Sep 18, 2026 · AI Visibility

Local AI Visibility Study: Why Google Maps Rankings Don’t Translate to AI Search Results

Local AI visibility study illustrated by a hiker and dog overlooking a mountain lake

Local businesses can track their visibility inside AI search engines the same way they track Google Maps, and a new analysis of 200,000 AI-driven local queries shows which platforms mention which businesses most often, how stable those mentions are across repeat searches, and how a strong Maps ranking often fails to carry over into AI answers. The findings give local business owners a practical baseline for measuring whether AI engines are actually recommending them.

What the study measured

Researchers ran 200,085 non-branded local searches across three AI platforms, ChatGPT, Google AI Mode, and Google AI Overviews, across 1,300 business locations. Each prompt was one a local business could realistically be recommended for (think “emergency dentist open late in Brooklyn tonight”), and every prompt was re-run several times over a 60-day window and from multiple points on a map to measure how much the answers changed. In total the study logged 1.9 million AI citations.

Three ideas kept coming up in the analysis:

Variance: how much the set of recommended businesses changes between runs of the same prompt.
Persistence: how often a business that appears once appears again on a repeat run.
Overlap: how much two platforms agree on which businesses to name.

AI answers change a lot, and businesses reappear only about half the time

Between repeated runs of the same prompt, only 20% to 33% of the businesses named matched across result sets. ChatGPT averaged 23% overlap, AI Mode 21%, and AI Overviews 33%. A business that showed up once was seen again on roughly half of subsequent attempts, 50% persistence for ChatGPT and AI Mode, and 58% for AI Overviews. Between 71% and 80% of businesses appeared in half the attempts or fewer, meaning inconsistent inclusion is the norm rather than the exception.

The variance is lower than general AI shopping searches (a separate January 2026 SparkToro study found brands repeated less than 1% of the time on broad product prompts), because the local pool of competitors is much smaller. The result still adds up to a moving target: a single business is likely to appear in two of four runs of the same prompt, but the full set of businesses named in those four answers will usually be different, and the number of businesses named changes too.

An illustrative prompt (“best pizza place in Manhattan for a tourist who only has time to try one pizza”) returned a different shortlist on each of four ChatGPT runs, with 43% similarity and 50% persistence on average. A few businesses appeared every time, while others appeared once and disappeared.

How many businesses does each platform actually recommend?

ChatGPT named the most businesses per response, Google AI Overviews the fewest:

ChatGPT: average 4.1 businesses per response, range 1 to 7, median 4.
Google AI Mode: average 3.5, range 0 to 6, median 4.
Google AI Overviews: average 2.5, range 0 to 4, median 3.

Empty responses were not rare: 10% of ChatGPT queries returned no businesses, and 11% of both Google surfaces returned none. The fluctuation in how many businesses each response includes is one of the main reasons there is no huge consistency between answers.

Google Maps rankings don’t carry over into AI

This is the headline finding for any local business owner relying on a strong Maps presence. Google Maps mentioned a tracked business on average 66% of searches, compared with 32% to 38% for AI surfaces (AI Overviews 38%, ChatGPT 33%, AI Mode 32%). Maps was also far more stable, with 99% agreement across repeat attempts versus 91% to 96% for AI platforms.

The AI platforms also disagree with each other. Google’s AI Overviews and AI Mode share only 29.3% of the businesses they name, despite both relying heavily on Google Business Profile as a source. ChatGPT overlaps even less, sharing just 20% of businesses with AI Overviews and 19% with AI Mode, because ChatGPT leans on Yelp and Bing rather than Google’s own listings.

The reason AI behaves differently comes down to how it answers. A traditional Local Pack runs one keyword-style local query. An AI engine runs multiple fan-out queries behind the scenes, breaks the prompt into several related searches, gathers answers, and blends them. So a green grid of Map rankings across a city is no guarantee that any specific AI engine will pick your business.

Moving across a city changes the answers

Running the same prompt from different points in the same town produced 36.2% overlap on ChatGPT, 46.9% on AI Mode, and 47.4% on AI Overviews. Only 3.5% of ChatGPT results were present at every grid point, compared with 8.8% for AI Mode and 25.0% for AI Overviews.

“Near me” is still a real signal. Roughly 53% to 59% of AI picks fell within 5 km of the searcher, and 72% to 78% within 10 km. ChatGPT’s circle of influence stretched far wider than Google’s: the radius covering 90% of its results reached 287 km, compared with 64 km for AI Mode and 38 km for AI Overviews. Google Business Profile made up 28.5% of all citations across the three platforms, and Yelp showed up across all three as well, with Facebook mattering more for Google’s surfaces than for ChatGPT.

What this means for a local business

A few practical shifts follow from the data:

Track AI visibility separately from Maps. Showing up in one of every two AI searches is workable. Showing up in one of every ten is a different problem, and only measurement will tell the difference.
Watch who else is being named. Recurring competitors in AI answers are the ones putting in the work. Use repeated AI queries to spot them, then close the gap with reviews, directory listings, and cleaner Google Business Profile categories.
Keep NAP consistent and address visible. Proximity still drives AI answers, so a clear address on the site and matching name, address, and phone details across listings directly affect whether a business is in range of those 5 km and 10 km cutoffs.
Build strong location pages. Because AI engines pull from real business websites more than many owners expect (93% of unique domains cited were business sites, making up 42% of all citations), well-written, location-specific pages give AI more to work with.

Rank position across a service area is what a geo grid report shows, and SEOScanPro’s GEO Grids map that out town by town so a business can see exactly where it shows up in Maps and where it does not, separately from how AI engines see it.

FAQ

Do Google Maps rankings affect AI search visibility?

Not directly. In this study Google Maps mentioned a tracked business on 66% of searches, while AI surfaces mentioned it on only 32% to 38% of searches. A strong Maps presence is no guarantee that ChatGPT, AI Mode, or AI Overviews will name the same business.

How often do AI platforms recommend the same businesses on repeat searches?

About half the time. A business that appeared in one run was seen again on roughly 50% of repeat runs for ChatGPT and AI Mode, and 58% for AI Overviews. Only 20% to 33% of the full set of named businesses overlapped between runs.

Which AI platform names the most local businesses?

ChatGPT named the most, averaging 4.1 businesses per response. Google AI Mode averaged 3.5 and Google AI Overviews averaged 2.5. ChatGPT also had the widest geographic spread, with 90% of its picks falling within a 287 km radius, compared with 64 km for AI Mode and 38 km for AI Overviews.

Related coverage

This article summarizes reporting from brightlocal.com.

← Back to the blog