A dated baseline of what the AI assistants answer

We track whether ChatGPT, Gemini, Claude and Perplexity name your business, which competitors appear and which sources they cite, on a dated question set.

“How can I get my company mentioned in ChatGPT?”

A question owners ask.

What the AI Visibility scan measures

An AI Visibility scan tests twenty agreed customer questions across selected assistants and records what each system returned on a specific date under a documented method: whether your business is named, where it appears in the answer, which competitors appear, whether material business facts are accurate, and which sources are cited when the assistant provides them. Position describes placement inside the response; it does not say how strongly the assistant recommended you.

Six of the twenty questionsChatGPTGeminiClaudePerplexity
Physiotherapy for a running injury near me4 of 4 runs named the business0 of 4 runs named the business2 of 4 runs named the business4 of 4 runs named the business
Which clinic treats vertigo0 of 4 runs named the business0 of 4 runs named the business1 of 4 runs named the business0 of 4 runs named the business
Physiotherapy clinic open on Saturdays4 of 4 runs named the business4 of 4 runs named the business, and an answer got a fact about it wrong4 of 4 runs named the business3 of 4 runs named the business
Direct billing physiotherapy nearby0 of 4 runs named the business4 of 4 runs named the business2 of 4 runs named the business0 of 4 runs named the business
Rehabilitation after knee surgery, which clinic4 of 4 runs named the business4 of 4 runs named the business4 of 4 runs named the business0 of 4 runs named the business
Pelvic health physiotherapy0 of 4 runs named the business0 of 4 runs named the business2 of 4 runs named the business, and an answer got a fact about it wrong0 of 4 runs named the business
  • 4 of 4 named in every run
  • 2 of 4 named in some runs
  • 0 of 4 not named in any run
  • an answer got a fact about the business wrong
Illustrative exampleOne scan: the same questions to four assistants, four times each, on one dateA trade drawn for the example. The cells are what a real run records: named or not, run by run, and whether an answer got a fact wrong. Which competitors were named goes in the report.

The scan shows what the assistants said on that date; whether anyone was persuaded, visited, called, booked or paid needs its own evidence.

How customer questions are selected

Questions start from the services that matter to you, in a customer’s words, and none is built only to force your name into an answer. Buying intent separates a useful question from a vanity one: a definition of a treatment produces general education, while a question about which practices offer it nearby is closer to a provider decision. The same twenty questions stay stable across scans unless the service scope changes, because changing the prompts and the date together makes the result impossible to interpret.

Why one screenshot is not a baseline

One screenshot

  • One system
  • One answer
  • One moment
  • One set of conditions nobody recorded

It proves that one answer was once produced.

A baseline

  • The same questions
  • Across the selected systems
  • On a recorded date, with the material settings
  • The response, the named businesses, the factual errors, the visible citations

It can be run again and compared.

What a screenshot proves, and what a scan records

Variation is part of the measurement, and the baseline does not pretend generated answers are fixed rankings.

How Google and AI Visibility are compared

Each question is sorted into one of four groups, from strong on both Google and the assistants to absent from both. Google-only visibility may point toward how clearly the business is defined online, or toward outside sources. The comparison is a guide rather than a score, because a keyword grid and a conversational question are not identical tests.

Each dated run reports the mentions and omissions, the competitors named, the facts an answer got wrong, the citations shown, and what changed since the run before.

What to fix first

Come with the project you already have in mind. Before we meet, we review your website, listings and other public-facing material. During the 30-minute session, you show us how the relevant part works today. We check whether the project addresses the real constraint and what it needs to connect to. You can hire us for one part.

Frequently asked questions

Can you track whether ChatGPT recommends my business?
Yes, we can record whether ChatGPT names your business for a repeatable set of customer questions. The result is dated because generated answers can vary, and a mention should not be treated as a permanent ranking.
Which AI assistants should a local business monitor?
Monitor selected assistants that your customers are likely to use and that can answer current local questions. The useful set can change as products, retrieval systems, and customer behavior change, so the report records which systems were tested.
How many customer questions should an AI Visibility scan include?
We use twenty high-intent questions for the defined scan. That set is large enough to cover priority decision themes while remaining stable and inspectable across dated comparisons.
Why does my business appear in one AI answer but not another?
Different assistants and runs can use different models, retrieval sources, context, dates, and response methods. Small changes in phrasing can also alter which sources and businesses appear.
Can an AI mention be attributed to a booked customer?
Sometimes, but only when source data or the customer provides credible evidence connecting the booking to the AI interaction. Many journeys cross an assistant, branded search, website, call, and offline conversation, so complete attribution is not always possible.

Primarily serves the Market Visibility and Insights pillars of the MOSAIC Six.

Request a working session