Do Google Reviews Affect AI Recommendations?

On Google's own AI, Google reviews sit beside nearly every local recommendation, and the businesses it names have strong ones. Across 9,435 Google AI Mode answers to 32 home-services questions in 100 US and Canadian metros, collected by Cheers between August 31 and September 20, 2026, the engine showed a Google star rating and review count for 90% of the businesses it named, 83% of those businesses had 4.8 stars or more, and the business it named first had a median of 342 reviews against 195 for the third. ChatGPT works differently: it cited Google in 0.1% of its answers and leaned on Expertise.com, Angi and BBB instead.

For a brand with 40 locations, that splits the review work in two. Google reviews are the evidence Google AI Mode puts beside each branch, while a separate set of review sites shapes how the same branch looks on ChatGPT and Perplexity. To see what each engine says about one of your locations today, run the AI Visibility Grader.

Last verified: September 23, 2026.

What Google says reviews do

Google is the only engine maker here that says outright that reviews affect where a business shows up. Its Business Profile help page on local ranking says prominence is "based on info like how many websites link to your business and how many reviews you have," and puts it in one line: "More reviews and positive ratings can help your business's local ranking." Its guide to generative AI features in Search, last updated July 10, 2026, says those features "are rooted in our core Search ranking and quality systems," and it points local businesses to their Business Profiles.

Neither page says how Google AI Mode weighs a review. What they establish is that Google's local ranking uses reviews and that its AI features sit on the same core systems, which fits what we measured.

Google AI Mode puts a rating next to almost every name

Google AI Mode showed rated business cards in 90% of its answers, and the share barely moved with the kind of question: 90% for "best X for Y" questions, 92% for emergencies and 93% for deals. For 90% of the businesses it named, the card carried a Google star rating and review count.

The ratings bunched near the top. Of the rated businesses it named, 82.7% had 4.8 stars or more, 12.8% had 4.5 to 4.7, 3.5% had 4.0 to 4.4 and 1.0% were under 4.0, with a median of 4.9. Review counts spread far wider, from 74 at the 25th percentile to 783 at the 75th, around a median of 226. Only 3.7% of answers named any business under 4.0 stars, while 23.5% named at least one with fewer than 20 reviews, so in this data a low rating was much rarer among the named businesses than a low count.

The first business named had the most reviews

The order of an answer followed review count on all three engines that show business cards. In Google AI Mode answers naming three or more rated businesses, the first had a median of 342 reviews, the second 237 and the third 195, and the most-reviewed business came first 36% of the time against 25% expected by chance. On ChatGPT's map cards, recorded September 18 to 22, the order ran 184, 139 and 125, with the most-reviewed business first 32% of the time against 19%. Gemini's cards, recorded September 14 to 22, ran 373.5, 277.5 and 222.5, with the most-reviewed first 33% of the time against 26%.

Ratings barely moved by position: Google AI Mode's first-named business averaged 4.86 stars and its third 4.83. Review count also tracks how big and how old a company is, so read this as a pattern among the named businesses, not proof that reviews move a branch up the list. Consistency followed the count as well; 43% of businesses with 1,000 or more reviews were named again in at least two of three weekly checks, against 26% of those with fewer than 20.

ChatGPT and Perplexity read reviews somewhere else

Google reviews live on Google's own properties, and the other engines rarely cite those. Across 9,434 ChatGPT answers, ChatGPT cited Expertise.com in 32.1%, Angi or HomeAdvisor in 29.4%, HomeStars in 16.0%, BBB in 14.5%, Yelp in 0.6% and Google in 0.1%. Google AI Mode cited Google properties in 95.5% of its answers and businesses' own websites in 92.5%, with Yelp at 19.7%, Angi or HomeAdvisor at 9.3%, BBB at 1.7% and Expertise.com at 0.2%. Perplexity cited BBB in 31.0% of its answers, Expertise.com in 30.9%, Angi or HomeAdvisor in 30.8%, Thumbtack in 20.0% and Yelp in 13.9%.

Counting every directory, review platform and "best of" list together, 84% of ChatGPT answers and 93% of Perplexity answers cited at least one, against 38% for Google AI Mode. Gemini is harder to read because we captured citations for only 44% of its answers, so I've left it out of this comparison.

Country changes the list. Angi and HomeAdvisor were effectively absent from Canadian answers (1 of 3,775 Canadian ChatGPT answers cited them), and HomeStars took their place, appearing in 40% of ChatGPT's Canadian answers. BBB was cited about twice as often in Canada as in the US.

ChatGPT also has a review channel that its citations don't show. About one in three of the rated business cards on its map linked to Yelp, and those cards ran lower than Google's; why ChatGPT shows a lower star rating than Google has the numbers.

The answers say the ratings out loud

Ratings aren't only on the cards. When we asked which electricians, HVAC companies, plumbing companies or roofers "have the strongest reviews and fair pricing," 82.3% of Google AI Mode answers, 78.9% of ChatGPT answers and 78.0% of Gemini answers mentioned a star rating in the text, against 43.0% for Perplexity. Across every question we ask, ChatGPT mentioned a review count in 53.6% of its answers. A mention counts wherever it appears, including general advice, so read these as how often answers talk about ratings, not as the reason an engine chose a business.

What it means for a multi-location brand

The review work splits by engine, and the split shifts with trade and country. For an HVAC group in the US, Google reviews are what Google AI Mode shows beside each branch, and the HVAC businesses it named had a median of 449 of them; the same group's ChatGPT picture leans on Expertise.com, Angi and each branch's Yelp page. For a plumbing brand with Canadian branches, HomeStars turns up in ChatGPT answers in a way it never will in Texas. For a roofing company the directories carry even more of the load, since ChatGPT cited the contractor's own website in only 17.8% of its roofing answers. Every factor we measured across the four engines is on one page in AI local ranking factors for 2026.

The rules differ by platform too. Google lets you ask every customer for a review and prohibits incentives, selective asking and staff review quotas, while Yelp asks businesses not to request reviews at all. The compliant review collection playbook covers the Google side in full.

What to do this week

Pick one market and the question customers ask most for your trade there. Ask it in Google AI Mode, ChatGPT and Perplexity, and for your branch and every business named ahead of it, write down the Google rating and review count, which review sites each engine cited, and whether ChatGPT's card links to Yelp. If your branch is missing from Google AI Mode and sits under 4.5 stars, fix the service complaint and reply to reviews before asking for more of them. If ChatGPT or Perplexity cites Angi, BBB or HomeStars for that question, make sure the branch's profile there is claimed and accurate. Then repeat the exercise in your next two markets with the same question, so the results stay comparable.

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.

Ratings and review counts here are what each engine displayed on its business cards for businesses also named in the answer text: Google AI Mode for the full window (34,017 rated business appearances in 8,523 answers), ChatGPT map cards from September 18 to 22, 2026, and Gemini place cards from September 14 to 22, 2026. The order figures use the 7,873 Google AI Mode answers and 2,085 ChatGPT answers that named three or more rated businesses. The rating-mention figures use 1,130 Google AI Mode, 1,129 ChatGPT, 1,128 Gemini and 681 Perplexity answers to the four strongest-reviews-and-fair-pricing questions. Google AI Mode's cards can hide business names in about 5% of answers. Citation shares are the share of usable answers citing a domain at least once, with URLs de-duplicated within each answer. Text mentions come from pattern matching on the answer text; a star rating counts only when the number sits next to "stars", "/5" or "out of 5", so Google AI Mode's bare card ratings are not counted there.

Sources

• Cheers Market Baseline, August 31 to September 20, 2026, 34,742 answers from four engines in 100 US and Canadian metros. Cheers Research methodology. Source of every rating, review, order, citation and text-mention figure above.
• Tips to improve your local ranking on Google. Google Business Profile Help, undated, read September 23, 2026. Source of the prominence and review sentences quoted above.
• Optimizing your website for generative AI features on Google Search. Google Search Central, last updated July 10, 2026, read September 23, 2026. States that Google's generative AI features are rooted in its core Search ranking and quality systems.
• Prohibited and restricted content. Google Maps user-generated content policy, undated, read September 23, 2026. Source of the incentive, selective-solicitation and staff-quota prohibitions.
• Don't Ask for Reviews. Yelp Support Center, undated, read September 23, 2026. Source of Yelp's rule against asking for reviews.

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