Reviews and AI recommendations: does review count matter?
Updated October 15, 2026
Not on its own. In our data, firms with 250+ Google reviews were no more likely to be named at all than firms with under 10. But among the firms that did get named, the ones with more reviews were named more often. Reviews look like one signal of many, and no assistant publishes how it weighs them.
What does our data show?
We matched businesses with a Google review count to the businesses the assistants named, using exact names only, in the same city. That gave 1,617 businesses across 9 cities (15 October 2026).
The share named at least once was flat: 7.5% for under 10 reviews, 6.8% for 10 to 49, 7.4% for 50 to 99, 15% for 100 to 249 and 8.5% for 250 or more. The bigger difference was in how often: named firms with under 10 reviews averaged 6.3 mentions, and those with 250 or more averaged 15.8.
- Under 10 reviews: 64 of 853 named, 6.3 mentions on average when named.
- 10 to 49: 33 of 482 named, 8.2 mentions.
- 50 to 99: 9 of 122 named, 12.4 mentions.
- 100 to 249: 17 of 113 named, 13.1 mentions.
- 250 or more: 4 of 47 named, 15.8 mentions.
How sure can you be about that?
Not very. Exact-name matching misses firms the assistants call something slightly different, so the true share named is higher than these figures. The top groups are small: 47 firms with 250 or more reviews. And a link isn't a cause. Busy, established firms collect reviews and get written about, so reviews and mentions may both follow from being well known.
What do the assistants say about reviews?
None of the four publishes a rule like "more reviews, more recommendations". Google documents that Gemini can draw on Maps data, which includes reviews. ChatGPT and Perplexity search the web, where review sites are among the pages they cite.
So what should you do about reviews?
Keep them coming, on the sites your customers and the assistants read. Recent reviews that mention the job and the town give an answer something specific to repeat. A big old total with nothing new is weaker than it looks.
Check your business for free
Enter your website. We ask ChatGPT, Claude, Gemini and Perplexity the questions your customers ask and email you who they recommend.
Common questions
How many reviews do I need to get recommended by ChatGPT?
There's no published number, and our data doesn't show a threshold. Firms with very few reviews got named too. Recent, specific reviews on more than one site are a better aim than a target count.
Do star ratings matter more than review count?
We haven't tested ratings separately yet, so we won't guess. Assistants do repeat what reviews say, so a pattern of complaints can show up in an answer.
Which review sites count?
The ones the assistants cite for your trade and country. In our stored Perplexity answers, US runs leaned on Yelp, BBB and Thumbtack. In the UK you'd look at Trustpilot and Checkatrade, in Australia hipages and ProductReview.
Can I check how my reviews compare?
Our report shows which firms get named instead of you, and where we hold review counts for your city we show how yours compare.