How AI Assistants Recommend Local Businesses

How do AI assistants decide which local businesses to recommend?

They look the businesses up. When someone asks for “an emergency plumber near me” or “a quiet Italian place for a birthday dinner in Pasadena”, the assistant searches business data: listings, ratings, reviews and websites. It filters the results by location and fit, then writes a short list with reasons. The recommendation is assembled at the moment of the question from sources you can see and, mostly, correct.

The catch is that each assistant reads a different mix. Google’s AI features draw heavily on Google Business Profiles. ChatGPT now uses licensed Yelp content. Microsoft documents Bing as the search source for Copilot in work accounts, and Siri and Apple Maps show Apple’s own place cards. So the same question can name different businesses in different assistants, and a listing you have ignored for years can decide whether you appear in one of them.

This matters because customers have moved fast. BrightLocal’s survey of 1,002 US adults, published in March 2026, found that 45% had used AI tools for local business recommendations, up from 6% the year before. Adults aged 30 to 44 led at 64%.

Where does each AI assistant get its local business data?

From the source its owner controls or licenses, plus the open web. The table below lists the main local source for each assistant, based on what the platforms document and what independent studies have measured. No platform publishes its full local recipe, so treat it as a map of where to spend effort, not a wiring diagram.

Google AI Overviews and AI Mode: Business Profile first

Google says there are “no additional requirements to appear in AI Overviews or AI Mode,” and its short list of worthwhile practices includes checking that your Business Profile information is up to date. Both features may use what Google calls query fan-out: “issuing multiple related searches across subtopics and data sources” to build one answer.

The data shows how much weight the profile carries. In BrightLocal’s study, Google Business Profile accounted for about 67% of the AI Overview citations in its tables and about 53% of AI Mode’s (our arithmetic: 108,018 of 161,094, and 455,123 of 861,263). AI Mode also read far more widely, citing 81,817 different domains against 8,541 for AI Overviews. That breadth is where your website, local news coverage and industry directories come in.

Gemini: grounded in Google Maps

Google’s Gemini help pages say the Gemini app automatically uses public information from Google services, including Google Maps (Keep Activity must be on). In practice, your Business Profile is your Gemini listing. SOCi’s 2026 Local Visibility Index observed that Gemini, “grounded in Google Maps as a primary source,” was more accurate and consistent than other assistants.

ChatGPT: Yelp, then the web

ChatGPT is the outlier, and the one most businesses misjudge. Yelp’s letter to shareholders of August 6, 2026 says “Yelp ratings and reviews recently began powering ChatGPT’s local experience in relevant categories,” and that its Request-a-Quote integration with ChatGPT was coming soon. BrightLocal’s data matches: Yelp appeared as a source in 80% of ChatGPT’s local answers and made up about 61% of the ChatGPT citations in its table (our arithmetic: 99,281 of 162,966). Google Business Profile was not among ChatGPT’s top sources; Apple’s listings were, though cited only 600 times.

Your own website still counts, if OpenAI can reach it. OpenAI says sites that opt out of its OAI-SearchBot crawler “will not be shown in ChatGPT search answers.” Our guide to getting your business mentioned in ChatGPT covers the non-local side.

Copilot: whatever Bing knows

Microsoft documents this for Copilot in work accounts (Microsoft Copilot and Microsoft Copilot Chat for organizations). Its page, updated October 29, 2026, says Copilot generates a search query “that it sends to the Bing search service” and passes along the user’s IP address, which Bing uses to localize results. We found no Microsoft documentation giving the same detail for the consumer Copilot app. Our working assumption is that Microsoft’s own search engine sits behind it too, so your Copilot visibility rests on your Bing Places listing and on Bing indexing your site. Setup is covered in our guide to Bing Places and Apple Business.

Perplexity: its own index, no local documentation

Perplexity runs its own index, which it says tracks over 200 billion URLs, and gives priority to authoritative domains. It has published nothing specific about how it sources local results, and the two big studies used here did not measure it. Our working assumption: it behaves like a web search engine, so your website and the major review sites carry the weight.

Siri and Apple Maps: Apple Business

Apple replaced Apple Business Connect with Apple Business on April 14, 2026, and moved existing data over automatically. It is free, and it controls your place card (photos, location details and hours) across Apple Maps, Safari, Spotlight and more; Apple also names Siri among the places businesses can reach customers. Apple announced Maps ads for the US and Canada, clearly marked, which you can only buy after claiming your location.

What makes an assistant pick one business over another?

No AI platform publishes its local ranking rules, so the best documented guide is Google’s own: local results are “mainly based on relevance, distance, and popularity.” The section heading on the same page still calls the third factor prominence, defined as how well-known a business is, based on links to it and how many reviews it has. Studies of AI answers add one more pattern: ratings act as a gate.

Relevance: does your information match the question?

Google says businesses with complete and accurate information are more likely to show up in local results. AI answers take this further because questions are longer: “a pediatric dentist that takes Delta Dental and is open Saturdays” can only be matched to a business whose profile, reviews or website state those facts. Fan-out turns one question into several searches, and each sub-search needs something to find.

BrightLocal found that around 93% of the domains cited in local AI answers were individual business websites. That counts domains, not citations: the big platforms still earn most of the links. But it shows assistants do read business sites, usually for the specific detail a listing lacks.

Distance: where the question is asked

Location shapes the answer. Microsoft says Copilot in work accounts sends the user’s IP address so Bing can localize results, and Google’s local ranking page says it uses what it knows about a searcher’s location when none is shared. BrightLocal ran each prompt from nine map points around every business it tracked. A plumber recommended on the north side of town may not appear from the south side.

Prominence: reviews as a gate, not just a score

Google says “more reviews and positive ratings can help your business’s local ranking.” SOCi’s 2026 index suggests AI answers are stricter. It ranked 171 enterprise retail brands by asking ChatGPT, Gemini and Perplexity to recommend ten businesses of a given type in each market. It reported that ChatGPT “has a high bar of 4.3 stars for inclusion,” and that recommended businesses averaged 4.2 stars on Perplexity and 3.9 on Gemini. SOCi also found only 45% overlap between the brands most visible in traditional local search and those most visible in AI recommendations.

Treat those figures as direction, not thresholds: the sample was retail chains, and SOCi’s exact scoring is not fully published. Our view: an assistant listing three to five businesses has little reason to include a 3.6-star option when 4.6-star ones exist nearby.

Which sites matter most in your industry?

Google Business Profile and Yelp lead almost everywhere, but the third and fourth sources change by industry. BrightLocal also listed the most-cited sources for selected industries, without saying which assistants each table covers.

• Restaurants: four sources made up 80% of citations: Google Business Profile 45.2%, Yelp 21.8%, Tripadvisor 7% and OpenTable 6.6%.
• Car repair and vets: Yelp was strong, at 19.4% and 19.5%. Facebook added 4.3% for vets.
• Dentists: Yelp 15.4%, with Zocdoc at 1%.
• Plumbers: Yelp 13.8%.
• Lawyers: Yelp 7.9%, then Justia at 3.2%, a sign that legal directories still matter.

Three caveats. About 20% of the locations were restaurants, which BrightLocal says may skew some directory figures; Gemini, Copilot and Perplexity were not studied; and the vet table’s counts exceed its stated total. Still, the practical move is simple: ask ChatGPT and AI Mode for your service in your town, click the cited sources, and treat the third-party domains you see as your listing priorities. Our guide to local citations covers which directories are still worth claiming and how to clean them up.

How accurate are AI answers about local businesses?

Accurate enough to be trusted, wrong often enough to cost you customers. SOCi’s 2026 index found that ChatGPT, Perplexity and other non-Google assistants were “only about 68% accurate when providing business contact info.” SOCi does not define the measurement in detail, so take the number as a warning light rather than a benchmark. Its explanation is credible, though: those assistants combine several sources into one answer, and when the sources disagree, the answer can pick the wrong one.

Google adds its own moving parts. Its Business Profile help says Google “gathers info from different sources, like user reports and licensed content,” and may update your profile when those sources say your details are wrong. When a suggested edit needs your review, you have 4 days to accept or reject it. If you don’t respond, Google may publish it when other public information, such as your business website, supports it.

Customers do check. Among the AI users in BrightLocal’s survey, 63% said they trust AI recommendations, but 97% sometimes double-check them against real reviews and 88% fact-check the reviews AI tools cite. A wrong phone number in ChatGPT and a different one on Google is the kind of mismatch that sends a cautious customer to the next name on the list.

The errors we see most

• Hours that never got updated for a schedule change or holidays, on one platform but not the others.
• An old address still live on Yelp or a data aggregator years after a move.
• A call-tracking number used on one directory that now competes with the real number.
• A closed second location that was never marked closed, so assistants still send people there.
• Discontinued services still listed on an old website page, which an assistant then repeats.

How do you fix wrong information in an AI answer?

Fix the source, not the answer. We found no documented way for a business to edit what ChatGPT, Gemini or Copilot says about it directly. What you can change is the listings and pages those assistants read, then give them time to re-read.

1. Record the wrong answerNote the question, the assistant, the date, roughly where you asked from, and every source the answer links to.
2. Trace it to a sourceOpen the cited links first. If none are shown, check the main source for that assistant: Business Profile for Google, Yelp for ChatGPT, Bing Places for Copilot, Apple Business for Siri.
3. Correct the listings you ownUpdate Google Business Profile, Yelp, Bing Places and Apple Business so name, address, phone, hours and categories match exactly. Review any pending Google updates within the 4-day window.
4. Make your website the referencePut the same details in plain HTML text on your contact and location pages, not only in an image or map widget. If you use LocalBusiness structured data, make sure it matches the visible text, as Google asks.
5. Clean up third-party copiesFix or remove duplicates and outdated entries on directories and data aggregators, starting with the domains assistants cite for your industry.
6. Check crawler accessMake sure robots.txt and your firewall let in Googlebot, Bing’s crawler and OAI-SearchBot. OpenAI says robots.txt changes take about 24 hours to register.
7. Re-test in two to four weeksAsk the same question again. If the error persists, look for a source you missed; one stale directory can keep it alive.

What should a local business do first?

Start with eligibility on the platforms that feed the assistants, then credibility, then specificity. Our team brings over 20 years of digital marketing experience and has served 500+ brands, and the local businesses missing from AI answers usually have the same problems that keep them out of the map pack. Most of the fixes are ordinary local SEO, covered step by step in our local SEO guide.

Weeks 1 and 2: the four core listings

Claim and verify Google Business Profile, Yelp, Bing Places and Apple Business. Google says verification makes a business more likely to show up in search results. Use one exact name, address, phone and website everywhere, matching categories, current hours (including special hours), real photos and a description that names your actual services and areas. Detailed setup is in our Business Profile guide.

Weeks 3 to 6: a website that answers the follow-up questions

In audits we often find location and service pages that are not indexed, or a dozen city pages that swap only the town name and compete with each other. Neither gives an assistant anything specific to quote. Picture a three-location physical therapy clinic whose insurance list sits in a PDF and whose hours appear only in an image. An assistant asked for “a physical therapist in Round Rock who takes Medicare” has no text to match.

Give each location one strong page with its address, hours, services, insurance or payment options, parking and neighborhoods served, all in HTML text. Add LocalBusiness markup that matches what the page shows. Then confirm the pages are indexed in Google and Bing.

Ongoing: reviews with detail

Ask every satisfied customer for a review on the platform that matters most for your category, and reply to reviews; Google says replying shows you value feedback. Reviews that mention the service, the neighborhood and the outcome give assistants the specifics that match long questions.

How do you track whether AI assistants recommend you?

By asking them, the same way every month, and logging what comes back. No assistant gives businesses a full local visibility report, so a fixed question set is the most reliable baseline you can build.

• Write 15 to 25 real questions. Combine your services, your towns or neighborhoods and the qualifiers customers use (“open Sunday”, “takes Medicare”, “free estimates”). Add brand checks such as “Is [your business] open on Saturdays?”
• Ask from where your customers are. Location changes the answer, so test from your service area, not only from the office.
• Log four things per answer: whether you were named, your position in the list, the sources cited and whether your details were correct.
• Read trends, not single runs. Answers vary between runs, so judge progress over several months.
• Connect it to results. Track AI referral visits in your analytics, Business Profile calls and direction requests, and ask new customers how they found you.

Our guide to tracking brand visibility in AI answers goes deeper on methods and tools. If you would rather have it done, our plans include AI question tracking: 25 questions on Starter (Google AI and ChatGPT) and 75 on Growth (adding Perplexity, Gemini and Copilot), alongside Business Profile and Bing Places work. To start with the website side, our free SEO + GEO report analyzes your website, including local and trust signals such as NAP, click-to-call and business schema, and which AI crawlers can read it. You get password-protected dashboard access within 24 hours.

Common questions

Sources

1. BrightLocal (August 26, 2026): Top sources and directories for local AI search
2. BrightLocal (March 10, 2026): Nearly half of consumers are asking AI for business recommendations
3. SOCi (February 12, 2026): The 2026 Local Visibility Index, SOCi100 (PDF)
4. Yelp Inc. Q2 2026 Letter to Shareholders (August 6, 2026), SEC filing
5. Google Business Profile Help: How to improve your local ranking on Google
6. Google Business Profile Help: Understand Google updates on your Business Profile
7. Google Search Central: AI features and your website
8. Gemini Apps Help: Use and manage Connected Apps in Gemini
9. Microsoft Learn (updated October 29, 2026): Data, privacy, and security for web search in Microsoft Copilot and Microsoft Copilot Chat
10. Perplexity (September 25, 2025): Architecting and evaluating an AI-first search API
11. Apple Newsroom (March 2026): Introducing Apple Business
12. OpenAI: Overview of OpenAI crawlers