"How many reviews do I need?" is the question we hear most from contractors. The honest answer is that there's no magic number, but the research gives you a useful direction and a target for how often to ask.
This guide pulls together what recent studies found about review count, review freshness and rating in AI recommendations, explains the rules for asking, and gives you a routine you can run every month. These are patterns in observational data, not proof that reviews cause AI to name you. Popular, busy businesses tend to have more reviews and more mentions everywhere.
What the research found
Count and freshness (Revfer, November 2026). A study of about 2,900 AI answers across 40 US cities and 12 industries found:
- Review count was the strongest predictor of being named, for every AI tool tested. Businesses with 250+ reviews had roughly six times the odds of being named by ChatGPT compared with those under 50.
- Freshness mattered. ChatGPT named about 1 in 5 businesses whose newest review was under a week old, and fewer than 1 in 50 when the newest was a year old or more.
- Businesses named by three or more tools had a median of 488 reviews with the newest about 5.5 days old. Businesses never named had a median of 101 with the newest about 22 days old.
- Replies, photos and booking links "mattered little" once count and freshness were accounted for (Revfer). The authors say the study is observational, and Revfer sells review-reply software.
Rating (Cheers, September 2026). Across 34,742 home-services answers, Google's AI Mode showed a star rating for 90% of the businesses it named. The median rating was 4.9, and the median review count about 226. Only 3.7% of those answers named a business under 4.0 stars, though 23.5% named at least one with fewer than 20 reviews (Cheers).
Customers (BrightLocal, 2026). In a survey of 1,002 US adults, 47% won't use a business with fewer than 20 reviews, 74% only look at reviews from the last three months, 31% need 4.5 stars or higher, and 89% expect owners to reply (BrightLocal).
What it adds up to: a high rating works like a filter, and volume and recency go with being named. Your own customers apply the same logic.
What we see in our own scans
In our November 2026 study of 390 Texas markets, ChatGPT usually named the same top business across its answers (the top business appeared in 87% of answers on average). That favorite is often a business with many reviews, but we haven't measured reviews directly in this study, so look at the pages yourself. For example, compare the businesses named for HVAC in Dallas or roofing in Mesquite with their review profiles. One roofer with no website and 48 Google reviews was named in every answer in one market, which is a reminder that reviews are only one of several things AI reads.
A practical target
No study gives a threshold, so use these as working goals:
- Get over the customer filter first: at least 20 recent reviews and a rating of 4.5 or better, because customers apply it directly.
- Then match your local competition. Look at the businesses ChatGPT names for your trade and city, and compare their review counts and newest-review dates with yours.
- Keep it fresh. A steady stream beats a once-a-year push. Aim for new reviews every week or two.
How to ask (and how not to)
Do
- Ask every customer, soon after the job, with a direct link by text or email. Google allows asking customers to leave a review with a link or QR code (Google).
- Make it easy: one tap, no login hunting.
- Ask for specifics: "Would you mention what we did and your city?" Reviews that name the service help customers and search.
- Reply to each review, thanking the customer by name.
Don't
- Don't offer discounts, gifts or payment for reviews. Google treats incentives as fake and misleading content and prohibits them.
- Don't write, buy or trade reviews, including from friends and staff. The FTC's 2024 rule bans fake reviews, buying reviews and review suppression.
- Don't only ask happy customers. Selective asking distorts reviews, and Google requires reviews to reflect genuine experiences.
- Be careful on Yelp. Yelp is known for discouraging solicitation and may not recommend reviews it believes were asked for (BrightLocal). Send review requests to Google, and let Yelp reviews arrive on their own.
A 20-minute monthly routine
| Week | Task | Time |
|---|---|---|
| 1 | Send review links to every customer from the last 30 days who hasn't reviewed | 10 min |
| 2 | Reply to any unanswered reviews | 5 min |
| 3 | Check your count, rating and newest-review date against two competitors | 5 min |
| 4 | Run the five-minute ChatGPT test and note any change | 10 min |
If you only do one thing, set up a text message that goes out after each completed job.
When reviews aren't the problem
If you have plenty of fresh reviews and ChatGPT still doesn't name you, look elsewhere:
- Your listings may be missing or inconsistent. See the directories ChatGPT reads.
- Your website may block ChatGPT's crawler. See OAI-SearchBot.
- ChatGPT may be citing "best of" sites you aren't on. See the sites ChatGPT keeps citing.
- Also read does ChatGPT use Google reviews.
FAQ
Is 4.9 stars better than 4.7 for AI? One study found ratings of 4.7+ did better with ChatGPT and Google's AI. Treat rating as a filter and keep it high by doing good work, not by filtering who you ask.
Do old reviews still count? For customers, recent ones count more. For AI, fresh reviews went with being named far more often in the study above.
Should I get reviews on other sites? Pick Google plus one or two directories that matter for your trade, and follow each site's rules.
See how your reviews compare. Run a free scan at localpickai.com. You'll see whether ChatGPT names you, who it names instead, which websites it read, and what to fix first. Free, no signup, about a minute.
