Review Signals in AI Recommendations: Why Your Google Rating Shows Up in ChatGPT

Ask ChatGPT, Perplexity or Google’s AI Overview for “the best physio in Brighton” or “a good Chinese accountant in Box Hill” and look at how the answer is built. Almost every recommendation comes with a reason, and the reason is very often reviews: “highly rated”, “customers mention friendly staff”, “4.8 stars from over 200 reviews”. For a Melbourne small business, reviews have quietly become part of how AI tools decide whether to mention you at all.

Where AI tools get review information

AI assistants don’t have a private feed of your reviews. They pick up review signals from what’s publicly available and what their search partners return:

• Google Business Profile ratings and review counts, which appear in Google’s own results and AI Overviews.
• Review text that search engines have indexed, including what customers repeatedly mention.
• Third-party sites such as industry directories, booking platforms and local listings.
• Your own website, if it describes your services clearly. AI tools often combine “what the business says it does” with “what customers say about it”.

The exact mix varies between tools and changes over time. None of them publishes a formula. What’s consistent is that businesses with a steady volume of recent, specific reviews get described more confidently.

Which review signals seem to matter

• Volume. Forty reviews give an AI tool more to work with than four. A handful of reviews reads as uncertain.
• Recency. Reviews from this year suggest the business is active. A profile whose newest review is from 2023 looks stale.
• Specific language. “They fixed my hot water system the same afternoon” tells an AI tool what you do and how well. “Great service!” tells it very little.
• Replies. Owner replies show the business is engaged, and a calm reply to a complaint reads better than silence.
• Consistency. Your name, address, phone and services should match across your profile, website and directories, so the reviews clearly belong to you.

What this means for a Melbourne business

The practical work is the same as good review management, just with a new reason to do it:

1. Ask every customer, the same way, with a direct review link. Never offer anything in return, and don’t only ask the happy ones; both break Google’s policies.
2. Encourage detail without scripting it. Asking “what did we help you with?” tends to produce reviews that mention the service and suburb naturally.
3. Reply to every review within a day or two.
4. Make your website match. If customers praise your weekend service, your service pages should say you’re open on weekends. If they mention a suburb, have a page that covers it.
5. Add structured data to your site so AI tools can read your business details clearly. Our guide on schema markup for small businesses covers which types matter.

How to check what AI tools say about you

Before changing anything, look at where you stand. Ask ChatGPT, Perplexity and Google the questions your customers would ask, such as “best dentist in Glen Waverley” or “Chinese-speaking accountant near Box Hill”, and note whether you are mentioned and how you are described.

Then compare that description with your reviews. If AI answers call you “affordable” but your reviews mostly praise “quick turnaround”, the tools may be leaning on old or third-party information. If you are not mentioned at all, check whether your Google Business Profile is complete, whether your reviews are recent, and whether your website clearly says what you do and where.

Repeat the same questions every month or two. AI answers shift, and a simple log of what each tool says is the easiest way to see whether your review and website work is having an effect.

What not to do

Buying reviews, swapping reviews with other businesses or posting them yourself will eventually be found, and when Google removes fake reviews it can take genuine ones with them. Review gating, where you only send review requests to customers you expect to be happy, also breaks Google’s rules. AI tools also tend to pick up on patterns: dozens of short five-star reviews in one week look nothing like a real customer base.

Reviews are one part of how AI tools decide who to recommend. The other part is whether your site answers questions clearly. That’s what our answer engine optimisation and AI SEO in Melbourne work covers. If you’d like us to look at how AI tools currently describe your business, get in touch.

Frequently asked questions