What is it costing us not to be recommended by ChatGPT?

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, 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 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.

#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 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.

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:

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 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 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.

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.

• 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.

The flip side is what being named is worth. For 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.
• 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, and present it with making the business case to your 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 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: 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