Yes, reviews still matter, and for AI search they arguably matter more than they ever did for rankings. What’s changed is the job they do. A Google results page put ten businesses in front of a customer and let them weigh it up. An AI answer names three or four, and reviews are part of how it decides who makes the cut. We test this on real local queries every month as part of our AI visibility work, and the pattern in our latest round was hard to miss.
What did our own testing show?
In our last spot-check across ChatGPT, Perplexity and Google’s AI Mode, using the kind of local service questions real customers ask. On Google’s AI Mode, the businesses being named tracked review counts closely. The ones that made the shortlist were consistently the ones with the deepest, most recent review profiles.
That makes sense, given AI Mode has Google’s own review data sitting right next to it. ChatGPT and Perplexity read more widely, which I’ll come back to.
I’ll be up front about the limitations. This is a spot-check, not a controlled study. AI answers vary between runs, and a deep review profile tends to travel with other things that help too: years in business, an active Google Business Profile, more mentions around the web. But when the same pattern holds across the categories we tested, it’s worth acting on.
Why do AI engines lean so heavily on reviews?
Think about the problem the engine is solving. Someone asks for a good electrician in Newcastle. There are dozens who could do the job, and the engine has to compress that down to a handful of names it’s prepared to put forward. It needs evidence, and reviews are the biggest body of independent, dated, plain-language evidence about a local business that exists anywhere.
Each part of a review profile tells the engine something different:
- Volume says plenty of real people have used you.
- Recency says you’re still trading and still doing good work, rather than coasting on a reputation from 2021.
- The review text supplies language the engine can match to a question.
That last one is the part most businesses haven’t clocked. As we explained in how AI assistants choose which local business to recommend, these engines break a question into a set of related searches and look for passages that answer them. A review that says “they rewired our 1950s house in Mayfield” contains a service, a property type and a suburb, written the way a customer talks. When the next person asks about rewiring an older home in Newcastle, that review is doing work your service page might not be.
A review that names the service and the suburb gives an AI engine something to match. Five stars and “great service” gives it almost nothing.
What does this change about asking for reviews?
For years the goal was the star rating. Get the average above 4.5, keep it there, move on. That still counts, because a poor average will keep you off a shortlist quickly. But the bigger lever now is what customers actually write.
A five-star review that says “great service, highly recommend” reassures a person. It gives an AI engine very little to work with. A review that says what you did and where gives the engine exact language to match against the next customer’s question.
The trick is asking for that without coaching people into copy that reads as fake. Scripted reviews all sound the same, to people and to machines. Here’s the approach we suggest to clients:
- Ask soon after the job. While the work is fresh and the customer is still pleased, not weeks later on an invoice footer.
- Make it one tap. Send the direct link to your Google review form, not a link to your profile that makes them hunt for the button.
- Give a prompt, not a script. Something like: “If you’ve got a minute, we’d really appreciate a Google review. Mentioning what we did and roughly where you are helps other people nearby find us.” That’s enough. They’ll write it in their own words, which is the whole point.
- Ask everyone. Don’t only ask the customers you’re sure were happy. Google’s policies don’t allow selectively asking for positive reviews, and a profile that’s nothing but glowing looks less credible.
- Reply, briefly and personally. A reply shows the business is active. If it naturally mentions the job, it adds context, as long as it doesn’t read like it was written for a search engine.
Do reviews on other sites matter for AI search?
Yes, and this is where ChatGPT and Perplexity part ways with Google’s AI. Google reviews carry the most weight on Google’s own surfaces. ChatGPT and Perplexity draw on a wider spread of sources: industry review sites, directories, local forums and social platforms.
In practice that means three things. Know which review platforms matter in your industry, whether that’s a trades marketplace, a health booking platform or a hospitality site. Keep a genuine presence on two or three of them, not just Google. And make sure your business details are identical on every one. The same name, address, phone number and service description everywhere is how an engine works out that reviews on four different sites are all about the same business. It’s the same consistency point we made in why ChatGPT recommends your competitors and not you, and it applies to reviews as much as listings.
Are reviews enough on their own?
No. Reviews are one signal among several, and a great review profile attached to a website the engines can’t read will still lose. The engine needs to fetch your site, understand what you do and confirm the details match. If your services are buried in images, your contact details are inconsistent or your pages never answer a question plainly, reviews alone won’t carry you onto the shortlist.
And don’t buy or incentivise reviews. Offering a discount or a gift for a review is against Google’s guidelines, and fake or misleading reviews can breach the Australian Consumer Law, which the ACCC does enforce. Google already removes reviews it judges to be fake and can restrict profiles that attract them. The real risk is being filtered out entirely, and in a shortlist of three that’s the same as not existing.
In a shortlist of three, being filtered out is the same as not existing.
The good news is that the legitimate version is simple and cheap. Do good work, ask every customer, and make it easy for them to say what you did and where.
Frequently asked questions
Do Google reviews affect whether AI recommends my business?
Yes. AI answers name a handful of businesses rather than showing a list, and review volume, recency and content are part of how they choose. In our own testing, Google’s AI results for local services closely followed review depth. Reviews are one signal among several, so a strong profile works best alongside a clear, readable website and consistent business details.
How many reviews does my business need?
There’s no magic number, because it’s relative. AI engines compare you with the other businesses they could name for the same question. Check the review counts of whoever is currently being recommended for your key queries; that’s your benchmark. Recency matters too, so a steady flow of new reviews beats a big batch from a few years ago.
Do reviews on sites other than Google matter for AI search?
Yes. ChatGPT and Perplexity draw on multiple sources, including industry review sites and directories. Google reviews carry the most weight for Google’s AI, but a footprint across several platforms with consistent business details helps everywhere.
What should I ask customers to say in a review?
Nothing scripted. Ask them to mention what you did and where. A review that says the service and the suburb in plain words gives AI engines exact language to match against what future customers ask. Never offer incentives for reviews, and ask every customer rather than only the ones you know were happy.
Want to know where you stand?
If you’d like to know whether AI engines are naming your business, and why or why not, our AI readiness review covers exactly this: where you appear across ChatGPT, Perplexity and Google’s AI, how your reviews and listings compare with the businesses being named, and what it would take to close the gap. It’s one half of what we described in the two questions business owners keep asking about AI.
Prefer to start on your own? Our guide to checking where you really stand in AI search takes about 30 minutes. Or get in touch with the team and we’ll take a look together.
By Brendan Brooks, Founder and Managing Director of HyperWeb. Brendan has spent 25 years watching search evolve and helping Australian businesses stay visible through every shift.
