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.
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.
What it is costing you.
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.
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 Note anything wrong or out of date.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.
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 ↗The short answers.
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.
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.
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.
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 WinsChatGPT 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 SellsThree things worth reading first.
How do we find out whether ChatGPT is recommending our firm?
31 March 2026
Why does ChatGPT sound unsure about our firm, and how do we fix it?
31 March 2026
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.