How to Get ChatGPT (and Other AI) to Recommend Your Business

To get ChatGPT to recommend your business, get onto the lists it reads and keep a review profile that holds up next to the businesses already on them. In the Cheers Market Baseline, 84% of ChatGPT's answers to 32 home-services questions in 100 US and Canadian metros, collected between August 31 and September 20, 2026, cited at least one directory, review platform or "best of" list; Expertise.com alone appeared in 32% of answers, and Yelp in under 1%. Google AI Mode works from a different pile, citing businesses' own websites in 92.5% of its answers, so for ChatGPT the first place to look is off your website.

Last verified: September 23, 2026.

This page is the practical summary of what we found when we asked four engines the same questions every week for three weeks. Each section gives the number that matters, what to do about it and the guide that goes deeper. The six findings below are the short version, and the free AI Visibility Grader shows where one of your businesses stands today across ChatGPT, Gemini and Perplexity. For all four engines side by side, factor by factor, see AI local ranking factors for 2026.

Get onto the lists ChatGPT reads

OpenAI's documentation says web search lets its models "provide answers with sourced citations", and those citations are the part of an answer you can actually inspect. The chart at the top of this page shows ChatGPT's source mix across 9,434 answers. Business websites appeared in 44.2% of answers, Expertise.com in 32.1%, Angi or HomeAdvisor in 29.4%, BBB in 14.5%, Reddit in 4.9% and Yelp in 0.6%. Ranking and "best of" sites alone showed up in 64.2% of the ChatGPT answers that cited anything. In US answers the two big directories were bigger still: Expertise.com appeared in 53.5% and Angi or HomeAdvisor in 49.1%.

Canada runs on different directories. HomeStars appeared in 39.9% of ChatGPT's Canadian answers, Angi and HomeAdvisor in 1 of 3,775, and BBB was cited about twice as often as in the US (21.5% against 9.8%).

So run the buyer questions that matter for your trade and cities, open every list and directory ChatGPT cites, and check three things on each one: whether your business is there, whether the name, phone, hours and service area match your Business Profile, and whether the review count and rating it shows match reality. Profiles you control, like Angi, BBB and HomeStars, can be fixed this week. Editorial "best of" lists choose their own entries, so learn how each one selects businesses before you spend time on it. The ChatGPT sources guide goes source by source.

Keep a rating and review depth that hold up

The businesses the engines named were strongly rated, and the ones named first had the deepest review histories. On Google AI Mode, where ratings were captured all three weeks, only 3.7% of answers with rated business cards named any business under 4.0 stars, while 23.5% named at least one with fewer than 20 reviews. In the 7,873 answers that named three or more rated businesses, the first one named had a median of 342 reviews against 195 for the third, and the most-reviewed business came first 36% of the time against 25% expected by chance.

ChatGPT showed the same order in the five days we captured its map cards, September 18 to 22. Its first-named business had a median of 184 reviews, against 139 for the second and 125 for the third, and the most-reviewed business came first 32.2% of the time against 18.7% by chance.

Review count also tracks how big and how old a business is, so treat this as a pattern, not a lever with a known return. The practical reading is to protect the rating first and build depth steadily by asking every customer the same neutral way. The Google AI Mode guide has the full rating and review profile of the businesses it named, and how many Google reviews you need for AI recommendations breaks the counts down by trade.

Check the Yelp page ChatGPT shows on its map

Yelp barely appears in ChatGPT's cited sources, yet about one in three rated business cards on ChatGPT's map (34%) linked to Yelp, and 72% of ChatGPT answers with rated cards showed at least one Yelp card. The effect was bigger in the US, where 46.4% of cards were Yelp cards, than in Canada at 12.8%.

Yelp cards often show a lower rating than Google does. For the same business in the same city, across 1,624 matched pairs, ChatGPT's Yelp-sourced cards averaged 4.3 stars against 4.9 on Google AI Mode, and 23.9% of them showed under 4.0 stars against 0.25% of the same businesses on Google AI Mode. ChatGPT's other cards matched Google AI Mode's rating within 0.1 star 98% of the time. Of the ChatGPT answers that showed any business under 4.0 stars on the map, 98% (617 of 630) involved a Yelp card.

Open the Yelp page for every location and check the category, hours, phone number and the rating a customer would see. Don't answer a low Yelp rating with a review drive, though. Yelp's own guidance is explicit:

Proactively asking for reviews may hurt your Yelp rating because our automated software may not recommend reviews that seem to be prompted or encouraged by the business.

Fix what the Yelp reviews complain about, respond to them, and keep asking for reviews on Google, where asking is allowed as long as you ask every customer and offer nothing in return. Why ChatGPT shows a lower star rating than Google goes deeper on the Yelp cards, and fixing wrong business information in ChatGPT covers the corrections side.

Expect the list to move, and measure it that way

ChatGPT was the steadiest engine we measured, and it still changed its mind. Asked the same question for the same city a week later, it named 58% of the same businesses again; Google AI Mode repeated 41%. Across engines the overlap was small: when all four answered the same question for the same city in the same week, 77% of the businesses named were named by only one engine.

That makes a single ChatGPT screenshot close to useless as evidence. Track the share of repeated checks that name each location, by engine, over several weeks. Why AI recommends different businesses every time has the full numbers and a measurement plan.

Put the website work into homepage, service and location pages

When an engine linked to a contractor's own website, it linked to the pages that describe the business and the job. Of 74,969 citations to contractor websites across the four engines, 43.1% went to the homepage, 17.1% to service-plus-city pages, 16.9% to service pages and 14.0% to location or city pages, 48% for those three together. Booking, offers or financing pages took 2.3%, blog posts 1.5% and review or testimonial pages 0.7%.

ChatGPT on its own followed the same shape, with 37.4% of its business-website citations going to homepages, 24.1% to service pages, 17.4% to service-plus-city pages and 10.7% to location pages. Page types come from URL rules, so treat the split as approximate. Before you commission another blog post, make sure the homepage names your services and areas and each location has a page that describes the jobs it does. The location page checklist covers what belongs on it.

For Google AI Mode, add the booking link

One lever works on Google AI Mode and almost nowhere else. Google's own AI guidance says Business Profiles "can help your products and services be visible in both AI responses and other Google Search results", and in our data 31% of Google AI Mode answers linked to a business's online booking page on field-service software such as Housecall Pro, ServiceTitan or Jobber, while ChatGPT never did. We think those links come from the booking link on the Business Profile, which is our inference rather than a verified fact, and the Google AI Mode guide explains the rules Google sets for that link.

Gemini needs its own plan

Gemini behaved differently again. In the answers where we captured its sources, 88% cited a business's own website, and 61% of all its answers mentioned years in business, against 20% for ChatGPT. Those are topics the answers covered, not reasons a business was picked. Gemini also shows up in more than one Google product, which is why it has its own guide: how to get your business recommended by Gemini.

What to do this week

• Ask ChatGPT the three questions customers most often ask about your trade, for your five busiest locations, on two different days, and write down every business and source it names.
• Open every directory and "best of" list it cites, then fix your entries on the ones you control and note the ones you're missing from.
• Check the Yelp page for each of those locations: category, hours, phone and the rating a customer sees.
• Flag any location under 4.5 stars on Google and fix the service problem its reviews describe.
• Confirm each Google Business Profile has a working booking link to that location's booking page.

Methodology

Every week, Cheers asks ChatGPT, Gemini, Perplexity and Google AI Mode the same 32 home-services questions in 100 metro areas across the United States and Canada, and records every business each engine names, with its star rating, review count and map position where the engine shows one, and every source it cites. This analysis covers the three weeks from August 31 to September 20, 2026: 34,742 usable answers (9,434 from ChatGPT, 9,435 from Google AI Mode, 9,364 from Gemini and 6,509 from Perplexity) across 60 US and 40 Canadian metros. We excluded empty or cut-off answers, most of them from a Perplexity collection outage. Each question was asked once per city per week, so week-to-week changes include normal answer-to-answer variation. Business names were extracted automatically; a hand check of 40 answers found the extraction caught about three in four of the businesses a reader would count, so counts are lower bounds. Star ratings and review counts are the figures each engine displayed; for ChatGPT and Gemini we captured them from September 18 and September 14 onward. Findings describe associations in this sample, not how any engine ranks businesses.

For this article: ChatGPT map-card figures, including the Yelp findings, come from September 18 to 22, 2026, when about 23 of the 32 questions ran. A Yelp-sourced card is one whose provider link is a yelp.com or yelp.ca business page. Same-business comparisons match a business by normalized name within the same metro and average its Google AI Mode ratings. Source groups come from hand-built domain lists; "business website" counts only domains we could match to a named business, so it is a lower bound. Gemini's sources were captured for 44% of its answers, so its shares use answers with captured sources as the base.

Sources

Checked September 23, 2026.

• Web search. OpenAI API documentation, checked September 23, 2026. Source of the quoted line on sourced citations.
• Optimizing your website for generative AI features on Google Search. Google Search Central, last updated July 10, 2026. Google's guidance that Business Profiles can help a business appear in AI responses.
• Don't Ask for Reviews. Yelp Support Center, checked September 23, 2026. Source of the quoted line on solicited reviews.
• Prohibited and restricted content. Google Maps User Generated Content Policy Help, checked September 23, 2026. Google's rules against selective solicitation, review incentives and staff review targets.
• Cheers Market Baseline, August 31 to September 20, 2026: 34,742 answers from ChatGPT, Gemini, Perplexity and Google AI Mode to 32 home-services questions in 100 US and Canadian metros, aggregate only. Research methodology. Source of every source, rating, review, Yelp, repeat, page-type and booking figure.

Dylan Allen-Arnegård is the CEO & Co-Founder of Cheers, the local search platform for multi-location service businesses.