How Many Reviews Does Your Business Need to Show Up in AI Search?

The Cheers baseline, collected from August 31 to September 20, 2026 and published September 28, 2026, asked home services questions like the ones your customers ask, then wrote down the star rating and review count shown for each business Google AI Mode named. Its findings, in plain terms:

Low counts get named. Low ratings almost never do. Of the answers that showed a rated business, 23.5 percent named at least one business with fewer than 20 Google reviews. Only 3.7 percent named any business rated under 4.0 stars. Across every business named, 83 percent carried 4.8 stars or more, and the median rating was 4.9. If your rating is under 4.5, fixing what customers are complaining about comes before chasing a bigger count, because more reviews at the same service level mostly add more reviews at the same average.

The middle is higher than most owners expect. The median business named had 226 Google reviews, and the business named first had a median of 342. The middle half of named businesses ran from 74 reviews to 783. Split by count, 8 percent of named businesses had fewer than 20 reviews, about 11 percent had 20 to 49, about 28 percent had 50 to 199, about 33 percent had 200 to 999, and about 21 percent had 1,000 or more. A business with a dozen reviews is not locked out. It is competing in lists where the typical name carries a few hundred.

The bar moves with the trade. By trade, the median named business had 449 reviews in HVAC, 364 in plumbing, 130 in roofing, and 113 in electrical work, across the questions Cheers asks each week. A count that is strong for an electrician is thin for an HVAC company. Benchmark against the businesses AI names for your trade in your town, not against a national number.

Count buys staying power, not entry. Cheers asked the same questions three weeks running. Among businesses named at least once, those with 1,000 or more reviews came back in at least two of the three weeks 43.3 percent of the time. Businesses with fewer than 20 reviews came back 25.7 percent of the time, and each band in between fell in order. One appearance can be luck. A bigger base of reviews is part of what turns an appearance into a habit.

A perfect score is not the goal. Businesses rated 5.0 were not named more consistently than businesses rated 4.9. The median 5.0 rated business in the data had 101 reviews, against 418 for the median 4.9 rated business. Twelve perfect reviews do less work than two hundred very good ones.

Count is not the only review signal, and for a business starting from zero it is not the first one. Seer Interactive analyzed 804,491 AI responses across 1,926 brands on ChatGPT, Google AI Mode, Gemini, and Perplexity, in research run mainly in March 2026. Review and trust sites were the second most cited source type in AI answers overall, and their share rose to about 24 percent of citations when the person asking was close to buying.

On Trustpilot specifically, brands with no profile had a median AI citation rate of 1 percent. Brands with even a minimal profile, as few as 1 to 13 reviews, sat at 53.5 percent. An active profile, with a median of 81 reviews, added roughly another 25 points. That study covers brands rather than local trades, and Trustpilot matters less for a plumber than Google reviews do. The shape of the finding still applies: the jump from nothing to something is the biggest jump there is. A claimed profile with a handful of real reviews is not a small step. In the data, it is most of the distance.

Four things do more work than the raw count once you are past the first handful.

Recency. Reviews age. BrightLocal's 2026 consumer survey work found 74 percent of customers only care about reviews written in the last three months. AI answers lean the same way, because recent reviews are the ones that describe the business as it runs today. Fifty reviews with none in the last year read as a business that used to be good. A steady five a month reads as a business that is busy now.

Words that name the work. A review that says "great service" tells AI almost nothing. A review that says the company replaced a water heater in Dixon on a Saturday tells AI what the business does, where, and how it behaves under pressure. That text is what an answer engine can match to the next homeowner's question. Asking for honest detail, without scripting the words, is worth more than asking for stars.

The rating floor. From the Cheers data above: under 4.0 stars is close to disqualifying, under 4.5 is a real handicap, and past 4.8 the last tenth of a star stops mattering. Protect the floor first.

Your market, not the internet's. AI builds a short list for one town and one trade. Your real benchmark is the review counts on the businesses it already names for your questions. If the three names it gives carry 80, 140, and 210 reviews and you carry 35, that gap, not the 150 or 240 in a national article, is your target.

Under 20 reviews. You are below the line where 47 percent of consumers will not consider a business at all, per BrightLocal's 2026 report, and AI names businesses in your band in under a quarter of answers, usually as the extra name, not the lead. The fix is not a campaign. It is a habit: a neutral ask at the end of every finished job, every time, until the flow sustains itself.

20 to 100 reviews. You are credible and nameable, and this is the band where the other signals decide. Recent reviews, reviews that name the service and the town, a rating held at 4.5 or better, and a claimed, consistent profile everywhere AI looks will move you more than the next 30 reviews alone.

100 or more reviews. Your count is unlikely to be the problem. If AI still does not name you, look elsewhere: whether your reviews are recent, whether your business facts match across your profile, directories, and website, and whether your pages answer the questions in words AI can quote. Those checks are what an AI visibility audit covers.