Yes. AI tools like ChatGPT, Gemini and Google’s AI Overviews read reviews when deciding which businesses to name, and they often repeat what reviewers say. A steady flow of recent reviews on Google and the main review sites in your trade matters, but the wording matters as much as the count: reviews that mention the specific service, the place and the outcome give AI something to match and quote.
Key takeaways
- Do they count? Yes. AI tools read reviews when deciding who to name, and they often repeat the reviewer’s own wording back in the answer.
- The strongest signal In Ahrefs’ study of 75,000 brands, mentions on other websites scored 0.664 against being named in ChatGPT. Domain rating scored 0.266.
- Wording over count A review that names the service, the place and the outcome gives AI something to match and quote. “Great service” gives it nothing.
- Pace Five new reviews a month, every month, beats fifty in one week. A burst looks unnatural to platforms and to readers.
- The rules Asking only happy customers, or offering anything in return, breaks Google’s rules and can get the reviews removed.
- Local and service businesses whose customers will leave reviews
- Owners who can ask after every job
- Trades and professions with their own review sites
- Businesses that can't contact customers afterwards
- Anyone looking to buy, filter or gate reviews
Do AI tools read reviews?
Yes, in two ways.
First, when an AI tool searches the web to answer “who’s a good physio in Calgary?”, review sites and Google Business Profiles are among the pages it finds. It reads the ratings and the review text the same way it reads any other page.
Second, reviews shape what the rest of the web says about you. Directories, “best of” lists and local articles often pick businesses by rating. In a study of 75,000 brands, being mentioned on other websites went with being named in ChatGPT more strongly than almost anything else.
Google’s own AI Overviews draw on the same local data as Google Maps, where reviews are a big part of how businesses are ranked. If you’re new to that side, here’s how AI Overviews pick sources.
Reviews won’t get you named on their own. They’re one part of AI SEO, alongside clear pages and mentions on other sites. But they’re often the part that tips an AI tool from “a dentist” to “this dentist”.
Which review sites matter?
Google is first for almost every business. After that, it depends on your trade and country:
| Trade | Review sites AI tools often cite |
|---|---|
| Most local businesses | Google Business Profile, Facebook, Yelp (US), Trustpilot |
| Dentists and clinics | Google, Doctify and iWantGreatCare (UK), Healthgrades and Zocdoc (US) |
| Law firms | Google, ReviewSolicitors (UK), Avvo and Martindale-Hubbell (US) |
| Trades and home services | Google, Checkatrade and TrustATrader (UK), Angi, Houzz and the BBB (US) |
| Agencies and B2B | Google, Clutch, G2, LinkedIn recommendations |
| Weddings | Google, The Knot and WeddingWire (US), Hitched (UK) |
A quick way to find yours: ask ChatGPT and Perplexity your main buyer question and look at the sources they show. If the same review site keeps coming up, that’s the one to work on. There’s more on this for builders and trades in AI SEO for trades.
Does the number or the wording matter more?
Both, in different ways.
| What | Why it matters |
|---|---|
| How many | A business with 8 reviews looks riskier than one with 80. Below a certain number, AI tools often skip you. |
| How recent | Reviews from this year show you’re still trading and still good. A wall of three-year-old reviews looks stale. |
| Star rating | Most tools favour businesses in the 4s. A 4.8 with real reviews beats a 5.0 with five. |
| What they say | The service, the place and the outcome give AI tools something to match to the question and quote back. |
| Your replies | Thoughtful replies show you’re active and fair, and add your own words to the page. |
The wording is where most businesses miss out. Compare these two reviews:
| Review | What an AI tool learns |
|---|---|
| “Great service, highly recommend!” | Someone liked them |
| “Dr Shah did my Invisalign in Leeds. Finished in 14 months, and the payment plan made it easy.” | Invisalign, Leeds, a named dentist, a timeline, and a payment plan |
The second review answers three buyer questions at once. Ten of those do more than fifty of the first kind.
How do you get more good reviews?
Ask at the right moment
Right after a good result: the job’s finished, the case is won, the photos are delivered. Not a month later.
Make it one tap
Send the direct Google review link by WhatsApp or email. Every extra step loses people.
Prompt the detail, not the stars
“If you have a minute, it would help others if you mentioned what we did for you and where.” Never ask for five stars, and never offer anything in return.
Ask everyone
Asking only happy customers (“review gating”) breaks Google’s rules, and a natural mix looks more trustworthy anyway.
Make it routine
Five new reviews a month, every month, beats fifty in one week and then nothing.
After a couple of months, check whether ChatGPT names you and how it describes you. You’ll often see your reviews’ wording in the answer.
What do you do about bad reviews?
- Reply to every one, calmly and briefly. Thank them, say what you’ll do, and take it offline.
- Don’t argue in public. Readers, and AI tools, see your reply as part of the story.
- Fix the pattern. If three reviews mention slow replies, that’s the real problem.
- Report fake reviews through the platform. Don’t try to bury them with fake positive ones.
- Keep the good ones coming. A few bad reviews among many recent good ones barely move the picture.
Where should reviews live on your own site?
Reviews on Google and review sites do the most work. But your own pages are where AI tools look for detail about each service, so put the right reviews in the right places:
| Page | Which reviews |
|---|---|
| Each service page | Two or three that mention that service by name |
| City or area pages | Reviews from clients in that place |
| Homepage | Your rating and review count, with a link to the source |
| Case studies | A quote from that client, word for word |
Always link to where the review came from, and keep the wording exactly as written. Edited reviews look edited, to people and to AI tools.
What are the most common review mistakes?
| Mistake | What happens |
|---|---|
| Asking only happy customers | It breaks Google’s rules and can get reviews removed |
| Offering a discount for a review | Same: incentives aren’t allowed on Google or most review sites |
| A burst of 40 reviews in one week | Looks unnatural; a steady few a month looks real |
| Ignoring the review sites in your trade | AI tools may cite Avvo, Checkatrade or Clutch more than Google for your questions |
| Never replying | You miss the chance to add your own words and show you’re active |
What else do people ask about reviews and AI?
Can I buy reviews to get recommended?
No. It breaks the rules of Google and every major review site, and it’s illegal in a growing number of countries, including the US and UK. Platforms also remove them.
Should I copy reviews onto my website?
Yes, with permission and a link to the source. Reviews on your own service pages help people decide, and give AI tools more detail. Don’t mark them up with review schema for your own business.
How many reviews do I need?
There’s no magic number. Aim to match or beat the businesses AI tools name for your main question, then keep adding a few every month.
Do replies to reviews matter?
Yes. Google says replying shows you value customers, and your replies add your own words about your services to the page.
