Who do AI assistants recommend in your city? A test you can run in an hour
· 6 min read
Asking an assistant about yourself tells you whether it knows you. Asking it for a recommendation tells you something harder: who it names instead, and why. This guide gives a fair way to run that test for your own profession and city, how to record and read the answers, and what the pattern usually means. We have not run this across cities and publish no rankings here, because answers change from run to run and a fair study would need many repeated runs. The method is the useful part, and it is yours to run.
The short answer
- Ask for recommendations the way a client would, not by name, in several wordings, and repeat each several times. One answer proves nothing.
- Record who is named, in what order, what reasons are given and which sources are cited. The sources tell you where assistants look.
- Answers differ between assistants, between runs, between places and between logged-in and logged-out use. Report ranges, not single results.
- If you are not named, check whether you are findable, consistent and described in plain sentences. If you are named, check that the details are right.
- We have not published results for any city. Treat any ranking of “who ChatGPT recommends” you see with caution unless it says how many runs it used.
What this test is for
- To learn whether an assistant names you when a client asks for someone like you.
- To see who it names instead, and what those people and firms have in common.
- To see which pages and sites the assistant relies on, when it shows its sources.
- To find out what it says about prices, availability and reputation in your market.
It is a different test from asking about yourself by name. For that, see check what AI says about you.
Why one answer proves nothing
- Assistants produce different wording, and sometimes different names, each time you ask the same question.
- Small changes in the question change the answer: “best” against “good”, “near me” against a city name, “affordable” against “experienced”.
- Location matters. Some assistants use your location, your settings or your past conversations; some do not.
- Assistants that search the web draw on whatever they find that day. Those that do not draw on older training and may be out of date.
- Different assistants behave differently. A result from one does not carry over to another.
So treat each answer as one sample. Several runs of several wordings give you a picture; one gives you an anecdote.
Set up a fair test
- Choose two or three assistants that your clients are likely to use. Include one that shows sources and one that does not.
- Use a fresh conversation for each run, so earlier answers do not shape later ones. Where possible, log out or use a private window.
- Note the date, the assistant and its mode (for example, with or without web search), and whether your location was shared.
- Write down your questions first and do not change them between runs.
- Run each question at least five times, spread over a few days.
- Use a second person, with a different device and account, for a handful of runs, to see whether your own history affects the result.
Twelve questions that sound like clients
Replace the profession and place with yours. Keep the wording natural.
- Who is a good physiotherapist in Bristol for a sports injury?
- I need a divorce lawyer in Austin who explains things clearly. Who should I look at?
- Can you recommend an architect in Pune for a home renovation?
- Which Steuerberater in Hamburg works with freelancers?
- Where can I find a therapist in Manchester, England, who offers evening appointments?
- Which dentist in Lyon is good with nervous patients?
- How much does a first appointment with a physiotherapist in Bristol cost, and who offers one?
- What should I look for when choosing a divorce lawyer in Austin, and can you name some options?
- Compare three architects in Pune and tell me why you chose them.
- Which professionals in my area are well reviewed? (Add your city.)
- Who offers fixed-fee work for [your service] in [your city]?
- Which of these is best for me, and why? (After a list, ask for reasons and sources.)
Record the answers
Date | Assistant and mode | Question (number) | Run | Names given, in order | Reasons given | Sources cited | Am I named? | Details correct?
- Write the names exactly as given, including any firm names and misspellings.
- Copy the reasons given, such as “well reviewed”, “specialises in”, “fixed fees” or “near the station”. They show what the assistant values.
- Copy every source it cites. These are the pages that, at least that day, shaped the answer.
- Mark any wrong detail: wrong city, phone, price or profession.
How to read what you find
| What you see | What it often means | What to do |
|---|---|---|
| You are never named | The assistant may not find enough clear, consistent information about you | Check that you have a plain page, a complete Google profile and the same details everywhere. See what is GEO |
| You are named sometimes | You are on the edge. Small improvements in clarity and consistency can tip it | Repeat the test after each change, and look at who is named when you are not |
| The same few names appear every time | They have strong, consistent presence and, often, many reviews and mentions | Read their pages and profiles. What do they state plainly that you leave out? |
| The sources are mostly directories | The assistant relies on listings | Complete and correct your profiles there. See directories that matter by country |
| The sources include your competitors’ own pages | Their pages state services, places and prices in plain text | Make yours do the same |
| You are named with wrong details | The assistant combined conflicting information | Fix the source. See why AI gets your details wrong |
| Answers vary wildly between runs | The assistant has little to go on in your market | A clear, consistent presence helps most where there is little else |
What not to conclude
- That an assistant’s list is a ranking of quality. It reflects what it found and how it weighed it, not who is the best professional.
- That you can pay or trick your way in. Tactics that stuff pages with keywords or invent claims risk harm and do not last.
- That the pattern is permanent. Assistants and their sources change.
- That a single run for a single wording is a result.
- That an assistant’s statement about price or availability is current. Check it yourself.
Rules for honest reporting
If you publish what you find, for example in a newsletter or a post, protect your readers and the people you name.
- Say which assistants, which dates, how many runs and which wordings.
- Report ranges: “named in 3 of 10 runs”, not “ranked third”.
- Do not name or shame individuals on the strength of a few runs.
- Do not present an assistant’s list as an endorsement, and do not claim that you were “recommended by” an assistant in your own advertising without checking the rules in your profession. See lawyer websites and doctor and clinic websites.
Where BeVisible fits
BeVisible builds a clear page with your name, role, place, services and contact route in plain text, and generates machine-readable files that describe your page. That gives assistants accurate material to draw on, but it cannot decide what any assistant says, and we make no promise of being named. You can build and preview free, and publish free for 7 days with no card needed; after that one yearly plan keeps a website online, and one plan covers one website. Start from a photo or type your details.
Frequently asked questions
How do I find out who ChatGPT or other assistants recommend in my city?
Ask the way a client would, in several wordings, in fresh conversations, and repeat each question at least five times over a few days. Record who is named, the reasons given and the sources cited, and report ranges rather than single results.
Why do assistants give different answers each time?
They generate wording afresh, small changes in the question change the answer, and assistants that search the web draw on whatever they find that day. Location, settings and history can also matter.
If I am not named, what should I check first?
Whether you have a plain page of your own, a complete Google Business Profile and identical details everywhere, and whether your page states your services, place and prices in clear sentences.
Is an assistant’s list a ranking of the best professionals?
No. It reflects what the assistant found and how it weighed it. Use it as a signal about your online presence, not a judgement on your work.
Can I say I was recommended by an AI assistant in my advertising?
Be careful. Answers vary between runs, and advertising rules in many professions restrict testimonials, endorsements and comparisons. Check your regulator before making such a claim.
Have you published results for particular cities?
No. A fair study needs many repeated runs of several assistants, which we have not done. This guide gives you the method to test your own market.