Where Each AI Actually Gets Local Business Names
ChatGPT, Gemini, Perplexity and Claude do not read the same sources. Here is where each one actually looks up local businesses, with dates.

Built BakingSubs to 162,500 Copilot citations and accelerating. Now teaching the system behind it.
- local-ai-visibility
- local-service
- chatgpt
- directories
Most articles about local AI visibility tell you to fix your Google Business Profile and stop there. That advice is not wrong, it is just aimed at one engine out of four. ChatGPT, Gemini, Perplexity and Claude each look up local businesses in different places, and if you only feed one of them, the other three will keep naming somebody else in your town.
Key takeaways
- ChatGPT leans on Foursquare Places (partnership announced December 16, 2024) and Yelp (licensing deal announced July 23, 2026), falling back to Bing Places. Google Business Profile is not the primary source for this engine.
- Gemini is grounded in Google Maps, so your Google Business Profile is the main lever here and almost nowhere else.
- Perplexity leans on Yelp, TripAdvisor and industry directories for your trade. Claude leans on Brave Search results.
- In BrightLocal's February 11, 2026 survey of 1,002 US adults, consumers using AI to find local businesses jumped from 6% to 45% year over year, with ChatGPT at 31% and Google AI Mode at 23%.
- Across 9,454 question-by-business checks we ran, the business was not among the sources the answer used in 84.3% of them. Being present in Google organic results for the question was the strongest predictor of clearing that stage (odds ratio 12.74).
- A Foundation and AirOps analysis of 28 million AI responses in Q4 2025 found that 91 to 97% of AI citations come from category questions ("best transmission shop near me") rather than brand searches, so defending your shop name is not the game.
The short answer: four engines, four different sources
Each of the four main AI engines pulls local business names from a different set of places.
- ChatGPT leans on Foursquare Places and, since summer 2026, Yelp. When it needs a map fallback, it uses Bing Places.
- Gemini is grounded in Google Maps and Google's own local index. Your Google Business Profile is the source here.
- Perplexity leans on Yelp, TripAdvisor and industry directories relevant to your trade.
- Claude leans on Brave Search results, which pull from an independent web index.
So the answer to "which listing matters" is: it depends on which engine your customer is using. If you only maintain a Google Business Profile, you are only reliably feeding Gemini. The other three may name you or may not, depending on what shows up on Yelp, Foursquare, TripAdvisor, Brave and the industry directories that cover your trade.
One important limit up front. We know which platforms these engines have partnerships or grounding relationships with, and we can watch which sources they name in their answers. We cannot tell you how much weight each engine puts on each source, and anyone quoting a specific percentage share (the "60 to 70% from Foursquare" figure floating around online, for example) is guessing. The partnership is a real fact. The share is not.
Why "just fix your Google Business Profile" leaves three engines out
Here is a plausible situation. Rafael owns an independent transmission shop in Tucson. He spent a Saturday cleaning up his Google Business Profile, adding photos, and asking his last twenty customers to leave a review. Two months later, Gemini started naming him when people asked about transmission work in Tucson. That is a real result and it matters, because Google's own AI mode is one of the two biggest AI destinations for local searches right now.
But when someone asked ChatGPT the same question, the answer named two other shops. When someone asked Perplexity, it named a national chain and a shop that has been on Yelp for eight years with 340 reviews. Rafael's Google Business Profile did nothing for either of those answers, because neither engine was reading Google Maps in the first place.
The BrightLocal survey published February 11, 2026 (n=1,002 US adults) found that consumers using AI to find local businesses climbed from 6% to 45% year over year. Inside that group, ChatGPT was the tool for 31% and Google AI Mode was the tool for 23%. Perplexity, Claude, and Copilot split most of the rest. If Rafael only feeds Google, he is showing up for the 23% and missing the other 77%.
What ChatGPT is actually reading (Foursquare and Yelp)
ChatGPT's local answers pull from two named partnerships and a fallback.
On December 16, 2024, Foursquare announced a partnership with OpenAI covering its Places API. Foursquare's dataset covers more than 100 million points of interest across more than 200 countries. This is the underlying map data ChatGPT uses when it needs to know that a business exists at a given address in a given town.
On July 23, 2026, Yelp announced a licensing deal with OpenAI that brought Yelp reviews, ratings and photos into ChatGPT. On August 10, 2026, Yelp added reservations and waitlist functionality inside ChatGPT itself. So when ChatGPT names a restaurant, salon, med spa or repair shop and mentions how it is reviewed, it is often quoting Yelp.
When neither of those has what it needs, ChatGPT falls back to Bing Places, which is Microsoft's local index and the one closest to a Google Business Profile equivalent for the Bing side of the web.
The practical read for a local owner: your Foursquare listing (which most small businesses have never touched or have never even seen) and your Yelp listing are the ChatGPT surface. Your Google Business Profile is not, at least not through the direct partnerships.
Two cautions. First, do not fall for any specific percentage share of ChatGPT results attributed to Foursquare. The "60 to 70%" number is repeated everywhere and has no published methodology behind it. Different vendors quote different numbers, which is the tell. Second, avoid any "you need N reviews before AI cites you" threshold. That number is also unsupported. Recent, specific reviews on the platforms an engine actually reads are what matters. The count is not the lever.
What Gemini, Perplexity and Claude are reading
Gemini is Google's model, grounded in Google's own local index. When you ask Gemini for a plumber in Phoenix, it is essentially asking Google Maps. Your Google Business Profile, your Google reviews, and your ranking in normal Google searches for the query are what feed this engine. If Google's local pack shows your business for "plumber in Phoenix", Gemini is very likely to be aware of you. If not, it is very likely not to be.
Perplexity leans on Yelp, TripAdvisor, and industry directories that cover your trade. If you run a med spa, that means RealSelf. If you run a hotel or restaurant, TripAdvisor is central. If you run a general local service business, Yelp does most of the work. Perplexity also cites its sources visibly at the bottom of its answers, which is useful for checking what it actually read to arrive at a name.
Claude uses Brave Search as its main web-search partner. Brave has its own independent web index, separate from Google and Bing, and it weighs review platforms, directories and the open web. If you are named in local best-of roundups, on your trade's industry directories, or in local news coverage, that is what tends to show up in Claude's answers. There is no Claude equivalent of a Google Business Profile you can "fix" directly. You get there by being covered elsewhere.
A concrete example of the difference. A Foundation and AirOps analysis of 28 million AI responses in Q4 2025 found that Yelp received 512,680 citations across those responses, roughly 3.4 times the next platform. Yelp is doing outsized work for both ChatGPT and Perplexity. If your Yelp listing is thin, has old reviews, or has no photos, two of the four major engines are looking at that thin profile when they decide whether to name you.
What our research says about the biggest lever across all four engines
We ran a study of 9,454 question-by-business observations, checking for each specific business whether it appeared among the sources a given AI answer used. In 84.3% of those checks, the business was not among the sources the answer used. 94% of the observed failure happened before any selection or ranking step. In other words, the engine did not have the business in its working set to begin with.
Small language reset here, because it matters. We do not say "the AI did not find you". We can see an engine's returned bibliography, we cannot see everything it retrieved and then discarded. The honest wording is "not among the sources this answer used".
The strongest predictor of clearing that first stage was Google organic presence for the question the person asked (odds ratio 12.74, McFadden R-squared 0.24). We tested nine technical on-page indicators. None of them survived Bonferroni correction. Which means the boring answer is the right one: businesses that already show up in normal Google search for a question tend to be present in the sources AI engines pull from for that question, and that pattern held whether the answer came from ChatGPT, Perplexity, Gemini or Claude.
Two honest caveats we would rather write than hide. This is an association, not proof of cause. We cannot tell you that fixing your Google presence will cause an AI citation, and anyone who tells you they can is guessing. And the finding says nothing about which listing feeds which engine, which is why the table above still matters. Google presence sets the floor across all four engines. Foursquare, Yelp, TripAdvisor, Brave and Bing Places determine which specific engine names you on which specific question.
The category-question shift, and why your business name barely matters
Here is the second thing almost nobody writes about correctly. The Foundation and AirOps analysis of 28 million AI responses found that 91 to 97% of AI citations came from category questions rather than brand searches.
A category question is "best transmission shop in Tucson", "24-hour plumber near me", "med spa with morpheus8 in Charlotte". A brand search is "Rafael's Transmission Tucson". Customers researching in AI are almost always asking the category question, because if they already knew your name they would have called you.
The consequence for a local owner is direct. Defending your brand term in AI (making sure your name pulls up your business) is almost irrelevant. The buyer is not asking. What matters is being one of the three to five businesses named when the buyer asks the category question for your town and your job. That is a very different task from "showing up when someone Googles your business name".
What this means for what you should actually check
Six things, in this order. Do them yourself, one afternoon, on a laptop.
- Search Google for "best [your trade] in [your town]" and note which businesses show up in the local pack and the first-page results. If you are not on that page, that is the biggest single gap, and it is the one our research says predicts everything else.
- Check your Google Business Profile: category correct, address and hours right, at least a dozen recent reviews, photos from the last six months. This is your Gemini surface.
- Look up your business on Yelp. If you do not have a listing, claim one. If you do, check whether it has photos, whether the category is correct, and whether the reviews are more than two years old. This is a shared surface for ChatGPT and Perplexity.
- Look up your business on Foursquare (foursquare.com, not the old Swarm app). Most local owners have never seen their own listing. Claim it. Correct the address, phone, hours and category. This feeds ChatGPT.
- Check the industry directories for your trade. Dentists have Healthgrades. Contractors have Angi and BuildZoom. Med spas have RealSelf. Restaurants have TripAdvisor. Auto shops have RepairPal and RepairPro. Perplexity and Claude both weigh these.
- Search "best [your trade] in [your town]" and see if any local best-of roundups, local news pieces, or community lists come up. If your name is not on them, that is a slow project, but it is the main way you get into Claude's answers.
A note on the word "citation"
If you have read anything about local SEO before, "citation" meant a directory listing of your business name, address and phone number on sites like Yelp, Yellow Pages, or Angi. In AI search, "citation" means an engine naming your business in its answer and linking to your page. Same word, two different things. Both matter for a local business, and unfortunately they got the same name.
Frequently asked questions
Which AI matters most for a local business in 2026?
ChatGPT and Google AI Mode do the most work in local searches right now, based on BrightLocal's February 11, 2026 survey (ChatGPT 31%, Google AI Mode 23% of AI-using consumers). Perplexity and Claude are smaller but growing, and they read different sources than either of the top two. If you can only work on one, work on Google presence for your category question in your town, because that is the strongest predictor of clearing the first stage across all four engines.
If I fix my Foursquare listing, will ChatGPT start recommending my shop?
We cannot promise that, and neither can anyone else. What we can say is that Foursquare's partnership with OpenAI (announced December 16, 2024) makes Foursquare Places one of the two named data sources ChatGPT uses for local businesses. Being on Foursquare with correct information puts you in the set the engine reads. It does not guarantee the engine will choose to name you on any specific question.
Why did the article not tell me a review-count threshold?
Because there is not a real one that has been published with a method behind it. You will see "150+ reviews before AI cites you" or similar in a lot of blog posts. None of it is sourced. Recency and specificity in reviews tend to matter more than raw count, and the platform the reviews live on matters as much as the reviews themselves. A dozen recent, specific reviews on Yelp is worth more to ChatGPT and Perplexity than a hundred old ones on a directory neither engine reads.
Do I need to do all of this myself?
You can start with the free AI Visibility Check, which will show you where your business currently stands across the four engines. If you want the full walkthrough of the six local signals with checklists and examples, that is The AI Citation Playbook at $27, which the visibility check will route you to if it fits. There is no rush to buy anything. The afternoon audit above is enough to see the shape of your gap.
What if I want to go deeper on the Google side specifically?
Google's local surface is its own topic. Google Business Profile in 2026 covers what Gemini and Google AI Mode read from that profile and what they ignore. For the broader picture of how local search shifted when AI started answering questions directly, Local SEO in 2026 has the fuller version.
Where to go from here
Spend an afternoon this week checking your listings on all four surfaces: Google Business Profile, Yelp, Foursquare, and your trade's main industry directory. Most local owners have never seen their Foursquare listing, and most Yelp listings have not been touched in years. Fixing what is stale is the fastest useful move you can make. When you are done, run the free AI Visibility Check to see which of the four engines currently name you when someone asks the category question for your town, and which do not.