Can we just do content, or do we need PR and a better website too?

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 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.

#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.

#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, 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.

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.

#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.

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. To run the sum with your own figures for a board, see how to get board approval for AI search.

#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 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.

#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, and what a B2B website should 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 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 and Reputation That Wins, and our prices are on 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.

#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.

#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: 30 minutes with David Adams, and three fixes in writing afterwards.

#Sources