AI visibility / GEO

When someone asks AI who to hire, does your name come up?

Buyers increasingly ask a model instead of opening a results page. This tool asks ChatGPT, Gemini, Claude, Perplexity and Grok the question a customer in your city would ask, and shows you what they answered - whether you were recommended, where you came, and who was named instead.

We ask each engine one real question and read the answer. Nothing is stored, and your business name is not sent anywhere except to the engines being asked.

We send one question - "which {sector} in {city} would you recommend" - to each engine that is switched on, in the language of this page, and read the businesses it names. An engine we cannot reach is reported as not checked and is left out of the score rather than counted as a zero. Models answer differently each time they are asked, so treat a single run as one sample rather than a ranking.

The guide behind this tool How to Find Out Whether AI Recommends Your Business A method for measuring whether ChatGPT, Gemini, Claude and Perplexity name your business when a customer asks — and what each possible answer means. Read the guide · 7 min read →

What this tool checks

One question per engine, asked the way a customer asks it.

  • Are you recommended

    Whether the engine names your business when asked who to use in your city. This is the only outcome that wins the job.

  • Where you come

    First is the answer a buyer acts on. Fourth is a name in a list they skim. The position is scored, not just the mention.

  • Known but not recommended

    The engine may know you exist and still put someone else forward. That is a different problem from being unknown, and it is reported separately rather than counted as a miss.

  • Who is named instead

    Every competitor each engine put forward, ranked by how often they came up across all of them.

  • Share of voice

    Of all the names the engines handed out, how many were yours. A recommendation among three is worth more than one among ten.

  • How many engines actually answered

    Printed on the result. A score from two engines is a weaker claim than a score from five, and the page says which it is.

How to read your result

Not mentioned by any engine is the common result, and it is not a verdict on your business. Models name what they have read about. If nothing on the open web describes what you do, where you do it and who says you are good at it, there is nothing for a model to retrieve.

Known but not recommended is the most useful result you can get. The model has your facts and is choosing someone else. That is usually a thin or unclear description of what you specialise in, or the absence of the third-party corroboration - directories, press, reviews - that models lean on when they have to pick.

Recommended, but fourth or later means you are in the consideration set and losing on evidence. Look at who came first and what exists about them that does not exist about you.

A different answer on a second run is normal, not a bug. These systems are not deterministic. Being named in three runs out of five is a real position; being named once is noise.

Why this is not the same as ranking on Google

A results page lists; a model chooses

Ten blue links let the buyer decide. A model hands over two or three names and a reason. There is no page two, and the businesses that are not named are not merely lower down - they are absent from the conversation.

Models answer from what has been written about you, not by you

Your own site establishes the facts: what you do, where, for whom. But when a model has to choose between businesses that all claim to be good, it leans on everything else - directories, local press, forum threads, review platforms, the sites of people you have worked with. A polished site with no corroboration anywhere else is a weak position.

Specificity beats reach

"Digital agency" is a category with thousands of members and no way for a model to separate them. "Multilingual e-commerce builds for health tourism clinics" is a category with few, and a model asked that question has an easy answer. The narrower the claim you can truthfully make, the more often you are the answer.

This is measurable, slowly

One run tells you where you stand today. Run it monthly, the same question, and the trend is the signal. A single score is a snapshot of a system that changes its mind.

Frequently asked questions

Why does the score change when I run it twice?

Because the models do. They are not lookup tables; the same question produces different wording and sometimes different names. We show the number of engines that answered so you can weigh it, and we would rather report that honestly than average it into something that looks stable and is not.

Some engines say "not checked". Why?

Each engine needs an API key on our server, and every run costs money at that provider. Whichever are switched on are asked; the rest are listed as not checked and left out of the score entirely, so a result from two engines is never presented as if it came from five.

We are not mentioned anywhere. Where do we start?

With the facts, in a form a machine can retrieve: what you specialise in, which places you serve, and structured data that states it unambiguously. Then corroboration - being listed and described somewhere other than your own site. The AI & LLM Visibility Checker on this site covers whether models can read you at all, which is the prerequisite for this one.

Is my business name stored?

No. It is sent to the engines being asked, used to read their answers, and discarded when the response is returned. Nothing about the query is written to a database or a log.

Can you make us the recommended answer?

Nobody can promise that, and anyone who does is selling you something. What can be done is the work that makes it likelier: a specific, defensible claim; the structured facts behind it; and corroboration off your own site. We will tell you on a call whether your category is winnable or already crowded.

Other free tools

SEO & Speed Audit

Real Core Web Vitals from Google's API, plus live on-page checks against your actual HTML.

Tell us what you are trying to grow.

One team. Three offices. Twenty-six languages. Send us the problem and you get a senior answer — not a sales script.

Get a proposal → Message us now

A written proposal within one business day, in your language.