What is an AI visibility checker? What it tests and what it cannot

What does an AI visibility checker actually do?

It runs a fixed set of tests and returns a score or a list of findings. Technical checkers look at crawler access, sitemaps, llms.txt, metadata and structured data. Brand checkers ask an assistant a question about your category and report whether you were named.

The two types answer different questions. A technical checker tells you whether an assistant is allowed to fetch your pages and whether those pages are easy to read. A brand checker tells you what an assistant said on one occasion about one question.

Many tools called visibility checkers are really one of the two. Knowing which you are using stops you drawing the wrong conclusion from the result.

What does a technical AI readiness check test?

It fetches your homepage, robots.txt, llms.txt and sitemap and checks what an AI crawler would see: whether the main AI crawlers are allowed, whether the page can be indexed, and whether it has a clear title, description and structured data.

• Crawler access

  Whether robots.txt blocks search and browsing agents such as OAI-SearchBot, PerplexityBot or Claude-SearchBot, which fetch pages to answer a question.

• Indexability

  Whether the page returns a normal status, is not marked noindex and has a canonical address.

• Machine-readable basics

  A descriptive title, a meta description, headings in order and structured data such as FAQ or Organization markup.

• llms.txt and sitemap

  A plain-text summary of the site for assistants and a sitemap that lists the pages you want found.

What can a free AI visibility checker not tell you?

It cannot tell you whether assistants recommend your brand. Recommendation depends on the answers an assistant generates, which change between runs, between engines and between phrasings of the same question. A technical score only says nothing is in the way.

A site can score perfectly on readiness and still never be named, because assistants decide what to say from the sources and training data behind them. The reverse also happens: a brand with a weak site can be named often because other websites talk about it.

If you need to know whether you are named, you have to ask the questions your customers ask and record the answers.

How do you check whether ChatGPT names your brand?

Write ten to twenty questions a customer would ask, run each one in the assistant, and note whether your brand appears, in what position and how it is described. Repeat on a different day, because the answers vary. One run is an anecdote.

Keep the question wording fixed so you can compare runs. Record the date, the engine and the full answer, not just a yes or no.

Doing this by hand works for a first look. It stops being practical once you need several engines, several markets or a weekly trend, which is where a tracker that runs the same questions on a schedule earns its place.

How should you read the result of an AI visibility check?

Treat a readiness result as a to-do list and a brand-mention result as a rate. Fix every blocked crawler and missing basic first. Then ask whether you were named in most runs of a question, not whether you were named once.

A single appearance can be luck. A pattern across repeated runs and several engines is a signal. Look at position and wording as well: being named fifth with a lukewarm description is different from being named first.

Finally, compare against competitors for the same questions. Your visibility only has meaning relative to who else is named.