On this page
In BrightLocal's 2026 survey of 1,002 US adults, 45% said they use ChatGPT or another AI tool for local business recommendations, up from 6% a year before. "AI visibility" is how often those answers name a business. Some people call the work of improving it generative engine optimization (GEO) or answer engine optimization (AEO). Whatever the name, measuring it works differently from measuring a Google ranking, and most of the mistakes come from treating it the same way.
What AI visibility means
For a local business, AI visibility is the share of answers, to the questions its customers actually ask, that name the business. This guide covers four: Google's AI Overviews at the top of search results, Google's AI Mode, ChatGPT and Gemini. Each one is measured on its own, because they read different sources and rarely agree, as our guide to how ChatGPT, Gemini and Google's AI choose local businesses explains.
Why there is no AI ranking to check
A Google map ranking is a position: third, fifth, twelfth. AI answers don't hold still long enough for a position to mean much. Rand Fishkin of SparkToro and Patrick O'Donnell of Gumshoe.ai, a company that sells AI tracking, ran 2,961 prompts through ChatGPT, Claude and Google's AI Overviews, with AI Mode where no Overview appeared, using hundreds of volunteers. Each prompt was run 60 to 100 times per platform. Search Engine Journal's report of the study gives the result: less than a 1 in 100 chance that ChatGPT or Google's AI would return the same list of brands twice, and the same list in the same order less than 0.1% of the time.
What was steadier was how often a brand appeared across many runs. That share was more consistent than any position, the researchers found, especially in tight categories, where the top brands appeared in most answers.
Google's own documentation explains part of why. AI Overviews and AI Mode may use different models and techniques, so the responses and links they show will vary. Both may use what Google calls "query fan-out": issuing several related searches across subtopics and data sources to build one answer. And AI Overviews only appear when Google's systems judge them useful, so they often don't show at all.
What Google says about getting in
Google's page on AI features is short on tricks. It says the best practices for SEO remain relevant, and that there are no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary. To be shown as a supporting link, a page must be indexed and eligible to be shown in Google Search with a snippet. For a local business, the studies in our guide to how AI assistants choose local businesses found the Business Profile among the main sources for Google's own AIs, so a complete and correct profile matters here too.
Why Search Console can't tell you
A natural first step is to look in Google Search Console. Google says sites appearing in AI Overviews and AI Mode are included in the overall search traffic there, in the Performance report under the "Web" search type. That means they are counted together with everything else, not shown apart. And Search Console counts your website's own appearances and clicks: an AI answer that names your business without linking to your site isn't part of that count. To know how often the AIs name a business, you have to ask them.
Which questions to ask
Measuring starts with the questions, and people ask an AI differently from how they type a search. Google's report on its AI Mode in the US, published in May 2026 from a year of its search data, says people now put the whole of what they need into longer questions: the average AI Mode search is triple the length of a traditional one. "I" ranks third among the first words of AI Mode searches, after "What" and "How". Searches beginning with "which" grew 40% faster than AI Mode searches overall in the six months before the report, "which of" and "which one" most of all, which Google reads as people using AI Mode to help decide. Searches starting with "where to" and "where should I" are growing too, and among the things people ask with them, repairing a car comes first.
Google's figures on words and phrases come from a sample of AI Mode searches from May 2025 to April 2026, as shares of all of them rather than counts, and its AI Mode data is not on the public Google Trends site, so they can't be checked from outside. They describe Google's AI Mode in the US; ChatGPT and Gemini may be asked in other ways.
What people do after the answer matters too. SOCi, a local marketing company for multi-location brands, surveyed more than 1,000 US consumers for its 2026 Local Discovery Index: 52% had used AI to find a local business in the last 30 days, against 9% in 2025. After an AI tool recommends a business, the most common next step is checking its reviews (33%), and 19% visit or contact the business directly.
Together, that gives four kinds of question worth asking about a local business, each in the words a customer would use, with the town in it:
- The whole need, in the first person: "I need a dentist in Austin, TX, who should I call and why?"
- Which one to pick: "Which dentist in Austin, TX has the best reviews?"
- Where to go: "Where should I go to find a good dentist in Austin, TX?"
- The business by name, the way people check an AI's pick: "What do people say about [the business] in Austin, TX?"
The first three are searches: count how often each AI names the business. The last one names the business itself, so it can't show how often the AIs bring it up; read it for what they say about the business and which sites they cite.
On a client's AI tab, on maps of 7×7 and up, Cornerlens suggests one question of each kind, written with the client's keyword, town and name, and asks the ones you add alongside the client's AI answers.
How to measure it by hand
- Write down the questions customers ask, in several wordings. The SparkToro study asked 142 people to write their own prompt for the same need and found almost no two alike, so one wording is not enough.
- Ask each AI separately, the way a customer in the business's town would, with the town in the question.
- Ask each question several times, on different days, and save every answer.
- For each AI, count the share of answers that name the business. That share is the number to keep.
- Note who was named instead, and which websites the answers cited. Those are the sources worth checking for missing or wrong facts.
- Repeat on a schedule with the same questions, and compare the shares over weeks, never one answer with another.
The study's authors left two questions open: how many runs are needed for a reliable reading, and whether questions sent through software vary in the same way as questions typed by hand. Read any single week's share, by hand or by tool, with that in mind.
What to do with the numbers
A low share is a prompt to check facts, not to rewrite a website for robots. Look at the sites the answers cited and at the businesses they named: if a directory the AIs rely on lists the business wrongly, or not at all, that is something to fix. Keep the Business Profile complete and correct. Then keep measuring the same way, and look at the direction over several weeks.
Cornerlens does this asking on a schedule. On maps of 7×7 and up, it asks Google AI Overviews, Google AI Mode, ChatGPT and Gemini about a client's keywords, and shows how often each one named the client, where it placed them, the answers themselves and the sites they cited.
Sources
- Google Search Central, AI features and your website, read October 19, 2026.
- Matt G. Southern, Search Engine Journal, AI recommendations change with nearly every query: SparkToro, January 30, 2026.
- BrightLocal, Local Consumer Review Survey 2026, February 11, 2026, updated September 14, 2026.
- Shivani Mohan, Google, How AI Mode is changing the way people search in the U.S., May 19, 2026.
- Google Search, How people are using AI Mode in the U.S. (PDF, pages 4, 7, 10, 11, 14 and 27), read October 20, 2026.
- SOCi, 2026 Local Discovery Index: The Verification Loop (PDF, pages 4, 7 and 15), read October 20, 2026.
