Skip to content

Does ChatGPT recommend your business? How to check AI visibility by location

Ask an AI assistant for a dentist near you and it names three or four. Whether you are one of them depends on where the question was asked from.

Ask ChatGPT for a dentist near you and it names three or four. Whether your business is among them depends on where the question was asked from, how it was worded, and on the run itself, because the same question asked twice can return two different sets. So the honest question is not “do we rank?” but “how often are we named, and from where?”.

How an assistant answers “dentist near me”

An AI assistant does not keep a ranked list of dentists. When it gets a local question it combines what it already knows with a live search, reads a handful of results, maps and review pages, and writes a short answer that names a few places with a line about each. The location it uses is whatever it has for you: the city you told it, the one it inferred, or none, in which case it asks or guesses.

Three things move the answer. The asker’s location, because the search it runs is local. The wording, because “dentist near me”, “a good dentist in Leeds” and “emergency dentist open now” pull different pages. And the run, because the model writes the answer fresh each time and does not always pick the same names from the same pages.

AssistantWhat it draws onWhere the answer changes
ChatGPTIts training plus live web search when it browses.The location it has for the asker, the wording, and from run to run.
GeminiGoogle Search and Google Maps.The asker’s location and the wording.
CopilotBing.The asker’s location and the wording.
PerplexityLive web search, with the sources cited in the answer.The wording, and what the cited pages say that day.
Google AI OverviewsGoogle Search results and Maps data, shown on the results page.The searcher’s location and whether Google shows an overview for that query at all.

Why the measure is how often, not a rank

On Google Maps a business is 4th or it is not, and a grid of points draws that as a map. An assistant gives three names in prose with no positions, so there is nothing to call a rank. What can be counted is whether you were named. Ask from one point ten times and you were named, say, 6 times. Ask from 25 points and the count becomes a map: named often near the practice, rarely across town, never from the next town.

That map behaves like a Maps rank that changes from street to street, because the assistant’s search is local too. It is just noisier, so one answer tells you little and a count tells you something.

How to check by hand

You can do this without any tool. Open the assistant, tell it where you are, and ask the question a customer would ask. Note the names. Change the location to the far side of your area and ask again. Repeat the whole set a few times on different days, and do the same in a second assistant. A spreadsheet with locations down the side and assistants across the top, and a tally in each cell, is the measurement.

The trouble is the time. Four assistants, 9 locations and 3 runs is 108 questions, each read and recorded by hand, and the answer is stale within weeks. Most owners do it once, see their name or do not, and never do it again. That single run is the AI equivalent of searching your own trade from your own counter: it tells you about one spot on one day.

AI Overviews and AI Mode are different

Google AI Overviews is a different kind of answer. It appears at the top of an ordinary Google Search results page, above the local pack, and it is built from Google’s own results and Maps data rather than from a separate assistant. AI Mode is a fuller version of the same thing, a chat-like tab on the search page. Both name businesses for some local queries and show nothing for others, and Google decides query by query.

Because they sit on the search page, they can be checked from a point the way the local pack is: search from the point, see whether an overview appears, see whether you are in it. That makes them the easiest AI surface to put on a grid, and the one most tied to how you already do on Google.

RankShark Track checks AI visibility this way: from the points of a grid it asks ChatGPT, Gemini, Perplexity and Google AI Overviews the question for your keyword and records whether the business is named, on the same map as the Maps rank. On Starter it asks AI Overviews and Gemini on grids up to 25 points; on Pro and up it asks every engine on grids up to 49 points, and the history runs to 25 months, so a count from this month can be set against the same count from last quarter.

What tends to make a business nameable

No formula is published for this, and the assistants change. What can be observed, by comparing the businesses that get named with the ones that do not, is a pattern rather than a rule.

  • A profile and a website that say plainly what you do and where. The assistant is reading. If the name, the category, the address and the services match across the Google profile, the website and the Bing listing, there is one clear thing to name. If they differ, there is doubt, and doubt is left out of a three-name answer.
  • Reviews that mention the service and the area. A review that says “fixed our boiler in Headingley the same day” is text an assistant can use to connect a trade to a place. A star with no words is not.
  • Being present on Bing too. Copilot draws on Bing, and some other assistants reach Bing’s index through their search partners. A business that exists on Google and not on Bing is invisible to those. The guide on Bing Maps and Bing Places covers the listing.

What not to do

Do not write pages for machines. A page that lists every suburb in a row, or repeats “dentist” forty times, reads badly to a person and gives an assistant nothing it can quote. Do not stuff the business name with keywords on the profile; Google refuses the edit and the mismatch with the website is exactly the doubt described above. And do not invent reviews. They are against every platform’s rules, they read as invented to a model trained on millions of real ones, and the risk is the listing, not the rank.

Measure before you change anything. Pick one keyword, count how often you are named from a few points, and repeat it on a schedule so the count has something to be compared with. The free audit shows the Google Maps side of the same picture for your business today.

Questions.

Does ChatGPT recommend local businesses?

Yes, when asked. For a question such as “dentist near me” it names a few businesses, drawing on live web search and whatever location it has for the asker. The names change with the location, the wording and from one run to the next.

Can I rank in ChatGPT?

Not in the way a business ranks in Google Maps. An assistant names a handful of places and does not keep a numbered list, so the useful measure is how often you are named from a given place, not a position.

How do I check whether AI assistants name my business?

Ask each one the question a customer would ask, from several locations, several times, and note the names. A tracker that asks from grid points automates this. RankShark Track asks ChatGPT, Gemini, Perplexity and Google AI Overviews from each point and records whether you are named.

Is Google AI Overviews the same as ChatGPT?

No. AI Overviews is a summary at the top of a Google Search results page, built from Google’s own results and Maps data. ChatGPT is a separate assistant that browses the web when it needs to. A business can be in one and not the other.

Read the grid. Run the journeys.