Asking ChatGPT yourself won't tell you. Your account already knows who you are, so it can name you back. I check six AI tools the way a stranger nearby would ask, and tell you who gets named instead.
No charge, and nothing to prepare.
I write three questions the way one of your customers would actually type them, not as keywords.
Then each one goes to ChatGPT, Claude, Gemini, Copilot, Google's AI answers and Perplexity. Every query is run from your area, not from wherever I happen to be, and from accounts with no history, because an account that already knows about you will tell you what it remembers rather than what it found.
That is the part most people get wrong. Ask ChatGPT about your own business while logged into your own account and it will often name you.
You get all eighteen answers, including the ones where you do not appear. Those are the important ones.
You can, and you should. Here is how to do it so the answer means something: a private window, signed out, a fresh chat for every question, and the same wording on every tool.
Ask from your own account and it will often name you. Not because you are visible, but because it remembers you.
That is the version most people run. It returns a confident answer in the flattering direction, which is worse than not checking at all.
That is not visibility. That is memory.
This is my own business, checked in August 2026, when the check ran across four tools rather than six. Same three questions, same controls.
| Run 001 | Question 1 | Question 2 | Question 3 |
|---|---|---|---|
| ChatGPT | named 1st | named 1st | named 1st |
| Google AI answers | not named | not named | not named |
| Perplexity | not named | not named | not named |
| Claude | nobody named | not named | nobody named |
One of the four names me first, every time. Three have never heard of me. Until I ran this I assumed I was either visible or I was not, and the truth turned out to be neither.
All twelve were run with memory off, from accounts with no history of my business, and my name appears nowhere in any of the three questions. The three ChatGPT runs were separate temporary chats. That is the whole reason those three count for anything: the easiest way to fake this test is to ask an assistant that already knows you.
Two of the blanks are a different thing, which is why they are a different colour. On Google and Perplexity I am absent and other businesses are named in my place. On two of Claude's three, nobody was named at all: it searched, and reported it could not confirm anyone doing this here. Being beaten and being in an empty category are not the same finding, and the second one is usually the more useful half.
Ask the same question next week and the wording will be different. This records what came back on the day, which is why it is dated.
Some honest context: this is a new site with no history, no reviews yet, and almost nothing pointing at it from anywhere else. I am at the start of exactly the problem I fix. If I were showing you a perfect score I would be showing you something you could not learn anything from.
And the spread matters more than the score. Four tools, the same three questions, and they do not agree with each other. Being findable is not one thing you either have or you do not.
No recommendations, no action plan, no proposal. This tells you what is true. What to do about it is a separate conversation, and only if you want one.
Whether each one named you, and where you came in the list
Every business that was recommended instead of you, and how often
The three questions in full, so you can run them yourself
The exact words each tool used, quoted, including the answers where you are absent
I run five of these a week. Each one is run by hand from your actual area, because the answer changes depending on where the question is asked from, and a tool cannot do that. If this week is full you go on the list and I will tell you when yours is coming. You will hear from me either way.
Your details are used to run and send your check. I do not pass them to anyone. Privacy policy
I am Eva Petrakova. I spent six years building operational systems for the Metropolitan Police, London Fire Brigade and British Airways, where a missed detail is a failure rather than an inconvenience. I run Local Business Systems from Surbiton, working with coaches and therapists in Kingston upon Thames and the surrounding area.
You will work with me directly. I look after every client personally.