the people you contact: who you are, how to check you are genuine, how to pay or dispute, where to get free advice and how to complain. 4. Publish one dated creditor guide covering what the Late Payment Act allows today and the status of the reforms. 5. Add a page for each country where you recover debts, naming your partner and memberships. 6. Invite a handful of satisfied creditor clients to review you on Google or Trustpilot. ## Frequently asked questions ### What percentage does a debt collection agency charge in the UK for commercial debt? We do not quote market rates here, because commission varies with the size and age of the debt, the sector, the debtor's location and whether legal action is included. What an agency should publish is how its fees work: what commission is charged on, how legal costs are handled and what happens to statutory interest and compensation. That answers the question creditors ask assistants and builds the trust that wins panel places. ### Should I use a debt collection agency or send a letter before action myself? For a recent, undisputed debt with a customer who is still trading, a firm letter claiming statutory interest and fixed recovery costs often works. An agency adds tracing, persistence and scale; a solicitor or court claim suits disputed or larger debts. ### Which debt collection agencies are CSA members and FCA authorised? Assistants answer this from the CSA member directory, the FCA Register and agencies' own websites, so all three need to agree. Show your CSA membership and FCA reference number on your website in plain text, with links to both records. ### How do I recover an unpaid invoice from a company in Ireland or Spain? Usually through an agency with a partner in that country, followed by local legal action if needed. The UK has been party to the Hague Judgments Convention 2019 since 1 July 2025, which can help enforce a UK judgment in EU countries. Agencies that publish country pages naming their partner and process are the ones assistants name. ### What statutory interest can I charge on a late invoice in 2026? Between businesses, statutory interest is 8% plus the Bank of England base rate, with a fixed recovery sum of £40, £70 or £100 depending on the size of the debt. The reforms confirmed in March 2026 would stop parties contracting out of statutory interest, but they need legislation first. If your website still undersells the work you do, [Website That Sells](/website-that-sells) is how arum's site was rebuilt. A free [Growth Review](/growth-review) is 30 minutes with David Adams, with three fixes in writing, and a good way to see what ChatGPT tells a finance director, or a debtor, about your firm today. ## Sources - [Department for Business and Trade, Late payments: tackling poor payment practices (July 2026)](https://www.gov.uk/government/consultations/late-payments-tackling-poor-payment-practices) - [Department for Business and Trade, Time to pay up: government response (March 2026)](https://www.gov.uk/government/consultations/late-payments-tackling-poor-payment-practices/outcome/late-payment-consultation-time-to-pay-up-government-response-web-version) - [GOV.UK, Late commercial payments: charging interest and debt recovery (September 2026)](https://www.gov.uk/late-commercial-payments-interest-debt-recovery/charging-interest-commercial-debt) - [Ministry of Justice, Pre-Action Protocol for Debt Claims (September 2026)](https://www.justice.gov.uk/courts/procedure-rules/civil/pdf/protocols/debt-pap.pdf) - [GOV.UK, Make a court claim for money (November 2024)](https://www.gov.uk/make-court-claim-for-money) - [Credit Services Association, About the CSA (September 2026)](https://www.csa-uk.com/) - [FENCA, Federation of European National Collection Associations (September 2026)](https://www.fenca.eu/) - [HCCH, Status table: 2019 Judgments Convention (September 2026)](https://www.hcch.net/en/instruments/conventions/status-table/?cid=137) --- Title: Why does ChatGPT never quote our blog, and how do we fix it? URL: https://fortitudemedia.ai/insights/why-ai-ignores-most-blog-posts Author: Ashley Kenerson Published: 2026-03-31 Last reviewed: 2026-09-24 Topic: Content that gets quoted Short answer: Usually one of three things, in this order. The assistant cannot reach the post, because AI crawlers are blocked in robots.txt or by Cloudflare, or the text only appears after JavaScript runs. The post does not rank in Google or Bing for the searches the assistant runs, so it is never a candidate. Or it is found but gives the assistant nothing to quote: vague, anonymous, badly structured or saying what every competitor says. Check in that order, then rewrite your twenty most important posts rather than all of them. Your firm has been blogging for years. There are two hundred posts on the site; some rank and a few bring in enquiries. Then someone asks ChatGPT the question one of those posts was written to answer, and the assistant quotes a trade body, a national firm and a software vendor. Try Gemini, Perplexity and Google's AI Overviews, and the result is the same. The usual explanation is that AI "judges content differently from Google". That is only half true. ChatGPT search, Perplexity, Copilot and Google's AI features start by running web searches, so they mostly quote pages they can find. In an [AirOps study reported by Search Engine Land in April 2026](https://searchengineland.com/chatgpt-citations-ranking-precision-length-study-474538), pages in the top search position were cited by ChatGPT 58.4% of the time, against 14.2% for pages in position ten. So an ignored post fails at one of three gates: the assistant cannot reach it, it never turns up in the searches the assistant runs, or it turns up and gives the assistant nothing worth quoting. Check them in order, because a rewrite fixes nothing if the first two are closed. ## Key takeaways - Check access first: AI crawlers blocked in robots.txt or by Cloudflare, or text that only appears after JavaScript runs, make every post invisible whatever its quality. - Ranking and citation go together: a post that does not rank in Google or Bing for the question, or for the related searches the assistant runs, is rarely a candidate. - Some questions are not blog questions: "best firms for X" prompts are answered from lists, reviews and directories, which a post rewrite cannot change. - When a post is found and still ignored, the causes are vagueness, no visible expertise or author, poor structure and no point of view. - Rewrite the twenty posts that matter most; in our experience a proper rewrite takes ten to twenty hours each. - Measure with a fixed list of buyer questions asked every month, and count a win only when it repeats. ## Why does ChatGPT cite a competitor's blog but not ours? Their post passed all three gates; yours stopped at one. Work through these checks in order and stop at the first that fails. 1. **Can AI crawlers reach the post?** Check robots.txt and your CDN. If not, fix access before anything else. 2. **Is the text in the page's HTML?** View the page source and search for your first paragraph. If it is missing, the text is rendered by JavaScript, which the main AI crawlers do not run. 3. **Is the post indexed in Google and Bing?** Search each for the post's exact title. If it is missing from Bing, submit your sitemap in Bing Webmaster Tools. 4. **Does it rank for the question and its follow-ups?** Search the buyer's question and two or three related ones. If you are nowhere, it is a ranking problem as much as a writing one. 5. **Is the question one a blog post can answer?** If buyers ask "who are the best firms for X", the answer comes from lists, reviews and directories. 6. **Only now: is it a content problem?** If the post passes all five checks and is still ignored, the four content reasons below apply. ## Is ChatGPT blocked from my website, and how do I check? Open yoursite.co.uk/robots.txt and look for the AI crawlers by name. [OpenAI says](https://help.openai.com/en/articles/9237897-chatgpt-search) a site must allow OAI-SearchBot, and its host or CDN must allow OpenAI's published IP addresses, to be eligible for ChatGPT search. The names that matter: | Company | Crawlers | Allow for AI search | |---|---|---| | OpenAI | GPTBot (training), OAI-SearchBot (ChatGPT search), ChatGPT-User (fetches a page when a user asks) | OAI-SearchBot | | Anthropic | ClaudeBot (training), Claude-SearchBot (search), Claude-User (user requests) | Claude-SearchBot | | Perplexity | PerplexityBot (search), Perplexity-User (user requests) | PerplexityBot | | Google | Googlebot (Search and AI features); Google-Extended controls Gemini training only and does not affect AI Overviews | Googlebot | | Microsoft | Bingbot (Bing and Copilot) | Bingbot | Then check your CDN. [Cloudflare changed its default to block AI crawlers in July 2025](https://blog.cloudflare.com/content-independence-day-no-ai-crawl-without-compensation/) and updated its defaults for new domains again on 15 September 2026, so look at the AI bot settings in your dashboard even if nobody on your side chose to block anything. Finally, check JavaScript. [Vercel and MERJ found in December 2024](https://vercel.com/blog/the-rise-of-the-ai-crawler) that GPTBot, ClaudeBot and PerplexityBot fetch JavaScript files but do not run them, so a blog whose text loads by script is blank to them. Googlebot does render JavaScript, which helps Gemini and Google's AI features. Our guide to [how AI crawlers read your website](/insights/ai-crawlers-vs-google-spiders) covers each check in detail. ## Does ChatGPT use Bing or Google to find the pages it cites? It uses partners, and Microsoft is one of them. OpenAI's help page says ChatGPT search "sometimes partners with other search providers" and rewrites your question into targeted queries it sends them, and it lists Microsoft among those partners. In a small [Seer Interactive test in February 2025](https://www.seerinteractive.com/insights/87-percent-of-searchgpt-citations-match-bings-top-results), 87% of ChatGPT search citations matched Bing's top results. Copilot draws on Bing; Gemini, AI Overviews and AI Mode draw on Google's index. So a post missing from Bing can be invisible to ChatGPT and Copilot. Check Bing Webmaster Tools as carefully as Search Console. ## Why can a post rank on Google and still not be quoted? Usually because it ranks for the main query but not for the related searches the assistant actually runs, or because it ranks and has nothing quotable in it. Ranking and citation are closely linked, as the AirOps figures show, but they are not the same. [Ahrefs found in March 2026](https://ahrefs.com/blog/ai-overview-citations-top-10) that only 38% of AI Overview citations came from pages in Google's top ten, down from 76% in July 2025, because Google now runs several related searches for one question. A post that answers only the headline question misses the follow-ups. The fix is to cover the follow-up questions a buyer asks, in their words. If the post ranks for them and is still passed over, it is a content problem. ## Is the question one a blog post can answer? Often it is not. Most blog posts answer informational questions, such as "how does X work?" or "what does X cost?". Buyer prompts that name suppliers, such as "best commercial property solicitors in Leeds" or "which firm should we use for X?", are answered from list articles, review sites, directories and press coverage. In [G2's April 2026 survey of 1,076 software buyers](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html), 45% said citations of review sites were the most confidence-inspiring signal in an AI answer. | Question type | Example | What usually gets cited | What changes it | |---|---|---|---| | Informational | "How long does a lease renewal take?" | Guides and blog posts | Rewriting the post | | Comparison | "Firm A or Firm B for X?" | Comparison pages, reviews | A fair comparison page on your site | | Shortlist | "Best firms for X in the UK" | Lists, directories, review sites, press | Reviews, listings and coverage | | Reputation | "Is Firm A any good?" | Reviews, press, your own site | Reviews and consistent facts everywhere | Comparison pages are the bridge. For [Helm Club](/work/helm-club), "We built the structured data AI engines read, then a comparison page for every question a founder asks when weighing up peer groups: Helm against YPO, Vistage and the rest." And "Every comparison page we have published ranks first for its question on Google and the AI engines. The pages not yet built are the ones where a competitor still writes the answer." For shortlist questions, see [why ChatGPT sounds unsure about your firm](/insights/ai-cross-references-business-sources). ## If the post is found, why is it still not quoted? In our experience, for one of four reasons. To prioritise across many posts, use the scoring rubric in our [guide to auditing existing content](/insights/audit-existing-content-ai-readiness), and for sites with hundreds of near-duplicate posts, read [why ChatGPT ignores sites with hundreds of pages](/insights/ai-penalises-thin-content). ### Reason 1: the post says nothing specific The most common failure by a distance. The post covers the topic without committing to anything: "Data governance matters. Firms should implement governance to protect data quality." All true, all useless, and it could sit on any competitor's site unchanged. Compare: "In the governance projects we have run, the ones that stalled almost always created policies before changing who was allowed to use which data." That is a mechanism, a pattern and a judgement an assistant can use and has to attribute. **How to tell.** Mark every sentence a competitor could have written. If most are marked, the post is generic. **How to fix.** Add figures from your own work, named patterns, the reasons things happen and where the advice does not apply. In a [Princeton and IIT Delhi study presented at KDD 2024](https://arxiv.org/abs/2311.09735), adding statistics raised visibility in AI-generated answers by about 34%; it was tested on a research engine, so treat it as directional. ### Reason 2: nothing shows who wrote it or why they know No author, or "by the team". No example that could only come from experience. No sources, no limits. Google's [guide to its generative AI features](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide), updated in July 2026, says content with "unique expert or experienced takes" counts for more than commodity content, and an anonymous post rarely shows either. **How to tell.** Is there a named author with a role? An example only a practitioner would know? A cited source? A paragraph on where the advice fails? Three noes is a weak post. **How to fix.** Assign each subject to one named person and put a proper byline on every post: name, role and a sentence of experience, linked to a profile page. Add two or three examples from practice, anonymised if needed, cite borrowed facts and say where the advice stops working. [What makes ChatGPT treat content as expert rather than marketing](/insights/what-makes-expert-quality-content) goes further. ### Reason 3: the structure makes it hard to use A post that wanders, with no clear headings and the answer buried at the end, is hard to lift from. A [2026 analysis of 1.2 million ChatGPT answers](https://searchengineland.com/chatgpt-citations-content-study-469483) found 44% of citations came from the first 30% of a page, so a post that saves its point for the conclusion wastes its best sentence. **How to tell.** Can you outline the post from its headings in two minutes? Is the answer in the first two hundred words? **How to fix.** Lead with the answer, turn each heading into the question it answers and keep sections short. [What an article that ChatGPT actually quotes looks like](/insights/anatomy-of-article-cited-by-ai) sets out the full shape and the evidence for it. ### Reason 4: it has no point of view A post that agrees with everyone gives an assistant no reason to choose it over the fifty others that say the same. A point of view is something noticed from the work: that the usual advice fails in a common situation, or that one step matters far more than the others. **How to tell.** Would a competitor read it and say "that is what I would have written"? Does it disagree with received wisdom anywhere? **How to fix.** Ask the practitioners what they believe the industry gets wrong, and what they have seen that supports it, then rebuild the post around that. This is the hardest fix, because it needs thinking as well as editing. ## How do we rewrite a post? Do this for the twenty posts that matter most: those on your core services, those that still get traffic and those that will stay relevant. In our experience the whole process takes ten to twenty hours a post, which is why you choose twenty rather than two hundred. ### Step 1: Diagnose it Run the six checks above and write down which gate the post fails. About an hour. ### Step 2: Gather what it is missing Talk to the practitioner, pull figures from your own records and find the sources worth citing. This is the step most people skip, and where the value is. Three to eight hours. ### Step 3: Outline the new version The buyer's question as the title, the answer in the first paragraph, the follow-up questions as headings. About an hour. ### Step 4: Rewrite it Write it fresh rather than patching the old text. Keep the specifics, cut the filler, add the reasons and the exceptions. Four to ten hours. ### Step 5: Add the structure and the author Key points near the top, the buyer's follow-ups at the end, links to related pieces, a proper byline and a last-reviewed date. One to two hours. ## How do I see whether ChatGPT has ever cited our site? Partly from analytics, fully only by asking. GA4 added a default ["AI Assistant" channel on 13 May 2026](https://www.nicelookingdata.com/blog/ga4-ai-traffic-chatgpt-referrals) covering visits from ChatGPT, Gemini, DeepSeek, Copilot and Grok, though not Claude or Perplexity, and clicks from AI Overviews and AI Mode count as Organic Search. Search Console's generative AI reports show your impressions in AI Overviews and AI Mode. Your server logs show ChatGPT-User visits, which happen when ChatGPT fetches a page to answer someone. None of that shows a citation nobody clicked. For that, keep a fixed list of 20 to 50 buyer questions, ask each several times a month in a clean session, and count a win only when it repeats. [How to find out whether ChatGPT is recommending your firm](/insights/track-ai-recommending-your-business) sets out the routine. ## Should we stop blogging and put the effort into other channels? No, but blog less and spread the effort. Posts that answer real buyer questions still feed the informational answers. The shortlist answers, though, lean on what others say about you: [Muck Rack found in May 2026](https://www.globenewswire.com/news-release/2026/05/07/3290268/0/en/generative-pulse-earned-media-consistently-drives-ai-citations-holding-at-84.html) that earned media made up 84% of the links AI tools cited. A sensible split is fewer, better posts, plus reviews, trade coverage and LinkedIn articles by named experts. Researched outbound, with every message written and approved by a person, reaches buyers before any assistant is asked, and works alongside all of it. ## What to do this month 1. Check robots.txt, your Cloudflare or CDN settings and the page source of three posts, and fix any access problem first. 2. Check that your twenty most important posts are indexed in Bing as well as Google. 3. For each of the twenty, search the buyer's question and two follow-ups, and note where the post ranks. 4. Sort the twenty by gate: access, ranking, wrong question type or content. 5. Book practitioner time for the top five content rewrites and start the first this week. 6. Build your fixed list of buyer questions and record who is named today, as the baseline. ## Frequently asked questions ### Should we delete the posts we are not rewriting? Not if they bring traffic or help a visitor. Keep what earns its place, and merge or retire posts that nobody visits and that repeat each other. The priority is the twenty that matter, not the tidiness of the archive. ### Can we fix one thing without a full rewrite? Sometimes. Access, indexing, bylines, headings and sources can all be fixed without touching the argument. Specificity and point of view cannot, because they change what the post says. If the diagnosis points at those, plan for a rewrite. ### What if every post we have is generic? That is common, and it usually means the expertise never made it onto the page. Ask the partners what they believe that competitors would argue with, and what supports it. That conversation produces the first five rewrites. ### Can an agency do the rewrites? An agency can do the structure, the sourcing and the writing, provided it has access to your practitioners. The arrangement that works is a writer interviewing an expert for forty minutes per piece, then drafting. The one that fails is a writer given a keyword and no expert. ### How many posts a month should we publish once the rewrites are done? Fewer than you think. One thorough piece a month by a named expert, answering a real buyer question, does more for a reader and for being quoted than four thin ones. [How often we need to publish for AI assistants to notice us](/insights/how-often-publish-content-ai) covers the evidence on frequency. If you would like an outside view of which gate your posts are failing, [book a free Growth Review](/growth-review): 30 minutes with David Adams and three fixes in writing. Turning a firm's expertise into content that gets it quoted is the work of [Reputation That Wins](/reputation-that-wins). ## Sources - [Search Engine Land, AirOps study of ChatGPT citations, ranking and length (April 2026)](https://searchengineland.com/chatgpt-citations-ranking-precision-length-study-474538) - [OpenAI Help Center, ChatGPT search (accessed September 2026)](https://help.openai.com/en/articles/9237897-chatgpt-search) - [Vercel, The rise of the AI crawler (December 2024)](https://vercel.com/blog/the-rise-of-the-ai-crawler) - [Cloudflare, Content Independence Day: no AI crawl without compensation (July 2025)](https://blog.cloudflare.com/content-independence-day-no-ai-crawl-without-compensation/) - [Seer Interactive, 87% of SearchGPT citations match Bing's top results (February 2025)](https://www.seerinteractive.com/insights/87-percent-of-searchgpt-citations-match-bings-top-results) - [Ahrefs, AI Overview citations from top-ten pages (March 2026)](https://ahrefs.com/blog/ai-overview-citations-top-10) - [G2, New G2 research: half of B2B software buyers now start their research with AI chatbots (April 2026)](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html) - [Muck Rack, Generative Pulse: earned media consistently drives AI citations (May 2026)](https://www.globenewswire.com/news-release/2026/05/07/3290268/0/en/generative-pulse-earned-media-consistently-drives-ai-citations-holding-at-84.html) --- Title: Why does ChatGPT recommend other energy firms when businesses ask? URL: https://fortitudemedia.ai/insights/ai-optimisation-energy-utilities Author: Ashley Kenerson Published: 2026-03-31 Last reviewed: 2026-09-24 Topic: Sector guides Short answer: ChatGPT names energy firms it can match to a specific job and verify. State your service, sectors, region and project sizes plainly, list accreditations such as MCS, NICEIC or OZEV authorisation with the scope each covers, and publish case studies that compare forecast with measured savings. Brokers should explain how they are paid. Dated guides to ESOS phase 4, energy standards and chargepoint grants make you the firm assistants quote on compliance. The finance director of a food manufacturer has a large roof, a larger energy bill and a board asking about net zero. She used to wait for the next cold call or ask a peer. Now she asks ChatGPT which commercial solar installers in the East Midlands have done large rooftop systems on food factories. A facilities manager at a school trust asks Copilot who specialises in heat pump retrofits for older buildings. An operations director asks Google whether to use an energy broker, and which ones can be trusted. This is now how business buyers work: [Forrester found in January 2026](https://www.forrester.com/blogs/b2b_buyers_make_zero_click_buying_number_one/) that 94% of business buyers use AI in their buying process, and that they are four times as likely as consumers to use Copilot. Each assistant replies with a handful of names. For the installer, broker, consultancy or EV charging firm that is not on the list, the enquiry never arrives. Buyers in this market are wary: many have been cold-called too often or sold savings that never appeared. Assistants reflect that caution and favour firms whose claims can be checked. This guide is for the founder or managing director of an energy services business selling to other businesses, and it also covers energy and climate tech software firms. ## Key takeaways - AI assistants recommend energy firms they can match to a specific job, sector, region and project size, so vague "net zero solutions" pages rarely help. - Accreditations such as MCS, NICEIC, OZEV authorisation and ESOS lead assessor status are independent proof, but only where they apply, so state the scope plainly. - Case studies that separate forecast savings from measured results, and explain the calculation, carry far more weight than promises. - Brokers who explain how they are paid, and which redress scheme they belong to, answer one of the questions business owners ask most. - Plain, dated guides to ESOS phase 4, energy standards for let buildings and chargepoint grants make you the firm assistants draw on for compliance questions. ## What do businesses ask ChatGPT about energy? The questions describe a site, a job and a worry, usually about cost or trust. Typical examples: - "Which commercial solar installers in the Midlands have done 500kW rooftop systems?" - "Heat pump retrofit specialists for schools in older buildings" - "How are energy brokers paid, and how do I see their commission?" - "Does ESOS phase 4 apply to my company, and when is the deadline?" - "Is the Workplace Charging Scheme still open for EV chargers in 2026?" - "Is [your firm] legitimate?" Finding questions ask who can do the job. Deciding questions ask whether it is worth doing and what the rules require. Checking questions ask whether a named firm can be trusted, and in energy they come often, usually after a proposal has landed. Most firms only prepare for the first kind. ## What does an AI assistant check before it names an energy firm? The firm that is easiest to verify usually beats the firm with the boldest claims. "We help businesses slash their energy costs" gives an assistant nothing to confirm. "We have installed rooftop solar on food and drink factories across the East Midlands since 2014; here are six projects with forecast and measured generation" gives it something solid. Our article on [what makes ChatGPT treat content as expert rather than marketing](/insights/what-makes-expert-quality-content) explains the difference. | What the assistant looks for | What good looks like | Where it cross-checks | |---|---|---| | Service, sector, region and size | "Rooftop solar and batteries for logistics sites in the Midlands, 100kW to 1MW" | Your service pages, Google Business Profile | | Accreditations, with scope | MCS, NICEIC or NAPIT, OZEV authorisation, ESOS lead assessors, ISO 9001 or 14001 | Scheme registers and your accreditations page | | Projects and results | Site type, system size, forecast against measured results | Your case studies, clients' own reports and news | | Client voice | Named quotes, reviews that mention the job | Google, Trustpilot, clients' sustainability reports | | Consistency | One name, one description everywhere | Companies House, scheme registers, directories | Energy firms often rebrand as they change technologies. If Companies House, your scheme listings, Google and your website describe you differently, an assistant struggles to connect the evidence and may name a competitor. ### Does MCS apply to us? It depends on system size. MCS's solar standard covers PV systems up to 50kWp, and its heat pump design standard covers units up to 45kWth each, with multiple units up to 70kWth in total ([MCS standards, issues dated December 2025 and March 2026](https://mcscertified.com/standards-tools-library/)). Above those sizes, buyers and assistants look for other evidence: electrical accreditation, trade association membership, ISO certification, insurance and completed projects of a similar size. Say which of your work falls inside MCS and which does not, with your certification number where it applies. ## Case studies that show measured results, not promises A case study earns trust when it separates what you forecast from what happened. Most stop at "estimated annual saving", the one number buyers have learned to discount. ### Step 1: Describe the site and starting point Site type, operating hours, annual consumption, roof or land available, existing equipment and the problem the client wanted solved. ### Step 2: Explain what you recommended, and what you advised against Telling a client a battery would not pay back on their load profile, or that insulation should come before a heat pump, is the judgement buyers and assistants trust. ### Step 3: Say what was installed System sizes, equipment, programme, how disruption was handled, and any grid connection or planning issues. ### Step 4: Compare forecast with measured results Show forecast generation or savings beside metered results for a stated period, explain any gap, and say whether savings account for tariff changes. ### Step 5: Let the client speak A named quote from the finance director or site manager carries more weight than any claim you make. ## How are energy brokers paid, and how can a client see it? Most brokers are paid commission by the supplier, and Ofgem's rules make that visible. Ofgem says information on broker costs included in the bill must be given to all businesses in the contract's principal terms and on request ([Ofgem, October 2022](https://www.ofgem.gov.uk/guidance/third-party-intermediaries-what-your-microbusiness-needs-know)). It also says a broker paid by a supplier must be a member of an approved redress scheme, and points customers to the Energy Ombudsman's broker register to check ([Ofgem](https://www.ofgem.gov.uk/your-energy-supply/how-manage-your-energy-supply/get-business-energy-contract)). Ofgem classes a microbusiness as one with fewer than 10 employees and turnover or a balance sheet of no more than £2 million, or using up to 100,000 kWh of electricity or 293,000 kWh of gas a year. If you are a broker, be more open than the rules require. Explain whether you take commission in the unit rate, a fixed fee or both, show how a client can see the amount, name your redress scheme and link to its broker register. Publish plain guides to contract terms and renewal dates. When a business owner asks an assistant whether a broker can be trusted, this is the evidence it looks for. ## Which compliance questions do buyers ask first? Many energy conversations start with an obligation, and the firm that explains it clearly often gets the work. Keep a dated page for each rule you help with, and link to the official guidance. | Rule or scheme | Who it affects | What is current (September 2026) | |---|---|---| | ESOS phase 4 | Large undertakings: 250 or more staff, or turnover above £44 million and balance sheet above £38 million | Qualification date 31 December 2026; notification deadline 5 December 2027; progress against action plans must be included ([GOV.UK](https://www.gov.uk/guidance/energy-savings-opportunity-scheme-esos)) | | ESOS phase 3 progress update | Organisations that qualified for phase 3 | Annual progress update due by 5 December 2026, signed off by a board director | | Energy standards for let commercial property | Landlords in England and Wales | EPC E minimum now; EPC B from 2031 for buildings over 1,000 square metres is proposed, not yet law ([GOV.UK, June 2026](https://www.gov.uk/government/consultations/non-domestic-private-rented-sector-minimum-energy-efficiency-standards-epc-b-implementation/outcome/minimum-energy-efficiency-standards-mees-in-the-non-domestic-private-rented-sector-interim-response)) | | Workplace Charging Scheme | Businesses, charities and public bodies | Extended for a final year to 31 March 2027; up to £500 per socket, 75% of costs, 40 sockets; claimed by an OZEV-authorised installer ([GOV.UK](https://www.gov.uk/guidance/workplace-charging-scheme-guidance-for-installers)) | | SECR | Quoted companies, large unquoted companies and large LLPs | Energy use and emissions reported in the annual report ([GOV.UK](https://www.gov.uk/government/publications/environmental-reporting-guidelines-including-mandatory-greenhouse-gas-emissions-reporting-guidance)) | Rules and deadlines change, so put a "last checked" date on each page. Pages that name the rule and answer inside the first paragraph are the ones assistants lift. When we rebuilt sector pages for [ID-Pal](/work/id-pal), the approach was the same: "Five industry pages rebuilt: fund administration, wealth management, legal, accountancy and estate agency. Each opens with the question a compliance lead types, names the regulator, and answers inside the first ninety words." Landlords facing the energy standards ask their agents too, which is why our guide for [commercial property firms](/insights/ai-optimisation-property-real-estate) covers them from the landlord's side. Installers share many issues with the trades, set out in our guide to [how construction firms get named](/insights/ai-optimisation-construction-trades). ## Energy and climate tech software firms Software buyers ask different questions: "best carbon accounting software for UK SMEs", "energy management platform for a multi-site retailer". Assistants answer them from review sites, comparison articles and vendor pages, and buyers trust the independent part most. In [G2's April 2026 survey of software buyers](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html), 45% said citations of review sites were the most confidence-inspiring signal in an AI answer. Name your category in the buyer's words, gather reviews where your buyers read them, and publish fair comparisons. Our guide for [B2B technology companies](/insights/ai-optimisation-technology-companies) covers the rest. ## Why your website may be hiding you, and how to tell Many energy websites are built around a savings calculator and a quote form. Case studies sit in PDFs, accreditations appear only as logos, and calculators load in widgets. The main AI crawlers do not run JavaScript, so key text must be in the page itself. Ask your web provider to confirm AI crawlers are not blocked, since Cloudflare has blocked them by default on new domains since July 2025. To see whether it is working, keep a fixed list of 20 to 50 buyer questions and run each several times a month in ChatGPT, Copilot, Gemini and Google in a clean session. Answers vary, so count a win only when it repeats. ## What to do this month 1. Rewrite your homepage and main service pages to state your service, sectors, region and typical project sizes in plain words. 2. List every accreditation in text on one page, with registration numbers and the scope each covers, including where MCS applies. 3. Rebuild your three strongest case studies with forecast and measured results side by side. 4. Publish a dated page on the rule your buyers ask about most, whether ESOS phase 4, energy standards or chargepoint grants. 5. If you broker energy, publish how you are paid and which redress scheme you belong to. ## Frequently asked questions ### Which commercial solar installers in the Midlands have done 500kW rooftop systems? An assistant can only answer with firms that publish projects of that size in that region. A project list giving site type, location, system size and year is what lets it name you. Systems that large fall outside MCS's solar standard, so buyers look instead at your electrical accreditation, insurance and comparable projects. ### Is the Workplace Charging Scheme still open in 2026? Yes, for now. GOV.UK says it has been extended for a final year until 31 March 2027, with grants of up to £500 per socket from 1 April 2026, and the installer must be OZEV-authorised. Put the date you last checked on any page that mentions it. ### Should an energy services firm publish its prices? Yes, as ranges with dates. Buyers ask assistants what commercial solar, heat pumps or EV charging cost, and an assistant can only quote figures it can find. Give typical ranges for the project sizes you do most, explain what moves the price and date the page. Say when a project does not make sense, too: that reads as expertise. ### How do we publish savings without overpromising? Separate forecasts from measured results and state the period. If a project underperformed, say why. Avoid "save up to" figures that cannot be traced to a project. ### How long before assistants start naming us? Assistants that search the web live, such as ChatGPT search, Copilot and Google's AI features, can reflect a rewritten page or a fixed listing within weeks. Answers drawn from a model's memory change only when it is retrained, which takes months. Being named reliably for broad questions takes several months of steady publishing, reviews and outside coverage. If you want a view on what to fix first, [book a free Growth Review](/growth-review). We will tell you plainly where you stand and what we would do next. ## Sources - [Forrester, B2B buyers make zero-click buying number one (January 2026)](https://www.forrester.com/blogs/b2b_buyers_make_zero_click_buying_number_one/) - [MCS, Standards and tools library: MIS 3002 and MIS 3005-D (December 2025 and March 2026)](https://mcscertified.com/standards-tools-library/) - [Ofgem, Third party intermediaries: what your microbusiness needs to know (October 2022)](https://www.ofgem.gov.uk/guidance/third-party-intermediaries-what-your-microbusiness-needs-know) - [Ofgem, Get a business energy contract (accessed September 2026)](https://www.ofgem.gov.uk/your-energy-supply/how-manage-your-energy-supply/get-business-energy-contract) - [GOV.UK, ESOS: find out if you qualify and how to comply (September 2026)](https://www.gov.uk/guidance/energy-savings-opportunity-scheme-esos) - [GOV.UK, Workplace Charging Scheme: guidance for installers (April 2026)](https://www.gov.uk/guidance/workplace-charging-scheme-guidance-for-installers) - [GOV.UK, Non-domestic private rented sector MEES: interim response (June 2026)](https://www.gov.uk/government/consultations/non-domestic-private-rented-sector-minimum-energy-efficiency-standards-epc-b-implementation/outcome/minimum-energy-efficiency-standards-mees-in-the-non-domestic-private-rented-sector-interim-response) - [G2 via PR Newswire, Half of B2B software buyers now start their research with AI chatbots (April 2026)](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html) --- Title: Why does ChatGPT recommend other insurance brokers and not us? URL: https://fortitudemedia.ai/insights/ai-optimisation-insurance-brokers Author: Ashley Kenerson Published: 2026-03-31 Last reviewed: 2026-09-24 Topic: Sector guides Short answer: ChatGPT recommends insurance brokers it can find clear evidence for: a stated specialism and plain, fair answers to the cover questions clients ask, including how you are paid and when going direct is fine. It weighs that against the FCA Register, BIBA, the Lloyd's Market Directory, trade press and reviews. Most brokers publish too little. Pick one or two niches, answer their real questions on your site and make every listing match. The operations director of a haulage firm opens her renewal and sees the premium has gone up again. Before she rings anyone, she types into ChatGPT: "who are the best fleet insurance brokers for a haulage company in the Midlands". She gets three names and a short paragraph on each. You place more fleet business than any of them. You are not on the list. That first conversation used to happen on the phone or through the client's accountant. Those routes still matter, but more early research now happens in ChatGPT, Copilot, Gemini and Google's AI answers. [Forrester found in January 2026](https://www.forrester.com/blogs/b2b_buyers_make_zero_click_buying_number_one/) that 94% of business buyers use AI in their buying process, and that business buyers are four times as likely as consumers to use Microsoft Copilot. By the time they call, they already have a shortlist and a view on the cover they think they need. You answer these questions every day for clients who trust you, yet most brokerages publish almost nothing that shows it, and assistants can only recommend what they can read. This guide explains what they read, why specialists get named more often, and how to write about cover in a way that helps clients, sits comfortably with the FCA and gives assistants something to quote. ## Key takeaways - Business owners now ask ChatGPT which broker to use and what cover they need before they ring anyone, so the shortlist forms without you. - AI assistants name brokers with a clear specialism, because "fleet insurance for hauliers" is a question they can match and "all your insurance needs" is not. - Plain, fair explanations of cover, including when a client might not need something, are what assistants quote and what FCA rules on clear, fair communication expect anyway. - Explaining how you are paid answers a question clients now ask assistants directly, and commercial clients are entitled to ask you for the commission figure. - Your details on the FCA Register, Companies House, BIBA, the Lloyd's Market Directory and review sites must match your website, or assistants treat you as less certain. - Coverage in Insurance Times, Insurance Age and your clients' own trade press counts for more than anything you say about yourself. ## What are business owners asking ChatGPT about insurance? They ask about their own situation, not about brokers, and the assistant then decides whether a broker belongs in the answer. The questions arrive at renewal, after a premium rise or after something has gone wrong, and they are specific: - "Do I need employers' liability insurance if my only staff are two part-timers?" - "How much professional indemnity insurance does a UK tech consultancy need?" - "Which brokers specialise in cyber insurance for SaaS companies in the UK?" - "Who are the best fleet insurance brokers for a haulage firm in the Midlands?" - "Do insurance brokers have to tell me how much commission they earn?" - "Is it cheaper to use a broker or buy business insurance direct?" - "What does directors' and officers' insurance cover for a private company?" Most are questions about cover: the assistant answers them and then, if it has evidence, suggests who could help. If your site answers the question well, you are the natural name to put next to it. The questions that do ask for a broker almost always name a niche and often a place. An assistant cannot match those words to a brokerage whose website promises "tailored solutions for businesses of all sizes". Our piece on [building content around the questions customers ask](/insights/content-around-customer-questions) sets out how to turn a list like this into pages. ## How does ChatGPT decide which brokers to name? It names the brokers it can describe with confidence, and that confidence comes from specific evidence that agrees across several sources. No single ranking decides it. ChatGPT has shown labelled ads to UK users on its Free and Go plans since 11 August 2026, but OpenAI says ads do not change the answer itself, so a sponsored slot beside the answer is a different thing from being named in it. Whether it draws on training or a live web search, it looks for four things: - **A stated specialism.** Pages that say plainly who you insure and for what, in the words clients use. - **Evidence you know the subject.** Explanations of cover, exclusions and claims for that niche that go further than a brochure. - **Named people.** Directors and account executives with profiles, qualifications and sector experience. - **Agreement from elsewhere.** The FCA Register, Companies House, BIBA, review sites, trade press and your clients' trade bodies saying the same thing about you. The last one catches out many brokers. Acquisitions, trading names, appointed representative arrangements and old addresses leave a trail of slightly different descriptions, and an assistant that sees conflict hedges. We saw the same pattern with [TCC Group](/work/tcc-group), a regulatory compliance group with three brands: "ChatGPT already placed TCC top for outsourced suitability reviews. But on Consumer Duty and Section 166, the two areas the group backs hardest, the assistants named other firms and not TCC. One had stopped recognising the UK firm altogether, and none of them connected the three brands." Our guide on [why ChatGPT describes firms wrongly](/insights/ai-conflicting-information-online) explains how to find and fix this. The other catch is volume without depth. A hundred thin product pages, each three paragraphs long and interchangeable with every other broker's, give an assistant nothing worth quoting. Ten pages that each fully answer one type of client's situation will do far more. ## Why do specialist brokers get named and generalists do not? Assistants recommend specialists because buyers ask specialist questions. A brokerage clearly known for one or two niches will be named in those niches far more often than a generalist of the same size is named anywhere. This is not an argument for turning business away. Most good brokerages already have two or three areas where they place a large share of their book. The job is to make that visible. For example, a regional broker that writes most of its business for manufacturers and logistics firms, with a growing book of care providers, but whose website leads with "commercial insurance for businesses across the North West", looks like a generalist. Rebuilt around three labelled specialisms, each with its own section, named team and client questions answered, the same firm becomes a candidate every time someone asks about insuring a manufacturer, a haulier or a care home in the region. If you really are a generalist, organise by client type rather than by product. "Insurance for professional services firms" gives an assistant something to match. "Our products" does not. ## A worked specialism: insurance for tech and SaaS firms Technology is a niche where the questions are specific and generalist pages fail. Founders of SaaS firms and tech consultancies ask about three covers in particular: | Cover | The question the founder types | What a strong page explains | |---|---|---| | Technology professional indemnity | "How much PI does a UK tech consultancy need?" | That the limit is usually driven by client contracts and project size, how to read a liability clause, and what "each and every claim" means | | Cyber | "Which brokers specialise in cyber insurance for SaaS companies?" | What cyber cover responds to, how it differs from tech PI, and what insurers ask about security before quoting | | Directors' and officers' | "Do we need D&O before we raise investment?" | Why investors often ask for it, what it covers for a private company, and what it excludes | The PI answer shows why depth wins. There is no single right figure for a tech consultancy, and a page that says so, then explains how contracts and the type of work set the limit, is far more useful than a quote form, and it is the answer an assistant can repeat. Name the account executives who handle technology clients, add anonymised examples of claims you have handled, and get listed where tech buyers look: BIBA's Find Insurance service has a dedicated cyber insurance directory, and your clients' trade press, not only the insurance titles, is where an assistant looks for experts in the niche. Our guide for [B2B technology companies](/insights/ai-optimisation-technology-companies) shows what those clients ask from their side of the desk. ## How to write about cover without sounding like a sales page Write the way you would explain cover to a client across the desk: plainly, fairly and including the parts that do not help you sell. That is what assistants trust, and it matches the FCA's rules. [ICOBS 2.2.2R](https://www.handbook.fca.org.uk/handbook/ICOBS/2/2.html) says that when a firm communicates information, including a financial promotion, it must be clear, fair and not misleading, and Principle 7 says the same. Where retail customers are involved, [the Consumer Duty applies](https://www.handbook.fca.org.uk/handbook/PRIN/2A/1.html) in place of Principles 6 and 7 and sets a higher standard. Financial advisers face the same tension, and our [guide for financial advisory firms](/insights/ai-optimisation-financial-advisory) covers how they handle it. ### Say plainly how you are paid Explain on your site whether you are paid by commission from insurers, a fee or both. You already owe clients much of this: under [ICOBS 4.3](https://www.handbook.fca.org.uk/handbook/ICOBS/4/3.html), an intermediary must tell a customer the nature of its remuneration before the contract is concluded, and the amount of any fee. Under [ICOBS 4.4](https://www.handbook.fca.org.uk/handbook/ICOBS/4/4.html), a commercial customer can ask for the commission you and any associate receive, and you must disclose it promptly, in cash terms and in writing. FCA guidance says that includes profit shares, volume-based payments and payments from premium finance companies. A clear page costs nothing and answers "do insurance brokers get paid by the insurer?" before the client asks an assistant. Premium finance is on the FCA's agenda too. Its [premium finance market study](https://www.fca.org.uk/publications/market-studies/ms24-2-premium-finance), which reported in February 2026, covered motor and home insurance customers and found the cost of paying monthly had fallen since the Consumer Duty came in. If you arrange credit for commercial clients, explain the cost and the alternative of paying annually. ### Compare fairly, including going direct Is it cheaper to use a broker or buy direct? The honest answer depends on the risk, and saying so makes you quotable. A broker that writes "for a one-person consultancy wanting basic professional indemnity, going direct is often fine; once you have staff and contracts with liability caps, advice starts to pay for itself" is far more believable than one that claims a broker is always better. ### Explain claims, not only policies Pages that walk through what happens after a flood, a cyber incident or an injury on site, anonymised and with permission, show the part of your job comparison sites cannot do. ### Keep it current Put a "last reviewed" date on each guide and revisit anything about pricing, capacity or regulation at least once a year. ## Which outside sources matter most for a broker? Assistants trust what others say about you more than what you say about yourself. In Muck Rack's analysis of more than 25 million links cited by AI tools, [earned media made up 84% of citations](https://www.globenewswire.com/news-release/2026/05/07/3290268/0/en/generative-pulse-earned-media-consistently-drives-ai-citations-holding-at-84.html) and paid content 0.3% (May 2026). For a UK broker, check these first: | Source | What it records | What to check | |---|---|---| | FCA Register | Firm name, trading names, address, permissions, appointed representative status | Matches your website exactly, including every trading name | | Companies House | Registered name and directors | Easy to connect to the name clients know | | [BIBA Find Insurance](https://www.biba.org.uk/find-insurance/) | A phone service and online search by type of insurance, with specialist directories such as cyber | Your membership is current and your specialisms are recorded | | [Lloyd's Market Directory](https://www.lloyds.com/findabroker) | Registered brokers, searchable by class and sub-class of business | Your classes of business reflect your real specialisms | | Chartered Insurance Institute | Chartered Insurance Broker status and individual qualifications | Named on your team pages as well | | Insurance trade press | Comment and coverage in Insurance Times, Insurance Age and Post | Regular comment on your niche, and niche-specific awards | | Your clients' trade press and bodies | Articles and talks in care, haulage or tech titles and events | For a specialist, often worth more than insurance press | | Reviews | Google, Trustpilot or Feefo | Reviews that mention the niche and the outcome | Reviews matter to buyers inside AI answers too: in [G2's April 2026 survey of 1,076 software buyers](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html), 45% said citations of review sites were the most confidence-inspiring signal in an AI answer. One relevant article or comment a month for a year does more than a flurry in the spring. ## What should a broker's specialism page look like? Each specialism needs a small cluster: one main page saying who you insure and how, and several pages that each answer one client question in full. A good main page for, say, insurance for manufacturers would: - Open with two sentences that say who it is for and what you do for them. - Name the covers that matter and explain each briefly, including common gaps, such as whether business interruption responds when a key supplier's factory burns down rather than your client's own. - Introduce the people who handle that book, with qualifications and years in the sector. - Include two or three anonymised examples of problems you have solved. - Answer six to eight common questions in plain prose, and link to the deeper question pages. Structured data, the labels in the page code that tell search engines this is an insurance broker, where it is and who works there, helps them read your facts correctly. Google says its AI features need no special markup, so keep it accurate rather than treating it as a ranking trick. Our guide to [schema markup for AI assistants](/insights/structured-data-schema-markup-ai) explains what to include. ## How long does it take, and how do you tell it is working? Fixes to listings and inconsistencies can show up within weeks in assistants that search the web live. Answers drawn from a model's memory change only when it is retrained, which takes months, and becoming the name an assistant reaches for in a niche depends on other people writing about you. Write down a fixed list of 20 to 50 questions your best clients would ask, including some that ask for a broker by niche and region. Run each several times a month in ChatGPT, Copilot, Gemini and Google in a clean session, because answers vary from run to run, and record whether you are named, who is named instead and how you are described. For [ID-Pal](/work/id-pal), the method was: "The same buyer questions run every month in ChatGPT, Gemini and Claude, in a fresh chat. A win only counts when it repeats, and losses are reported too." A spreadsheet is enough to start, and enquiries that mention an assistant are the commercial signal. ## What to do this month 1. **Make your details match.** Check your FCA Register entry, Companies House record, BIBA and Lloyd's listings, Google Business Profile and website agree on your name, trading names, address and what you do. 2. **Choose your niches.** Pick the one or two client types where you have the deepest book and the best stories, and agree internally that these lead the website. 3. **Publish how you are paid.** One plain page on commission, fees and how a commercial client can ask for the figures. 4. **Write the first three question pages.** Take three questions your account executives answer every week in that niche and publish full answers with a named author and a review date. 5. **Ask for niche-specific reviews.** After the next few renewals or settled claims in that niche, ask the client for a short review that says what you did. ## Frequently asked questions ### Do insurance brokers have to tell me how much commission they earn? For a commercial customer, yes, if you ask. ICOBS 4.4 says an intermediary must, on a commercial customer's request, promptly disclose the commission it and any associate receive, in cash terms and in writing. Every customer must also be told the nature of the broker's remuneration before the contract is concluded, and the amount of any fee. ### How do we compete with comparison sites and direct insurers? By answering the questions they cannot. Comparison sites handle simple, standard risks well and struggle with anything that needs judgement: contract wording, unusual exposures, a claim that goes wrong. When an assistant answers a complicated question, it needs a specialist to point to. ### Should our account executives publish under their own names? Yes. Assistants and clients both trust named people with visible experience more than an anonymous brand. A short profile for each executive, with qualifications and the sectors they handle, plus a few articles in their niche, ties that expertise to the firm. ### Does publishing this kind of content cause problems with the FCA? It should not. Educational content that is clear, fair and balanced sits comfortably with ICOBS 2.2. Where a page makes product or price claims, or could be read as a promotion, run it past your compliance lead. ### We are part of a larger group. Does that help or hurt? It can do both. A group name brings standing, but it confuses assistants if your trading name, the group brand and your appointed representative status are described differently in different places. Decide how you want to be described, say it clearly on your own site, and make the group site and your listings match. The broker who gets named is rarely the biggest. It is the one whose specialism is easiest to see and easiest to check. If you want to know how you compare with other brokers in your niche today, [book a free Growth Review](/growth-review): 30 minutes with David Adams, and three fixes in writing afterwards. ## Sources - [Forrester, B2B buyers make zero-click buying number one (January 2026)](https://www.forrester.com/blogs/b2b_buyers_make_zero_click_buying_number_one/) - [FCA Handbook, ICOBS 4.4 Commission disclosure for commercial customers (accessed September 2026)](https://www.handbook.fca.org.uk/handbook/ICOBS/4/4.html) - [FCA Handbook, ICOBS 4.3 Remuneration disclosure (accessed September 2026)](https://www.handbook.fca.org.uk/handbook/ICOBS/4/3.html) - [FCA, MS24/2 Premium finance market study, final report (February 2026)](https://www.fca.org.uk/publications/market-studies/ms24-2-premium-finance) - [BIBA, Find Insurance (accessed September 2026)](https://www.biba.org.uk/find-insurance/) - [Lloyd's, Market Directory (accessed September 2026)](https://www.lloyds.com/findabroker) - [Muck Rack via GlobeNewswire, Earned media consistently drives AI citations (May 2026)](https://www.globenewswire.com/news-release/2026/05/07/3290268/0/en/generative-pulse-earned-media-consistently-drives-ai-citations-holding-at-84.html) - [G2 via PR Newswire, Half of B2B software buyers now start their research with AI chatbots (April 2026)](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html) --- Title: Why does ChatGPT sound unsure about our firm, and how do we fix it? URL: https://fortitudemedia.ai/insights/ai-cross-references-business-sources Author: Ashley Kenerson Published: 2026-03-31 Last reviewed: 2026-09-24 Topic: Press, reviews and being trusted Short answer: Because it has found too little independent evidence that agrees with what you say. ChatGPT, Gemini, Claude and Copilot compare your website with LinkedIn, Companies House, business listings such as Bing Places, Wikidata, directories and press coverage. When a claim appears only on your own site, or sources disagree, they attribute it ("according to its website") or leave you out. Fix it by using one description everywhere you control, completing your listings and earning independent coverage that repeats the same facts. Ask ChatGPT about your firm and read the tone of the answer. For some firms it is crisp, for example: "Example Partners is an employment law firm in Leeds that advises hospitality and retail groups." For others it hedges: "Example Ltd appears to offer consultancy services, according to its website." Sometimes it says it has little information, or names you last in a list with a description that could fit anyone. That tone is not an accident. AI assistants such as ChatGPT, Gemini, Claude, Copilot and Perplexity state things plainly when several independent sources agree, and hedge when the main evidence is your own website. Independent evidence is most of what they cite: in [Muck Rack's May 2026 analysis](https://www.globenewswire.com/news-release/2026/05/07/3290268/0/en/generative-pulse-earned-media-consistently-drives-ai-citations-holding-at-84.html) of more than 25 million links cited by ChatGPT, Claude and Gemini, earned media made up 84% and paid content 0.3%. Buyers notice the hedge. In the [TrustRadius 2026 B2B Buying Disconnect Report](https://www.prnewswire.com/news-releases/trustradius-2026-b2b-buying-disconnect-report-reveals-ai-has-changed-how-buyers-research-but-not-what-they-trust-302825792.html), 63% of technology buyers had used AI in a purchase, and 94% of those checked what it told them at least some of the time. A buyer who reads "appears to" goes looking for a firm the assistant is sure about. If the problem is facts that are plainly wrong rather than vague, start with [how to fix wrong facts ChatGPT repeats about your firm](/insights/ai-conflicting-information-online). ## Key takeaways - AI assistants publish no confidence score, but you can read their confidence in how plainly or vaguely they describe your firm. - Hedges such as "appears to" and "according to its website" usually mean a claim appears only on your own site, or that sources disagree. - Assistants compare your website with LinkedIn, Companies House, business listings, databases such as Wikidata, directories and press coverage. - Copilot draws on Bing, so a stale Bing Places listing can make one assistant sure of you and another vague. - A five-step test, repeated monthly, shows where the hedges are and which sources the assistants use instead of yours. - The fix is one consistent description everywhere you control, complete listings, and independent coverage that repeats the same facts. ## What does it mean when ChatGPT hedges about our firm? It means the assistant has not found enough independent evidence to state your claims as fact. No assistant shows a confidence number, but the signs are consistent: - phrases such as "appears to", "seems to", "according to its website", "describes itself as" or "claims to specialise in" - a generic description that could fit any firm in your sector - out-of-date details, such as an old office or a service you stopped offering - your firm missing from shortlists where competitors appear, or named last with less detail - an admission that it has limited information about you Each tells you the same thing. Confident answers state facts; hedged answers attribute them. If a claim appears on your website and nowhere else, the assistant tends to credit it to you. If your site, a trade article and your professional body's register all say it, the assistant can say it plainly. For a buyer, the difference between "a specialist in X" and "appears to offer services" is the difference between a firm worth calling and one worth skipping. ## Why do assistants hedge about some firms and not others? Because the evidence about some firms is thin, comes only from them, or disagrees. That is how a careful buyer behaves too. Evidence reaches an assistant by two routes. The first is training: the model has read a large snapshot of the public web, and a firm described the same way in many places leaves a firmer impression than one that appears mainly on its own site. The second is live search. Gemini, Google's AI Overviews and AI Mode start from Google's index; Copilot starts from Bing's; and ChatGPT search has drawn on Bing, with 87% of its citations matching Bing's top results in a small [Seer Interactive test in February 2025](https://www.seerinteractive.com/insights/87-percent-of-searchgpt-citations-match-bings-top-results). So the index matters: a firm that is well described in Google but patchy in Bing can get a confident answer from Gemini and a vague one from Copilot. On both routes, your website is necessary but not enough. It is where the assistant finds the facts; other sources are where it learns whether anyone else agrees. Independent sources carry weight with buyers as well: in [G2's April 2026 survey of 1,076 software buyers](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html), 45% said citations of review sites were the most confidence-inspiring signal in an AI answer. Why earned coverage counts for more than your own pages is set out in [whether ChatGPT trusts press coverage more than your website](/insights/earned-owned-paid-authority-signals). ## Which sources do AI assistants compare? Almost anything public that names your firm. For a UK B2B firm, these matter most: | Source | Whose voice it is | Why it matters for hedging | |---|---|---| | Your website (home, about, service, team and contact pages) | Yours | Where the assistant finds your claims; it must be the fullest, most current account | | LinkedIn company page and founders' profiles | Yours | Often read alongside the website; a mismatch is the most common conflict | | Companies House | Official record | Registered name, office, directors and nature of business, copied by many other sites | | Google Business Profile, Bing Places, Apple Business Connect | Yours, on someone else's platform | The listings behind Google, Bing and Apple Maps; Bing feeds Copilot | | Business data sites (Endole, Dun & Bradstreet, Crunchbase) | Third party | Often built from filings or contributors, and rarely checked by the firm | | Wikidata | Open database anyone can edit | Structured facts, free for anyone to reuse | | Trade directories and professional body registers | Third party | Confirms your category and regulated status | | Press and trade coverage, client case studies, reviews, speaker pages | Independent | The corroboration that lets an assistant drop the hedge | Most firms of your size do not have, and should not try to create, a Wikipedia article. Wikipedia has strict rules on notability and on people writing about their own organisations, and a company-written article is likely to be challenged or removed. Wikidata is different. It is the open database of facts behind Wikipedia, anyone can edit it, and its data is released into the public domain, so search engines and AI developers are free to reuse it. A firm does not need a Wikipedia article to have a Wikidata item, but the item needs references to public sources, and other editors will check it. Our guide to [fixing wrong facts](/insights/ai-conflicting-information-online) lists who can edit each of these sources and the order to update them in. ## How do assistants settle a disagreement? They weigh sources rather than count them. None publishes its rules, so what follows is what we observe in the answers we record, not a documented method. **Independent sources often beat your own.** If your website says one thing and a respected trade publication says another, assistants often go with the publication, since you have a reason to flatter yourself and it does not. That cuts both ways: an old article that describes you wrongly can outweigh your accurate website. **Newer beats older when it explains the change.** A source that says "Example Ltd, which moved its head office to Manchester last year" lets an assistant reconcile old and new. Without one, it sees a contradiction and hedges. **Specific beats vague.** "Founded in 2009 by two former tax managers, with offices in Leeds and York" gets repeated. "An established practice serving Yorkshire" gets softened or ignored. ## What does this look like for a real firm? [TCC Group](/work/tcc-group) provides regulatory compliance for UK financial services through three brands: TCC for advisory and managed services, Recordsure for compliance AI and Momenta for specialist resourcing. Its case study puts the problem in one line: "Three brands, one group. AI could only see the pieces." "ChatGPT already placed TCC top for outsourced suitability reviews. But on Consumer Duty and Section 166, the two areas the group backs hardest, the assistants named other firms and not TCC. One had stopped recognising the UK firm altogether, and none of them connected the three brands." The evidence came from "36 recorded answers from ChatGPT, Gemini and Claude, every conversation kept and reproducible." And the cause was structural. "It was not a content problem." "What failed was underneath: a site organised around internal teams rather than the questions buyers ask, pages that half-answered the same question so neither was cited, and machine-readable data that declared a name, a logo and nothing else." That is the name-consistency problem at group scale. The plan answered it with "Ten solution pages, one for each problem a buyer types, from Section 166 and FCA intervention to vulnerable customers, plus sector hubs and one group page telling the story once." "TCC chose to build it with its own team, and the new tcc.group is live, built to the plan, with nothing about it depending on Fortitude." If your firm has merged, rebranded or runs several brands, the lesson carries over: publish one page that names every brand and how they relate, declare the same relationship in your structured data, and make each brand's profiles point to it. ## Why does ChatGPT give the wrong number of employees for my company? Because headcount is one of the facts most often left out of date. In our experience assistants piece it together from your LinkedIn company page, business data sites, old press coverage and your website, and those sources rarely agree. State a headcount range on your about page with the month, for example "about 90 people, September 2026", set the company size band on your LinkedIn page to match, and check what the business data sites show. Then the assistant has one current, dated figure to repeat instead of three old ones to reconcile. ## Does Companies House data affect what AI assistants say about my business? Yes, mostly indirectly. Companies House holds your registered name, registered office, directors and nature of business, and business data sites copy it, so an old registered office or a departed director can echo across dozens of pages that assistants read. Keep your filings current, and if your trading name differs from your registered name, say so on your website so an assistant can connect the two. A departed director still listed on data sites is a common source of confusion about who leads the firm. ## How do I update my business details in Bing so Copilot gets them right? Claim your listing in Bing Places for Business and match it to your fact sheet, then check that Bing has indexed your key pages, using Bing Webmaster Tools if it has not. Copilot draws on Bing, and ChatGPT search has too. This matters more than its share of web traffic suggests, because Copilot is what many buyers have at work. Microsoft 365 Copilot passed 30 million paid seats in July 2026, according to [Microsoft's results](https://www.microsoft.com/en-us/investor/earnings/fy-2026-q4/press-release-webcast), and [Forrester found in January 2026](https://www.forrester.com/blogs/b2b_buyers_make_zero_click_buying_number_one/) that business buyers are four times as likely as consumers to use Copilot. Check Apple Business Connect while you are there, since Apple Maps draws on it. ## How to tell whether your firm has a confidence problem You can run this test in under an hour, and it is worth repeating every month. ### Step 1: Ask the assistants about your firm directly Ask ChatGPT, Gemini, Claude, Copilot and Perplexity what your firm does, who its typical clients are and whether it is a good choice for your main service. Use a fresh chat each time and save each answer with the date. ### Step 2: Ask the question your buyers ask Ask the category question, such as "Which firms should I consider for this service in this region?", and note whether you appear, where, and how you are described compared with competitors. ### Step 3: Mark the hedges and errors Highlight every "appears to", every "according to its website", every generic description, wrong fact and omission. These are your confidence gaps. ### Step 4: Look at the sources the assistant used Perplexity, Copilot and ChatGPT when it searches show the pages they drew on. Note which sources describe your competitors and whether any mention you. That is the start of your target list. ### Step 5: Record it and repeat Keep a fixed list of 20 to 50 buyer questions in a spreadsheet and ask each several times a month in a clean session, because answers vary from run to run. Count a change only when it repeats. Our guide to [finding out whether ChatGPT recommends your firm](/insights/track-ai-recommending-your-business) sets this up in more detail. ## How do we fix a confidence problem? Make the sources agree, then give the assistant independent evidence that repeats what they say. **Settle one description and use it everywhere.** One plain sentence saying what you do, who for and where, used on your website, LinkedIn, listings, directory entries, the boilerplate you give journalists and your speaker biographies. The independent coverage you earn will then tend to repeat it. **Complete the listings and databases.** Google Business Profile, Bing Places, Apple Business Connect, Companies House, the business data sites and, where you can reference it, Wikidata. **Label the facts on your site.** Structured data states your name, former names, address and official profiles, and for a group how its brands relate, in a form software reads directly. It helps search engines read your facts correctly; Google says its AI features need no special markup, and OpenAI and Anthropic have not said they read it, so keep it accurate and do not treat it as a ranking trick. Our guide to [schema markup for AI assistants](/insights/structured-data-schema-markup-ai) explains what to include. **Earn independent evidence.** Most founder-led firms do it through a few repeatable routes: expert comment for the trade titles buyers read, original data about your market, guest articles an editor accepts on merit, case studies on clients' own sites, and roles with professional bodies. [Which websites that mention you count with AI](/insights/ai-evaluates-quality-referring-domains) and [digital or traditional PR for being named by ChatGPT](/insights/digital-pr-vs-traditional-pr) cover the trade-offs. Avoid anything that promises links in bulk: it adds to the count without adding to the evidence. ## What to do this month 1. Run the five-step test above on your firm and your main category question, and save every answer with the date. 2. Write one plain sentence saying what you do, who for and where, and update your website, LinkedIn page and listings to match. 3. Claim or correct your Bing Places, Google Business Profile and Apple Business Connect listings, and check your Companies House details. 4. Ask the publishers of the three most visible out-of-date descriptions of your firm to correct them. 5. Choose one route to independent coverage, such as expert comment for your main trade title, and commit to it for three months. ## Frequently asked questions ### How do I change what ChatGPT says about my company? Change the sources it reads. Telling ChatGPT it is wrong in your own chat may change that conversation, but not what other people see. Make your website, listings and profiles agree, correct third-party sources, and earn independent coverage that states the same facts. You can also flag a wrong answer with the thumbs-down button. ### Should my company have a Wikidata entry, and how do I create one? Often yes, if you can support it with public references such as Companies House and independent coverage. Create an account, search to check no item exists, then add your firm with a neutral description, what kind of organisation it is, country, founding date, official website and headquarters, each backed by a reference. Write it like a fact sheet, not a brochure, because other editors will check it. ### Do mentions without a link help at all? Yes, in our observation. Assistants read text, so a trade article that names your firm and describes what you do adds to the evidence even without a link. A link adds a little more, because it helps search engines connect the mention to your site. ### Can a critical article make assistants less likely to recommend us? It can colour the answer, especially if little else has been written about you. Fix the underlying issue if the criticism is fair, ask for a correction if it is not, and build enough balanced coverage that one article is not the main thing an assistant finds. ### We rebranded. How do we stop ChatGPT using our old name? You may not stop it entirely, and you do not need to. Aim for assistants to connect the two names: say plainly on your website that you were formerly known by the old name, list it as an alternative name in your structured data, and update every profile. A trade press announcement gives assistants an independent source that links both names. ### How long before new coverage changes what assistants say? Assistants that search the web live (ChatGPT search, Perplexity, Copilot, Google's AI features) can reflect a new article within weeks. What a model knows from its own memory changes only when it is retrained, which takes months. In our experience, steady coverage over several months shifts answers more reliably than one big article. If the answers you get back are vague or missing, [Reputation That Wins](/reputation-that-wins) is built to change them, starting with the sources the assistants already read about your competitors. To see where you stand first, [book a free Growth Review](/growth-review): 30 minutes with David Adams and three fixes in writing. ## Sources - [Muck Rack, Generative Pulse: earned media consistently drives AI citations (May 2026)](https://www.globenewswire.com/news-release/2026/05/07/3290268/0/en/generative-pulse-earned-media-consistently-drives-ai-citations-holding-at-84.html) - [TrustRadius, 2026 B2B Buying Disconnect Report (July 2026)](https://www.prnewswire.com/news-releases/trustradius-2026-b2b-buying-disconnect-report-reveals-ai-has-changed-how-buyers-research-but-not-what-they-trust-302825792.html) - [G2, New G2 research: half of B2B software buyers now start their research with AI chatbots (April 2026)](https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html) - [Seer Interactive, 87% of SearchGPT citations match Bing's top results (February 2025)](https://www.seerinteractive.com/insights/87-percent-of-searchgpt-citations-match-bings-top-results) - [Forrester, B2B buyers make zero-click buying number one (January 2026)](https://www.forrester.com/blogs/b2b_buyers_make_zero_click_buying_number_one/) - [Microsoft, FY26 fourth quarter earnings press release (July 2026)](https://www.microsoft.com/en-us/investor/earnings/fy-2026-q4/press-release-webcast) --- Title: Why is our website traffic falling when our rankings have not? URL: https://fortitudemedia.ai/insights/zero-click-search-your-business Author: Ashley Kenerson Published: 2026-03-31 Last reviewed: 2026-09-24 Topic: How buying is changing Short answer: Your traffic is falling because more Google searches now end on the results page: AI Overviews, AI Mode, featured snippets and direct answers give the searcher what they need without a click. In a Pew Research Center study, people clicked a traditional result on 8% of visits with an AI summary, against 15% without. Explanation and comparison searches are hit hardest. Aim to be the source Google cites and names, stay in AI Overviews, and measure mentions and enquiries rather than visits alone. The monthly report has a line that does not make sense. Rankings are steady and organic visits are down again. Your marketing manager has checked the tracking twice. Nothing is broken. The people who used to click are not clicking, because Google answered their question before they reached your link. This is what people mean by zero-click search: a search that ends on the results page, satisfied by an AI Overview, AI Mode, a featured snippet (the short extract Google lifts from a page and shows in a box at the top) or a direct answer, with no visit to anyone's website. [Ahrefs found in February 2026](https://ahrefs.com/blog/ai-overviews-reduce-clicks-update) that AI Overviews cut click-through to the top-ranked result by 58%, comparing December 2025 with December 2023 across 300,000 keywords. For a founder, the question is what this does to a business that has treated visits as the measure of marketing, and what to measure instead. This guide covers the mechanisms, the searches most affected, how to see the effect in Search Console and whether to opt out of AI Overviews. ## Key takeaways - More Google searches now end on the results page, answered by AI Overviews, AI Mode, snippets or direct answers, so steady rankings deliver fewer visits. - In a US study by Pew Research Center, people clicked a traditional result on 8% of visits when an AI summary appeared, against 15% when it did not. - Explanation and comparison searches lose the most clicks, and they are the searches most B2B content programmes were built to win. - Being cited in an AI Overview brings more clicks than not being cited, and puts your name in front of the buyer even when nobody clicks. - Search Console now shows impressions in AI features separately; clicks from them sit inside your normal totals. - Most B2B firms should stay in AI Overviews rather than use the new opt-out, because being named is the point. - Measure whether you are named in AI answers, branded visits and enquiries by source, with organic visits as context. ## What is actually happening on the results page? There are now four mechanisms, and they stack: - Featured snippets lift a short passage from one page and show it in a box above the results. - Direct answers handle facts and figures, such as a definition, a date or a calculation. - AI Overviews, [in the UK since August 2024](https://blog.google/products-and-platforms/products/search/new-ways-to-connect-to-the-web-with-ai-overviews/), write a summary in Google's own voice at the top of the results, combining several pages, with links to the sources. - AI Mode, [live in the UK since 28 July 2025](https://blog.google/around-the-globe/google-europe/united-kingdom/ai-mode-search-uk/), is a separate conversational search where the buyer asks follow-up questions and Google runs several related searches for each one. Unlike an Overview, it gives an AI answer in place of the usual page of links. The clicks follow. [Pew Research Center tracked the searches of 900 US adults](https://www.pewresearch.org/short-reads/2025/07/22/google-users-are-less-likely-to-click-on-links-when-an-ai-summary-appears-in-the-results/) in March 2025 and found that people clicked a traditional result on 8% of visits when an AI summary appeared, against 15% when it did not. Only 1% clicked a link inside the summary itself. And 60% of searches starting with a question word got a summary. All four turn a search that would have produced a click into one that produces, at best, a citation. The wider rebuilding of the page is covered in [what Google AI Mode is and how it differs from AI Overviews](/insights/end-of-ten-blue-links), and the Overviews themselves in [our guide to Google AI Overviews](/insights/google-ai-overviews-changing-search). ## How much traffic do AI Overviews take from websites? On the searches where they appear, a lot: Ahrefs measured a 58% fall in click-through to the top result. Both Ahrefs and Pew studied American searches of every kind, so your own figure will differ; the reliable number is the one in your Search Console. The effect is uneven. Semrush data [reported by Search Engine Land in December 2025](https://searchengineland.com/google-ai-overviews-surge-pullback-data-466314) found AI Overviews on 6.5% of queries in January 2025, just under 25% at the July peak and under 16% by November. The share of AI Overview queries that were commercial rose from 8% to 18% between January and October 2025, so buying searches are no longer untouched. ## Which searches lose the most clicks? Searches that ask for an explanation or a comparison lose the most, fact searches lose many, and searches with clear intent to hire lose the fewest, though that last group is starting to change. | Type of search | Example | What now answers it | Effect on clicks | |---|---|---|---| | Informational | "what is", "how does", "best practice for" | AI Overviews, AI Mode, snippets | Largest fall | | Comparison | "X versus Y", "best options for" | AI Overviews and AI Mode write the comparison | Large fall; clicks come later, when the buyer is ready | | Fact | A definition, a date, a simple figure | Direct answers and snippets | Large fall for simple facts; detailed B2B specifications and pricing still need a page | | Buying | "supplier for", "quote for", a named service in a named place | The results, the map and business profiles, with AI features appearing more often | Smallest fall, for now | Most B2B content programmes of the last decade were built for the first two groups, and that is the traffic shrinking fastest. ## Why is my click-through rate falling in Search Console when rankings hold? Because your position is measured in a list fewer people scroll to, while the searcher's attention now sits above it. There is a counting effect too. Google's [Search Console help](https://support.google.com/webmasters/answer/7042828?hl=en) says an AI Overview occupies a single position in the results, and every link inside it is given that same position; clicks on those links count as clicks, and standard impression rules apply. So a page cited in an Overview can gain impressions at a high position without gaining clicks, which pushes the click-through rate down while the average position looks healthy. So the symptoms look contradictory: position stable, click-through rate down, visits down. It is not a fault or a penalty. The same logic applies to the assistants outside Google, which we set out in [why ranking first on Google does not get you mentioned](/insights/google-rankings-wont-protect-you). ## Is it zero-click, or something else? Rule out the other causes before you blame the results page. A zero-click fall shows up as steady positions and falling click-through on question-style searches. A fall with dropping positions, or a sudden fall across every page on one date, usually has another cause: a site migration, a redirect problem, a change of name or a technical fault. ID-Pal, an identity verification firm we work with, is an example of the second kind. "Traffic to the pages that bring in enquiries had fallen by around 40 per cent since the NorthRow move, and nobody could say why." The cause was not the results page: "The redirect from northrow.com was technically correct, but nothing on either site told Google or the AI engines that NorthRow and ID-Pal were the same company." The detail is in [the ID-Pal case study](/work/id-pal). ## Does being cited in an AI Overview bring any clicks? Yes, fewer than a top ranking once did, but more than not being cited. [Seer Interactive's April 2026 study](https://www.seerinteractive.com/insights/aio-impact-on-google-ctr-2026-update) of 53 brands found that being cited inside an AI Overview brought about 120% more organic clicks per impression than not being cited. Per million impressions, it counted about 33,500 clicks on queries with no AI Overview, about 20,743 when the brand was cited and about 9,445 when it was not. The bigger value is the name. Someone asking "how do managed IT contracts for multi-site firms usually work" is early in a purchase. If the Overview cites you, and ChatGPT or Copilot names you when they ask who to talk to, you are on the shortlist without a visit. The trap is being a source without being a name. That thread runs through our [AI-first future topic](/insights/pillar/ai-first-future). ## Should we opt out of AI Overviews? For almost every B2B firm, no. Being cited and named is the position to aim for, and opting out removes you from the answer while your competitors stay in it. The option now exists. On 3 June 2026 the [Competition and Markets Authority required Google](https://www.gov.uk/government/news/cma-secures-fairer-deal-for-publishers-and-improves-google-search-services-in-uk) to let publishers opt out of their content being used to power AI features such as AI Overviews, and to attribute publisher content with clear links; Google has nine months to put all of it in place. Search Console already has a [Search generative AI control](https://support.google.com/webmasters/answer/16908024?hl=en) that excludes a site from AI Overviews, AI Mode and generative AI features in Discover. Inclusion is the default, exclusion takes effect within one to two days for most content, and Google says the control is not used as a ranking signal for the rest of Search. The control suits a publisher that sells the content itself, where a summary replaces a sale. For a firm that sells a service, the summary advertises your expertise to buyers who want the service, not the article. ## How do I see AI Overview traffic in Google Search Console? You can see impressions in AI features separately, but not yet clicks. Google [added generative AI performance reports](https://developers.google.com/search/blog/2026/06/gen-ai-performance-reports) to Search Console on 3 June 2026 and rolled them out to all websites by 31 August 2026. They show impressions, pages, countries, devices and dates for AI Overviews, AI Mode and Discover. The same data stays inside the main Performance report, so clicks from AI Overviews and AI Mode are counted in your normal organic totals. In Google Analytics 4, the default "AI Assistant" channel added in May 2026 covers ChatGPT, Gemini, DeepSeek, Copilot and Grok, but not Claude or Perplexity, and clicks from AI Overviews and AI Mode still count as Organic Search ([Nice Looking Data, May 2026](https://www.nicelookingdata.com/blog/ga4-ai-traffic-chatgpt-referrals)). ### Step 1: Compare click-through rate before and after August 2024 For your top twenty queries, compare click-through rate before and after AI Overviews reached the UK in August 2024, and mark the queries where position held and click-through rate fell. Search Console keeps [16 months of performance data](https://developers.google.com/search/blog/2018/01/introducing-new-search-console), so earlier months need an export or a report saved at the time; failing that, use your analytics. ### Step 2: Open the generative AI report Note which pages earn impressions in AI Overviews and AI Mode, then search those queries yourself to see whether you are cited and named. ### Step 3: Check your enquiries Falling visits with steady enquiries is a different problem from falling visits with falling enquiries. ## What should we change in how we produce content? Produce content to be cited and named: specific, evidenced, clearly structured, on questions narrow enough that you can be the definitive answer. ### From broad keywords to specific questions A general guide to "project management" gets summarised into nothing. A precise answer to "how should a fifty-person engineering consultancy run resource planning across three offices" can be the source an Overview cites, because there are few such answers. Cover the follow-up questions too. ### From volume to evidence Overviews draw on pages that say something checkable: original data, a real case with figures, a method you use. The structure of a piece that gets picked up is in [what an article that ChatGPT quotes looks like](/insights/anatomy-of-article-cited-by-ai). ### From landing pages to answer pages Put the answer near the top, under a heading that is the question, with short paragraphs and a named author. Structured data helps Google read the page's facts correctly; it is not a ranking trick. ### Protect the pages that still get clicks Service and location pages still receive visits from buyers who are ready. Make them fast, clear and specific, because they now carry more of the conversion load. ## What should we measure instead of visits? Measure whether you are named, branded visits and enquiries by source, and keep organic visits as context. A sensible monthly set: - Named or not: for a fixed list of 20 to 50 buyer questions, whether Google's AI features, ChatGPT, Copilot, Gemini, Claude and Perplexity name your firm, asked several times a month because answers vary. - Branded and direct visits, which rise when your name is seen in answers. - Enquiries by source, with "an AI assistant" and "a Google summary" as options on the form and in the first-call question. - Conversion rate on buying pages, because fewer, warmer visitors should convert at a higher rate. - Organic visits and rankings, kept but read as context. Asking works. Helm Club, a founder membership community we work with, tracks where new members come from: "1 in 5 new members this year came through an AI recommendation." How to judge a site on outcomes rather than traffic is in [how to measure the return from your website](/insights/measure-website-roi-b2b). ## What does this do to the economics of search? It lowers the value of a ranking on its own and raises the value of being cited and named. Effort spent making your firm the clear, evidenced answer keeps producing after the spend stops; the comparison is in [whether being named by ChatGPT is cheaper than Google Ads](/insights/economics-ai-visibility-vs-advertising). None of this means switching search off: keep the part that still produces buyers healthy. ## What to do this month 1. Using Search Console and any saved reports, compare click-through rate for your top twenty queries before and after August 2024, and mark the queries where position held and clicks fell. 2. Open the generative AI performance report and list the pages that earn impressions in AI Overviews and AI Mode. 3. Search those queries yourself, note what sits above the links and whether you are cited and named, then ask ChatGPT, Copilot, Gemini, Claude and Perplexity the same questions. 4. Choose three questions your firm should be the definitive answer to, and write or rewrite one page for each with the answer near the top and evidence beneath. 5. Add "an AI assistant" and "a Google summary" to your enquiry form's source options, and brief the team to ask on first calls. ## Frequently asked questions ### Is search engine optimisation finished? No. Winning broad keywords and counting visits is worth less every year, but the foundations remain: a fast site crawlers can read, clear pages and sensible internal links. Google's AI features start from its index, so ranking still helps you get cited. The goal changes to being the source that is cited and named. ### Should we stop publishing guides and explainers? No, but publish them for a different reason. A specific, evidenced guide is what an Overview cites and an assistant draws on when it names you, so it works even when nobody clicks. ### Does this affect paid search too? Less, for now. Google's ads in AI Overviews were [not yet listed as available in the UK](https://support.google.com/google-ads/answer/16297775?hl=en) in September 2026, though Google is [testing new ad formats in AI Mode](https://blog.google/products/ads-commerce/google-marketing-live-search-ads/). The searchers who click adverts are the same people whose questions the page now answers, so keep paid search for buying queries where it works and do not expect it to replace lost informational traffic. ### How do we explain a falling traffic line to the board? Show position alongside click-through rate for the same queries, so the board sees that the position held and the page changed. Then show the new measures: named or not, branded visits and enquiries by source. ### How quickly can we become a cited source? Ranking helps but is not a gate: [Ahrefs found in March 2026](https://ahrefs.com/blog/ai-overview-citations-top-10) that only 38% of AI Overview citations came from pages in the top ten, down from 76% in July 2025, because Google runs several related searches for one question. A clear page that answers a specific question can be cited within weeks of Google recrawling it. Assistants that search live, such as ChatGPT search and Perplexity, can also reflect a new page within weeks; answers drawn from a model's memory take months, until it is retrained. The falling line is the page changing shape above your link, and the firms that do well from here are the ones whose name is in the answer. To find out whether yours is, [book a free Growth Review](/growth-review): 30 minutes with David Adams and three fixes in writing. If your pages need rebuilding to be cited, that is what [Website That Sells](/website-that-sells) does. ## Sources - [Pew Research Center, Google users are less likely to click on links when an AI summary appears in the results (July 2025)](https://www.pewresearch.org/short-reads/2025/07/22/google-users-are-less-likely-to-click-on-links-when-an-ai-summary-appears-in-the-results/) - [Ahrefs, AI Overviews reduce clicks: 2026 update (February 2026)](https://ahrefs.com/blog/ai-overviews-reduce-clicks-update) - [Seer Interactive, AIO impact on Google CTR: 2026 update (April 2026)](https://www.seerinteractive.com/insights/aio-impact-on-google-ctr-2026-update) - [Ahrefs, AI Overview citations from top-ten pages (March 2026)](https://ahrefs.com/blog/ai-overview-citations-top-10) - [Search Engine Land, Google AI Overviews surge and pullback, Semrush data (December 2025)](https://searchengineland.com/google-ai-overviews-surge-pullback-data-466314) - [Google Search Central, Introducing Search generative AI performance reports in Search Console (June 2026)](https://developers.google.com/search/blog/2026/06/gen-ai-performance-reports) - [Competition and Markets Authority, CMA secures fairer deal for publishers and improves Google search services in UK (June 2026)](https://www.gov.uk/government/news/cma-secures-fairer-deal-for-publishers-and-improves-google-search-services-in-uk) - [Nice Looking Data, GA4 AI traffic and ChatGPT referrals (May 2026)](https://www.nicelookingdata.com/blog/ga4-ai-traffic-chatgpt-referrals) --- Title: Will a cheap AI search package get us recommended by ChatGPT? URL: https://fortitudemedia.ai/insights/cheap-ai-optimisation-doesnt-work Author: David Adams Published: 2026-03-31 Last reviewed: 2026-09-24 Topic: Costs, returns and the business case Short answer: Usually not. UK providers advertise AI search work from a few hundred pounds a month, and entry retainers run from £800 to £2,000 (MarGen, June 2026). After a provider's costs, a £500 to £750 fee buys roughly five to seventeen hours of work, enough for monitoring and some content but rarely the expert time and earned coverage that get a firm named. Use low-cost options for tracking and narrow fixes. The email arrives most weeks. "Get your business recommended by ChatGPT. AI search packages from a few hundred pounds a month." It sounds reasonable. You know buyers are asking AI assistants who to hire, and you know you are not being named. A few hundred pounds a month is less than the office coffee bill. What is the harm in trying? The offers are real and the prices are too. [Stagg Studios, a UK provider, put agency and freelancer prices](https://staggstudios.com/blog/how-much-does-ai-seo-cost-uk.html) for AI search work from about £300 a month in July 2026, and [MarGen's June 2026 pricing guide](https://www.margen.net/geo-agency-pricing-guide-2026-uk/) puts entry retainers at £800 to £2,000 a month. So a quote of £500 or £750 sits at the bottom of a real market, not outside it. The problem is rarely dishonesty. It is that the work which gets a firm recommended by ChatGPT, Google's AI features, Copilot, Gemini, Perplexity and Claude takes expert time, and a small monthly fee cannot pay for much of it. This guide shows what a low fee buys, what it leaves out, the red flags to watch for, and the jobs where a low-cost option is exactly the right choice. ## Key takeaways - UK providers advertise AI search work from a few hundred pounds a month, and entry retainers run from £800 to £2,000 a month (MarGen, June 2026). - Do the arithmetic on any fee: after the provider's overheads and margin, a £500 to £750 monthly fee buys roughly five to seventeen hours of work, depending on your assumptions. - Getting named depends on coverage and proof beyond your own website, which is the part low-cost packages usually leave out. - Automated "AI audit" PDFs, bulk directory submissions and paid "best of" list placements are red flags, not coverage. - Low-cost options are the right choice for narrow jobs: tracking what the assistants say, a listings clean-up, or a specific technical fix. - Nobody can guarantee what an AI assistant will say; judge any provider, cheap or expensive, by the hours and outputs it commits to. ## What does a £500 a month AI SEO package actually buy? A few hours of work a month. A fee has to cover the provider's sales, management, tools, overheads and profit before any of it pays for work on your account, and at the bottom of the market what is left buys very little. Example: assume half of a £750 fee goes on the provider's costs and margin, leaving £375 for delivery, and that the people doing the work cost the provider £50 an hour. That buys seven or eight hours a month. These are our round, illustrative assumptions rather than published figures, so the table below also shows a generous case where 70% of the fee reaches delivery at £30 an hour. | Monthly fee (example) | Hours a month, central case (50% to delivery at £50 an hour) | Hours a month, generous case (70% to delivery at £30 an hour) | |---|---|---| | £500 | About 5 | About 12 | | £750 | About 7 or 8 | About 17 | | £2,000 | About 20 | About 47 | | £4,000 | About 40 | About 93 | Seven or eight hours covers drafting two blog posts with an AI tool, running a site-checking tool and a templated report. Even seventeen hours rarely stretches to an hour interviewing your managing partner, a morning rewriting your services pages around the questions buyers ask, and the weeks of pitching it takes to land one good piece of trade press coverage. Those are the parts that get you recommended. The same arithmetic applies at every price, including ours. A higher fee is only better value if the extra hours go on those parts, so ask any provider to state the monthly hours and outputs in writing. Our own fees are on [what it costs](/what-it-costs). ## What should an AI SEO package include? It should include the three kinds of work that change who the assistants name, plus a way to measure it. Check any proposal for these: - **Time with your experts**, to capture the specific knowledge that makes content worth quoting. - **Pages rewritten around buyer questions**, in plain language, with named case studies and sector detail. - **A consistent description elsewhere**: directory listings, association profiles, LinkedIn and review platforms that match your website. - **Earned coverage**: trade press commentary, awards and speaking. [Muck Rack's May 2026 analysis](https://www.globenewswire.com/news-release/2026/05/07/3290268/0/en/generative-pulse-earned-media-consistently-drives-ai-citations-holding-at-84.html) of more than 25 million links cited by ChatGPT, Claude and Gemini found earned media made up 84% of them and paid content 0.3%. - **Measurement** of what the assistants say for a fixed set of buyer questions, and a link to your enquiries. For comparison, MarGen describes its entry band as "basic AI visibility monitoring and lightweight content optimisation", typically two to four content pieces a month, basic schema markup and monthly reporting. That is a fair description of what the low end of the market buys. It covers monitoring and some content, and leaves out the coverage. We explain the whole picture in [how to get ChatGPT to recommend your firm](/insights/what-is-llm-optimisation). ## What are the red flags in a cheap AI search package? The clearest red flags are outputs that look like progress but do not change who the assistants trust. Watch for these: - **Automated "AI audit" PDFs.** A tool-generated report full of scores, with no record of what the assistants actually said when asked your buyers' questions. A useful audit shows the answers, question by question. - **Bulk directory submissions.** Hundreds of listings on sites nobody reads. Google's spam policies list ["low-quality directory or bookmark site links"](https://developers.google.com/search/docs/essentials/spam-policies) as link spam. What helps is an accurate entry on the few directories and review sites your buyers actually use. - **Paid "best of" list placements sold as coverage.** A place on a list you paid for is advertising. [Google's guide to AI search](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide) tells site owners they can ignore chasing inauthentic mentions. When we reviewed UK "best agency" lists in September 2026, several were published by an agency that ranked itself first. - **AI-drafted posts at volume.** Google's spam policies describe "using generative AI tools or other similar tools to generate many pages without adding value for users" as scaled content abuse. - **Guaranteed places in ChatGPT's answers.** Nobody can guarantee what an assistant will say. A provider can guarantee effort, reporting and outcomes it controls. ## Why does volume content not get quoted? Because it says nothing an assistant cannot find somewhere else. An assistant has already read thousands of general articles on your topic; what it picks up is specific: a clear view from someone with experience, a worked example, original figures, a direct answer to a narrow question. Google's guide makes the same point: it says non-commodity content, with first-hand expertise, will influence AI search presence more than anything else. For example, a firm of chartered surveyors could publish "Ten things to know about commercial property surveys", which competes with every other firm's version. Or it could publish a page on what its team finds in pre-1950 warehouse conversions, with the three problems that come up most and what they cost to fix. The second takes an hour of a surveyor's time. The first takes ten minutes of an AI tool's. A site full of thin, similar posts also blurs what you are expert in, as we explain in [why ChatGPT ignores most blog posts](/insights/why-ai-ignores-most-blog-posts). ## What does a failed cheap package really cost? The fee is the smallest part. For example, twelve months at £750 is £9,000, plus a few hours of your own time each month reviewing drafts. If the package does not change who the assistants recommend, you have also spent a year while competitors who did the harder work became harder to displace. Then there is the internal cost. After a year of polished reports and no enquiries, the board concludes that "AI search does not work for us", and the next proposal, even a good one, starts from that scepticism. Our guide to [what inaction is costing you](/insights/cost-of-inaction-losing-right-now) shows how to put a figure on the ground you lose. ## When does a low-cost option make sense? When the job is narrow, with a clear task and a clear finish. Cheap is wrong only when it is asked to build a reputation on its own. - **Tracking what the assistants say.** Software does this well and cheaply, as the table below shows. - **A one-off listings clean-up.** Correcting your details on the directories and profiles that matter so they match your website. - **A specific technical fix.** Making sure AI crawlers can read your site, or correcting how search engines understand your firm. - **Supporting your own team.** If you have a capable marketer, a little outside advice or tooling can be good value. A narrow fix can matter a great deal when it targets the right problem. For [ID-Pal](/work/id-pal), which had bought NorthRow, a UK business verification platform, the fix was "About four hours of structured data work declaring NorthRow part of ID-Pal on both domains. Live in week one, with no copy changes and no risk to the redirect." The hours were few; finding the cause took the expertise. In the words of Dan Brooksbank, ID-Pal's Marketing Director: "Fortitude undertook specialist research and identified the primary cause within the first week of working together and had the fix live in the first month." Tracking tools, with list prices as shown on each vendor's page in September 2026: | Tool | Entry price | What the entry plan covers | |---|---|---| | [Otterly.AI](https://otterly.ai/pricing) | $29 a month | 15 prompts across ChatGPT, Google AI Overviews, Perplexity and Copilot | | [Ahrefs](https://ahrefs.com/pricing) | $129 a month (Lite) | Brand Radar with 5 custom prompts, alongside Ahrefs' SEO tools | | [Semrush](https://www.semrush.com/pricing/) | $199 a month (Starter) | 50 prompts tracked daily, alongside Semrush's SEO tools | | [Peec AI](https://peec.ai/pricing) | Plans by number of prompts, from 50 | A choice of three assistants on the smaller plans | | [Profound](https://www.tryprofound.com/) | Not shown on its home page | Built with enterprise features for large teams | ## What is the minimum budget to get recommended by ChatGPT? There is no single floor, because it depends on the gap between you and the firms already named, and on how much your own team can do. A firm with a strong website and existing trade coverage may need little more than tracking and a few targeted pages. A firm that needs its site fixed and its reputation built needs the expert hours in the table above, whether it buys them or supplies them. MarGen also warns that low monthly prices paired with claims of complete AI search work usually signal rebranded SEO. Our guide to [what AI search optimisation costs in the UK](/insights/ai-search-optimisation-cost-uk) shows how to set a budget from what a new client is worth. ## Which two questions tell us whether a provider is worth the fee? Ask every provider, cheap or expensive, these two: 1. **What will you do each month, in hours and outputs?** You want specifics: interviews with your team, pages rewritten, publications pitched, listings corrected. 2. **How will you measure what the assistants say about us, and connect it to enquiries?** Ask to see the question set and a sample report. The fuller checklist is in our guide to [choosing a firm to help you get recommended by ChatGPT](/insights/how-to-compare-ai-optimisation-providers). ## What to do this month 1. Ask ChatGPT, Google, Copilot and Gemini ten of your buyers' main questions and note who is named, so you know the size of the gap. 2. Work out what a new client is worth to you, so you can judge any fee against it. 3. For any package you are considering, get the monthly hours and outputs in writing and run the arithmetic above. 4. Check each proposal against the red flags: automated audits, bulk directories, paid list placements, volume posts, guarantees. 5. Buy the narrow jobs, such as tracking and listings, on their own terms, and resource the reputation work separately. ## Frequently asked questions ### Are £500 a month AI SEO packages worth it? For a narrow job, sometimes. For getting recommended, rarely: on our illustrative assumptions £500 buys about five to twelve hours a month, enough for monitoring and a little content but not the expert interviews and earned coverage that change who the assistants name. Ask what the hours will be spent on before you sign. ### Can we do AI search work ourselves, using ChatGPT to help? Some of it, yes. Your team can use ChatGPT to check what it says about you, rewrite service pages around buyer questions, correct your listings and track results each month. Use it to speed up writing that starts from your experts' knowledge; generating pages in bulk is what Google treats as scaled content abuse. The harder part to do alone is steady trade coverage. ### Do bulk directory submissions help AI visibility? No. Google lists low-quality directory links as link spam, and a hundred listings on sites nobody reads give an assistant no reason to trust you. An accurate, consistent entry on the handful of directories, associations and review sites your buyers use is worth far more. ### Is it worth trying a cheap package for three months as a test? Only if the test has a clear question and measure. Three months of generic posts will not tell you whether AI search can work for your firm. A better test is narrow: fix your core service pages and listings, then measure whether the assistants describe you more accurately. ### Why do some cheap providers show impressive results? Look at what is being measured. Traffic, posts published and tool scores can all rise with no change in who the assistants recommend. Ask to see the actual answers for a fixed set of buyer questions, before and after, and the enquiries that followed. If you want to see what the reputation side of the work involves when it is properly resourced, our [Reputation That Wins](/reputation-that-wins) page sets out how we build coverage and proof for founder-led firms, and [what it costs](/what-it-costs) lists our fees. ## Sources - [Stagg Studios, How much does AI search and technical SEO cost in the UK? (July 2026)](https://staggstudios.com/blog/how-much-does-ai-seo-cost-uk.html) - [MarGen, Agency pricing guide 2026: what UK businesses should expect to pay (June 2026)](https://www.margen.net/geo-agency-pricing-guide-2026-uk/) - [Muck Rack, What Is AI Reading? Earned media drives 84% of AI citations (May 2026)](https://www.globenewswire.com/news-release/2026/05/07/3290268/0/en/generative-pulse-earned-media-consistently-drives-ai-citations-holding-at-84.html) - [Google Search Central, Spam policies for Google web search (August 2026)](https://developers.google.com/search/docs/essentials/spam-policies) - [Google Search Central, Optimizing your website for generative AI features on Google Search (July 2026)](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide) - [Otterly.AI, Pricing (checked September 2026)](https://otterly.ai/pricing) - [Ahrefs, Pricing (checked September 2026)](https://ahrefs.com/pricing) - [Semrush, Pricing (checked September 2026)](https://www.semrush.com/pricing/) --- Title: Will ChatGPT and Google quote content written with AI or for SEO? URL: https://fortitudemedia.ai/insights/thought-leadership-vs-keyword-stuffing Author: Ashley Kenerson Published: 2026-03-31 Last reviewed: 2026-09-24 Topic: Content that gets quoted Short answer: Only if it is useful. Google says appropriate use of AI is not against its guidelines, but producing many pages mainly to manipulate rankings breaks its scaled content abuse policy, however they were written. Keyword stuffing does not help in AI answers either: in a 2024 Princeton and IIT Delhi study it lowered a page's visibility by about 8%, while adding quotations raised it by about 41%. What gets quoted is specific, first-hand explanation, whoever typed the first draft. Somewhere in your content folder there are now three kinds of article. The first was written to a keyword brief: a target phrase, a word count, headings that repeat the phrase and a paragraph or two under each. The second came out of ChatGPT or a similar tool, lightly edited, and reads smoothly without saying anything you would recognise as yours. The third was written by one of your partners after a difficult job, explaining what went wrong and what they would do differently. The first two are most of what many firms publish. The third is what a client forwards to a colleague. Will ChatGPT and Google quote the first two? Keyword engineering does not help, and it can hurt. In a [Princeton and IIT Delhi study presented at KDD 2024](https://arxiv.org/abs/2311.09735), keyword stuffing lowered a source's visibility in AI-generated answers by about 8%, while adding quotations from relevant people raised it by about 41% and adding statistics by about 34%. The test used a research engine, so treat the numbers as directional, but the direction is clear. AI drafting is a different question. Google does not penalise a page for being written with AI. It penalises pages made mainly to manipulate rankings, however they were produced, and a generic draft gives an assistant nothing it cannot find on a hundred other pages. Below: what Google has said, how assistants choose what to quote, the three kinds of page side by side, and what to do with old keyword pages. ## Key takeaways - Keyword stuffing does not help in AI answers: in the best-known test it lowered visibility by about 8%, while adding quotations raised it by about 41%. - Google says appropriate use of AI is not against its guidelines and judges content on quality, not on how it was produced. - Mass-producing pages to manipulate rankings breaks Google's scaled content abuse policy, whether people, AI or both wrote them. - Ranking still matters: ChatGPT search and Google's AI features start from search results, so a page has to be found before it can be quoted. - What gets quoted is specific, first-hand explanation: the mechanism, the exceptions and a real example, whoever typed the first draft. - Keep old keyword pages that still bring traffic or links, merge near-duplicates, and rewrite the ones that sit on questions buyers ask. ## Can Google or ChatGPT tell if our content was written by AI? Google says it does not need to. Its [guidance on AI-generated content](https://developers.google.com/search/blog/2023/02/google-search-and-ai-content) (February 2023) describes a "focus on the quality of content, rather than how content is produced", and says: "Using AI doesn't give content any special gains. It's just content." Unhelpful pages are demoted whether a person or a model wrote them. OpenAI, Anthropic and Perplexity have not said their assistants detect or penalise AI-written pages, and we know of no published study showing that they do. What the evidence suggests they reward is what an unedited AI draft usually lacks: quotations, figures, named sources and first-hand detail. Google's [guide to its generative AI features](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide) (updated July 2026) calls the alternative commodity content, which "is often based on common knowledge, which could originate from anyone, and typically adds little unique insight for readers". An unedited draft is commodity content almost by definition, because it is assembled from what is already published. ## Will Google penalise AI-generated content on our website? Not for being AI-generated. Google acts against pages made mainly to manipulate rankings. In March 2024 it announced a [scaled content abuse policy](https://developers.google.com/search/blog/2024/03/core-update-spam-policies) that applies "no matter whether content is produced through automation, human efforts, or some combination of human and automated processes", and said sites that break its spam policies "may rank lower in results or not appear in results at all". Its current [spam policies](https://developers.google.com/search/docs/essentials/spam-policies) (updated August 2026) list "Using generative AI tools or other similar tools to generate many pages without adding value for users" as an example. The AI features guide adds that creating separate content for every possible variation of a query breaks the same policy. | Practice | Google's position | Our view of the risk | |---|---|---| | Expert interview, AI-assisted first draft, expert review | Appropriate use of AI is not against its guidelines | Low, if the expert adds the substance | | AI draft, lightly edited, one page per real buyer question | Judged on quality like any other page | Rarely quoted, because it says what everyone says | | Dozens of AI pages a month, one per keyword variation | Named as scaled content abuse | High: pages can rank lower or disappear | | Keyword-stuffed pages, however written | Keyword stuffing is a named spam practice | High, and it lowers AI visibility too | | Location or sector pages with only the name swapped | Scaled content abuse if made mainly to rank | High, and the pages compete with each other | Should you say you used AI? Google says disclosures "are useful for content where someone might think 'How was this created?'" For an article drafted from an interview and checked by the named expert, a line saying so is usually enough. ## Does keyword stuffing still work for Google in 2026? No. Google's spam policies define keyword stuffing as "filling a web page with keywords or numbers in an attempt to manipulate rankings in Google Search results", and it is one of the named practices that can push a page down or out. The old idea that Google only counts words is years out of date. Google's [guide to its ranking systems](https://developers.google.com/search/docs/appearance/ranking-systems-guide) describes BERT as an AI system that lets it "understand how combinations of words express different meanings and intent", alongside RankBrain and neural matching, which relate words to concepts rather than matching exact strings. Google's AI features guide says plainly that "AI systems can understand synonyms and general meanings". Repeating a phrase adds nothing a language model needs, and it makes the page worse to read. ## How do ChatGPT and Google's AI answers choose which pages to quote? They search first, then choose. ChatGPT search, Copilot and Google's AI features start from a search index, retrieve a set of pages, and lift the passages that answer the question best. So ranking and substance both count. Ranking helps: [AirOps found in April 2026](https://searchengineland.com/chatgpt-citations-ranking-precision-length-study-474538) that pages ranking first were cited by ChatGPT 58.4% of the time, against 14.2% for pages ranking tenth. It does not guarantee a mention: [Ahrefs found in March 2026](https://ahrefs.com/blog/ai-overview-citations-top-10) that only 38% of AI Overview citations came from top-ten pages, down from 76% in July 2025, because Google now runs several related searches for one question (query fan-out). The Princeton and IIT Delhi study also found that lower-ranked pages gained most from changes such as citing sources. A page that ranks but says nothing specific can lose the citation to one that explains the mechanism. Our guide on [whether you still need SEO now that buyers ask ChatGPT](/insights/seo-vs-llm-optimisation) covers the wider picture. ## What do the three kinds of page look like side by side? For example, say a mid-sized accountancy practice wants a page on helping clients move to a new payroll provider. **Version one, written to a keyword brief:** "Payroll migration is a key challenge for businesses considering payroll migration. Our payroll migration service supports payroll migration for companies of all sizes. A successful payroll migration requires payroll migration planning. Contact us for payroll migration support." The phrase appears seven times and nothing is explained. **Version two, an unedited AI draft** (our own illustration of the pattern, not a real output): "Switching payroll providers can be a complex process, but with the right planning it can be smooth and stress-free. First, assess your current payroll needs. Next, research potential providers and compare their features and pricing. Communication is key: keep employees informed throughout the transition. Finally, test the new system thoroughly before going live." Every sentence is reasonable. None could only have been written by this firm, and the advice matches what many other pages already say. **Version three, written from experience:** "Most payroll moves go wrong for the same reason: the old provider's data is exported before anyone has checked what is missing. Pension opt-outs, salary sacrifice arrangements and year-to-date figures for people who joined mid-year are the usual gaps, and each takes a week to reconstruct once the old system is switched off. We now insist on a parallel run for one full pay period, which clients resist because it costs a month, and which has prevented every serious error we have seen since we introduced it. If your provider will not offer a parallel run, that is the first sign to look elsewhere." | | Keyword version | Unedited AI draft | Expert version | |---|---|---|---| | Times the phrase appears | Seven | Once or twice | Once | | Explains why moves go wrong | No | No | Yes: data exported before anyone checks it | | Specific detail | None | None | Pension opt-outs, salary sacrifice, year-to-date figures | | Trade-off admitted | None | None | A parallel run costs a month | | A judgement a buyer can act on | None | Generic steps | Look elsewhere if there is no parallel run | | What an assistant can quote | Nothing | Nothing it could not find elsewhere | Almost every sentence | An AI tool can help produce version three. Record the practitioner explaining the parallel run, have the tool draft from the transcript, and have the practitioner check it. The substance still comes from the person who did the job. ## What gives keyword-led and generic AI writing away? In our experience, a handful of patterns. None is a rule any assistant has published; they are simply what thin text looks like, and they leave a reader or a model nothing to use. - **Repetition nobody would say aloud.** The same phrase five times in four sentences. - **Sentences that contain the phrase and say nothing.** "A data strategy is important for firms implementing a data strategy." - **Even, shallow depth.** Five subsections of the same length. Real explanation is uneven: the crucial point gets six paragraphs and the minor one a line. - **No reasons.** It says what to do, never why, or why the obvious approach fails. - **No exceptions.** Nobody who has done a job for twenty years thinks their method always works. - **Steps that fit any firm.** Generic AI drafts lean on "first", "next" and "finally" where reasons should be. [Why ChatGPT ignores sites with hundreds of pages](/insights/ai-penalises-thin-content) shows what happens when a whole site reads this way, and [what makes content read as expert rather than marketing](/insights/what-makes-expert-quality-content) covers the signals to aim for. ## How do we write for ChatGPT without losing our Google rankings? Write one page for both. Google's AI features guide says optimising for its generative AI features is still SEO, and "You don't need to write in a specific way just for generative AI search." The habits that get a page quoted are the ones Google's quality guidance already asks for. - Build each page around a buyer's question, using the buyer's words once in the title and the first paragraph. - Put the direct answer in the first paragraph and the key figure near the top. - Keep pages that rank. Add the missing specifics, examples and limits rather than rewriting from scratch. - Cover the follow-up questions, because Google's AI features run related searches for each question. ## Should we delete old SEO blog posts written around keywords? Not all at once. Check each page's traffic, its links and whether it sits on a question buyers ask, then keep, rewrite, merge or delete. | What the page is | What to do | |---|---| | Brings traffic or links, and sits on a real buyer question | Rewrite with the mechanism, examples and limits; keep the address | | Brings traffic, but on a question you do not care about | Leave it, or tidy the worst repetition | | One of several near-duplicates on the same subject | Merge into the strongest page and redirect the others | | No traffic, no links, no buyer question | Delete, or redirect to the closest useful page | We know of no published evidence that ChatGPT or Perplexity score a whole site when choosing a citation, so the case for pruning is practical: near-duplicates compete with each other, and a buyer who lands on a thin page judges the firm by it. [How to audit existing content for AI assistants](/insights/audit-existing-content-ai-readiness) sets out the scoring method. ## How do we shift a content programme without losing what works? Keep the pages that still earn traffic and change how new work is produced. ### Step 1: Choose topics by difficulty, not volume Ask what buyers struggle to understand, where the received wisdom is wrong, and what you know that is not written down anywhere. These questions may have low search volume, and they are often the ones with no good source yet. ### Step 2: Put a practitioner in the room A forty-minute recorded conversation with the person who has done the job, and a writer, with or without an AI tool, who turns it into prose without sanding off the specifics. ### Step 3: Write to the length the subject needs Hard subjects need the mechanism, the exceptions and an example; narrow questions need a short, exact answer. [Whether ChatGPT prefers long or short articles](/insights/long-form-vs-short-form-ai-prefers) sets out what the studies say. ### Step 4: Put a name on it A byline such as "by the head of payroll, sixteen years in practice" beats "by the team", and the same name and role should appear on LinkedIn. ### Step 5: Bring a number when you have one A finding from your own records is something an assistant can only attribute to you; see [whether publishing your own research helps](/insights/original-research-data-ai-trust). ### Step 6: Link the pieces to each other Link related pieces so readers and crawlers can follow the subject; [organising content so AI assistants see you as the expert](/insights/topic-clusters-ai-understands) explains how. ## What to do this month 1. Read your ten most visited articles aloud for thirty seconds each and mark any that repeat a phrase or say nothing a competitor could not. 2. Write down your rules for AI drafting: which steps a tool may do, who adds the substance and who checks every figure. 3. Pick the three questions your clients struggle with most and check whether any page answers them with reasons and exceptions. 4. Record a forty-minute conversation with one practitioner on one of those questions and publish it as one thorough piece under their name. 5. Change the brief you give your writer or agency: no target phrase, no word count, one question to answer completely and an expert to interview. ## Frequently asked questions ### Will writing this way hurt our Google rankings? In our experience it helps them, because pages people read, share and link to tend to be the expert ones. Leave old keyword pages that still bring traffic in place and let the new work overtake them. ### How long before we see a difference? Assistants that search the web live, such as ChatGPT search, Perplexity, Copilot and Google's AI features, can reflect a new page within weeks. Answers drawn from a model's own memory change only when it is retrained, which takes months. ID-Pal's case study gives a dated example: "UK buyer questions where AI names ID-Pal: none in May, two in August, four in September 2026." See the [ID-Pal case study](/work/id-pal). ### Do we need a famous author? No, a credible one. A name, a role, a sentence of background and a body of work on one subject is enough. What matters is that the same identifiable person keeps explaining the same field well. ### Can an outside writer produce expert content? Yes, provided a real expert is the source. The writer interviews, structures and makes it readable; the expert supplies the mechanism, the examples and the exceptions. [How to outsource content without losing your voice](/insights/outsourcing-content-brand-voice) sets out the briefing method. ### We are in a commodity sector. What is there to be expert about? The use of the product rather than the product itself: where implementations go wrong, what the hidden costs are and which situations suit which option. A freight forwarder cannot be a thought leader on containers, but can be on why customs delays happen and how to avoid them. If you want a second opinion on how your content reads to buyers and assistants, [book a free Growth Review](/growth-review): 30 minutes with David Adams and three fixes in writing. Turning your partners' expertise into content that gets quoted is part of [Reputation That Wins](/reputation-that-wins). ## Sources - [arXiv, Aggarwal and others, Princeton and IIT Delhi: source visibility in AI-generated answers, KDD 2024 (June 2024)](https://arxiv.org/abs/2311.09735) - [Google Search Central Blog, Google Search's guidance about AI-generated content (February 2023)](https://developers.google.com/search/blog/2023/02/google-search-and-ai-content) - [Google Search Central, Guide to optimizing for generative AI features on Google Search (July 2026)](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide) - [Google Search Central Blog, What web creators should know about our March 2024 core update and new spam policies (March 2024)](https://developers.google.com/search/blog/2024/03/core-update-spam-policies) - [Google Search Central, Spam policies for Google web search (August 2026)](https://developers.google.com/search/docs/essentials/spam-policies) - [Google Search Central, A guide to Google Search ranking systems (December 2025)](https://developers.google.com/search/docs/appearance/ranking-systems-guide) - [Search Engine Land, AirOps study of ChatGPT citations, ranking and page length (April 2026)](https://searchengineland.com/chatgpt-citations-ranking-precision-length-study-474538) - [Ahrefs, AI Overview citations and top-ten rankings (March 2026)](https://ahrefs.com/blog/ai-overview-citations-top-10) --- Title: Will UK and EU regulation change who AI assistants recommend? URL: https://fortitudemedia.ai/insights/regulation-ai-search-uk-eu Author: Ross Williams Published: 2026-03-31 Last reviewed: 2026-09-24 Topic: How buying is changing Short answer: Not directly. No UK or EU rule gives a firm a right to be recommended. The EU AI Act adds transparency duties from August 2026, the Digital Services Act now covers ChatGPT as a very large search engine, and the CMA has required Google to let publishers opt out of AI Overviews and to credit them with clear links. The practical effect is that accurate, evidenced, consistent information about your firm becomes more valuable, and unsupported claims become riskier. Do not wait for the law. A director on your board reads that the EU's AI law is now in force and asks a reasonable question: does this mean the AI assistants will have to treat us fairly? If ChatGPT keeps naming three competitors and not us, can we complain to someone? And do we need to do anything ourselves to stay on the right side of it? They are good questions, and the rules are moving quickly. On 31 August 2026 the European Commission [designated ChatGPT a very large online search engine](https://digital-strategy.ec.europa.eu/en/policies/list-designated-vlops-and-vloses) under the Digital Services Act, reporting 159.1 million average monthly users in the EU, and in June 2026 the UK's Competition and Markets Authority imposed its first rules on Google's AI Overviews. None of this hands any firm a right to be recommended, and a business that waits for regulation to fix its invisibility will wait a long time. This guide sets out what the rules cover, which laws already apply to what you publish, and what to do. Where the law is unsettled, we say so. It is general information, not legal advice. ## Key takeaways - No UK or EU rule gives a firm a right to be named by an AI assistant, and we know of none proposed. - The EU AI Act's transparency duties apply from August 2026, while the May 2026 "AI Omnibus" agreement moved most high-risk duties to December 2027. - The strongest constraints on the platforms come from competition and platform law: ChatGPT was designated under the Digital Services Act in August 2026, and Google is bound by the EU Digital Markets Act and UK competition rules. - The CMA designated Google with strategic market status in October 2025 and in June 2026 required it to let publishers opt out of AI Overviews and to credit them with clear links. - Claims to business buyers fall under the Business Protection from Misleading Marketing Regulations 2008 and the CAP Code, and an assistant repeating your claim does not make it someone else's problem. - The rules reward what the assistants reward: accurate, evidenced, consistent information about your firm. ## What does the EU AI Act actually cover? The EU AI Act regulates AI according to the risk of the use it is put to. General-purpose assistants answering business questions sit in the lower tiers, where the obligations are mostly about transparency, and recommending a supplier is not a high-risk use. The Act entered into force on 1 August 2024 and applies in stages. The [AI Omnibus](https://digital-strategy.ec.europa.eu/en/policies/regulatory-framework-ai), a package to simplify the Act, reached political agreement on 7 May 2026 and entered into force on 27 July 2026, pushing back the high-risk rules. | What applies | From when | |---|---| | Banned uses and AI literacy duties | 2 February 2025 | | Duties on providers of general-purpose AI models, such as the models behind ChatGPT, Claude and Gemini | 2 August 2025 | | Transparency duties, including labelling of AI-generated content (Article 50) | August 2026 | | High-risk uses in sensitive areas such as recruitment, credit and education (Annex III) | 2 December 2027 | | High-risk AI built into regulated products (Annex I) | 2 August 2028 | Source: European Commission, AI Act page, updated August 2026. What the Act does not do is treat an assistant's answer to "which accountancy firms should I talk to" as a high-risk decision. There is no provision obliging a platform to explain why a business was not named, no right of appeal against being left out, and no duty of equal treatment between suppliers. Anyone telling you otherwise is describing a law that has not been written. The duties on model providers still matter to you indirectly. Providers must document their models, publish a summary of the content used to train them and respect copyright holders' opt-outs from training, which makes the models easier to study. Whether an assistant can read your site at all is a separate, technical question, covered in [how AI crawlers read your website](/insights/ai-crawlers-vs-google-spiders). ## Does the EU AI Act apply to UK businesses that use ChatGPT? Only at the edges. [Article 2](https://ai-act-service-desk.ec.europa.eu/en/ai-act/article-2) applies the Act to firms outside the EU when the output of the AI system is used in the EU, so a UK firm using ChatGPT to draft material for EU clients is within its reach in principle. In practice the duties that bite on ordinary business use are few: the transparency rules on labelling some AI-generated content, covered below, and the high-risk rules if you use AI for decisions such as recruitment, which now apply from December 2027. Using an assistant to draft proposals or research suppliers is not a high-risk use. ## Where does the real constraint on platforms come from? The rules with most effect on which firms an assistant names are not AI rules at all. They are the EU's platform and competition rules, and the UK's new competition powers, which restrict how the largest companies rank and favour services. The Digital Markets Act [designated Alphabet a gatekeeper](https://digital-markets-act.ec.europa.eu/gatekeepers_en) in September 2023, with Google Search among its core services, and prohibits gatekeepers from ranking their own services more favourably than rivals'. The Digital Services Act requires very large platforms and search engines to explain the main parameters of their recommender systems (the software that decides what you are shown) and to assess the risks their services create. Google Search and Bing have been designated since 2023; ChatGPT joined them on 31 August 2026, the first AI assistant on that list. For a UK B2B firm the practical effect is modest but real. If an AI feature systematically favoured its owner's products or paid partners over merit, that is the behaviour these rules target, and regulators now have ChatGPT as well as Google in view. None of it helps you against a competitor who is simply better evidenced than you. How the assistants differ is set out in [which AI assistants buyers actually use](/insights/ai-search-landscape-chatgpt-perplexity-google-claude). ## How does the UK approach differ? The UK has not passed an AI-specific Act. It asks existing regulators, among them the Information Commissioner's Office, the CMA, Ofcom and the Financial Conduct Authority, to apply broad principles such as safety, transparency, fairness and accountability within their own remits. Those principles guide regulators; they are not rights a business can enforce. The AI Security Institute evaluates frontier models for safety risks and does not regulate how they recommend suppliers. The most relevant UK tool for founders is competition law. The Digital Markets, Competition and Consumers Act 2024 lets the CMA give firms "strategic market status" in a digital activity and impose conduct requirements on them. The UK is lighter-touch than the EU, and large platforms tend to build for the stricter regime and apply it everywhere, so UK firms feel many EU effects second-hand. ## What is the CMA doing about Google AI Overviews? It has designated Google and imposed its first rule on AI Overviews. On 10 October 2025 the CMA [confirmed Google has strategic market status](https://www.gov.uk/government/news/cma-confirms-google-has-strategic-market-status-in-search-services) in general search and search advertising, and said AI Overviews and AI Mode are within the designation; the Gemini app is not, though the CMA will keep that under review. On 3 June 2026 it [imposed a conduct requirement](https://www.gov.uk/government/news/cma-secures-fairer-deal-for-publishers-and-improves-google-search-services-in-uk) obliging Google to let publishers opt out of their content being used to power AI features such as AI Overviews, and of its use for fine-tuning AI models, and to attribute publisher content with clear links in AI-generated results. Google has nine months to put all of it in place and must publish compliance reports every six months in the first year. As with the EU rules, the target is the gatekeeper's behaviour, not the fairness of any individual answer. ## Can I stop Google using my website in AI Overviews? Yes, and for most B2B firms you should not. Search Console already has a [Search generative AI control](https://support.google.com/webmasters/answer/16908024?hl=en) that excludes a site from AI Overviews, AI Mode and generative AI features in Discover; inclusion is the default, and Google says the control is not a ranking signal for the rest of Search. Google's [preview controls](https://developers.google.com/search/docs/appearance/ai-features), such as nosnippet, also apply to its AI features. Blocking Google-Extended in robots.txt stops your pages being used to train Gemini but does not remove you from AI Overviews. The opt-out suits a publisher that sells its content, where a summary replaces a sale. A firm that sells a service wants the opposite: to be the source the summary draws on and the name the buyer reads. We cover the trade-off in [why traffic falls when rankings hold](/insights/zero-click-search-your-business). ## Can we pay to appear in AI answers, and which rules apply? You can pay for an ad beside the answer, not for a place in it. ChatGPT has shown labelled ads to UK users on its Free and Go plans since 11 August 2026, and [OpenAI says](https://openai.com/index/testing-ads-in-chatgpt/) the ads are kept separate and do not change the answer itself. Being named in the answer is still earned. If you advertise inside an assistant, the usual advertising rules follow you. The [CAP Code](https://www.asa.org.uk/type/non_broadcast/code_folder/scope-of-the-code.html) covers online advertisements in paid-for space and defines a consumer as anyone likely to see an ad, "whether in the course of business or not", so B2B ads are covered. Your claims must be accurate and your ad must be recognisable as an ad. ## Which laws already apply to what we publish? The rules most likely to affect a founder-led firm are old ones, applied to claims an assistant may now repeat. For marketing aimed at other businesses, the main law is the [Business Protection from Misleading Marketing Regulations 2008](https://www.legislation.gov.uk/uksi/2008/1276/contents), which prohibit advertising that misleads traders and set conditions for comparing yourself with competitors. The CAP Code also covers marketing claims on your own website. Consumer protection law applies where your audience includes consumers. If your website says "the UK's leading provider" or "award-winning" without a basis, that was always a problem. It becomes a sharper one when ChatGPT repeats the claim to a buyer as fact, because the claim now travels further than your page and the source is still you. If two of your pages contradict each other, an assistant may pick either, or say the information is unclear. The mechanics are in [why ChatGPT describes a firm wrongly](/insights/ai-conflicting-information-online). Data protection law covers personal data, not business information, so it cannot be used to stop assistants reading your website. You can block particular crawlers in robots.txt (a small text file that tells crawlers what they may read), but for most B2B firms that is self-defeating, because being read is how you get named. ## Is it legal for ChatGPT to say false things about my company? No law requires an assistant to be right, and there is no regulator that corrects individual answers. Defamation law can apply to false statements about a company, but under [section 1 of the Defamation Act 2013](https://www.legislation.gov.uk/ukpga/2013/26/section/1) harm to a business that trades for profit counts as serious only if it has caused or is likely to cause serious financial loss, and how that applies to AI answers is not settled. Take legal advice before going down that route. The quicker fix is upstream. Errors in AI answers usually trace back to a source you can correct: an old page of your own, an outdated directory listing or a stale profile. Correct the source, use the platform's feedback route, and check the answer again after the assistant has had time to search again. ## Do we need to label AI-generated content on our website? In the UK there is no general legal duty to label AI-assisted writing, but everything you publish must still be accurate under the advertising rules. The EU AI Act goes further for content used in the EU. [Article 50](https://ai-act-service-desk.ec.europa.eu/en/ai-act/article-50) requires anyone publishing AI-generated text "with the purpose of informing the public on matters of public interest" to disclose it, unless it "has undergone a process of human review or editorial control" and someone holds editorial responsibility for it. AI-generated images, audio or video that amount to deep fakes must be disclosed. The practical rule for a B2B firm is simple: have a named person review and take responsibility for everything you publish, and label any synthetic images or video. That keeps you inside the exception and is what makes content worth citing anyway. ## What does this mean for a founder-led firm? Regulation will not get you recommended, but it makes the evidenced approach safer and the inflated approach riskier. Three conclusions follow. First, do not wait: no rule on the horizon will require an assistant to name you, and meanwhile a competitor is becoming the answer, as we argue in [why you cannot wait a year on AI search](/insights/first-mover-advantage-ai). Second, audit your claims, because assistants and regulators now both read them. Third, make your information consistent, because assistants cross-check, buyers verify, and transparency rules make inconsistency more visible. If your buyers are in the EU as well as the UK, the advice is the same. If AI agents that buy on a customer's behalf reach your sector, the evidence they weigh is exactly what the rules encourage you to publish, as we describe in [what happens when an AI agent builds the shortlist](/insights/ai-agents-change-b2b-buying). ## What to do this month 1. Read every claim on your homepage and service pages and mark each as verifiable, replaceable or removable, then fix the last two groups. 2. Ask ChatGPT, Gemini, Copilot, Claude and Perplexity what they say about your firm, note anything out of date or wrong, and correct the source page. 3. Check your robots.txt, any Cloudflare setting and Search Console's generative AI control to make sure you are not keeping assistants out by accident. 4. Agree who reviews and signs off AI-assisted content before it is published, and label any synthetic images or video. 5. Brief the board in one page: what the rules cover, what they do not, and what the firm is doing about being named regardless. ## Frequently asked questions ### Can we complain if ChatGPT keeps recommending our competitors? You can use the platform's feedback routes, but there is no legal right to be recommended and no regulator that rules on individual answers. The productive response is diagnostic: look at what the assistant cites for your competitors, work out what evidence they have that you lack, and publish it. That usually changes the answer faster than a complaint. Our guide to [tracking whether AI tools recommend you](/insights/track-ai-recommending-your-business) shows how to record it month by month. ### Will regulation force the assistants to show their sources? Partly. The CMA now requires Google to credit publishers with clear links in AI results, and platform rules push towards transparency. The assistants have mostly moved there themselves: Perplexity numbers its sources, and ChatGPT, Claude, Copilot and Google's AI features link to pages when they search. Visible sources are the most useful development for a firm, because they show you what to fix. ### Should we block AI crawlers to protect our content? For most B2B firms, no. Blocking the search crawlers means the assistants cannot read your service pages, so they cannot name you. Keep confidential material off the public web rather than relying on crawler instructions, which are requests, not barriers. ### Is there any risk in what the assistants say about us? Some. If an assistant repeats an old price, a service you no longer offer or a claim you cannot support, a buyer may act on it. Keep your own pages accurate and consistent, correct third-party listings, and check monthly what the assistants say. ### Will any of this slow the shift to AI search? Not noticeably. The rules add transparency and constrain the largest platforms, but they do not stop buyers asking assistants who to hire. The direction for a founder is unchanged: be the clear, evidenced answer, and check regularly that you are. The law will make the assistants a little more open and the largest platforms a little more careful. It will not decide who gets named; the evidence will. To see where your firm stands today, [book a free Growth Review](/growth-review): 30 minutes with David Adams and three fixes in writing. If the gap is what others say about you, [Reputation That Wins](/reputation-that-wins) is built to close it. ## Sources - [European Commission, AI Act: regulatory framework and application timeline (August 2026)](https://digital-strategy.ec.europa.eu/en/policies/regulatory-framework-ai) - [European Commission AI Act Service Desk, Article 50: transparency obligations (2026)](https://ai-act-service-desk.ec.europa.eu/en/ai-act/article-50) - [European Commission, List of designated very large online platforms and search engines (September 2026)](https://digital-strategy.ec.europa.eu/en/policies/list-designated-vlops-and-vloses) - [Competition and Markets Authority, CMA confirms Google has strategic market status in search services (October 2025)](https://www.gov.uk/government/news/cma-confirms-google-has-strategic-market-status-in-search-services) - [Competition and Markets Authority, CMA secures fairer deal for publishers and improves Google search services in UK (June 2026)](https://www.gov.uk/government/news/cma-secures-fairer-deal-for-publishers-and-improves-google-search-services-in-uk) - [legislation.gov.uk, The Business Protection from Misleading Marketing Regulations 2008 (2008)](https://www.legislation.gov.uk/uksi/2008/1276/contents) - [Advertising Standards Authority, CAP Code: scope of the Code (2026)](https://www.asa.org.uk/type/non_broadcast/code_folder/scope-of-the-code.html) - [OpenAI, Testing ads in ChatGPT (updated August 2026)](https://openai.com/index/testing-ads-in-chatgpt/)