Google says its AI Mode has passed one billion monthly active users. In US usage data published in May 2026, the average AI Mode query was about three times the length of a traditional search. Nobody types “plumber Leeds” into a box of that kind. They ask a full question, and they get a full answer, with a short list of businesses inside it.
That is a different game from ten blue links. The question for a small business is no longer where you rank. It is whether you are in the answer at all.
1. The shop window has changed
For two decades, reviews were written for humans. A potential customer scrolled, squinted at the one-star rant, and made a judgement. Now software reads them first. Google says Gemini in Google Maps is grounded in information about 250 million places, which means your listing, your reviews and your replies are the material it works from.
A question like “an emergency plumber near me who is good with old boilers and answers quickly” cannot be settled by a star rating. It is settled by what customers actually said. The review text, and how you responded to it, is the raw material.
2. Where the answers come from
Different systems pull from different places, but the pattern is consistent: they read what is public, specific and recent.
| Where it appears | What it does | What it reads |
|---|---|---|
| Google AI Mode | Answers long, specific local questions | Maps data, reviews, your profile |
| Gemini in Google Maps | Answers questions about places | Information on 250 million places |
| ChatGPT and other assistants | Recommend businesses in conversation | Public reviews and web content |
None of these tools needs you to be technical. They need you to have something worth reading.
3. The volume problem is real
Google reports that the Maps community contributed more than one billion reviews in 2025, and that its systems blocked or removed more than 292 million policy-violating ones. The scale cuts two ways. Genuine reviews are plentiful, but so is noise, and AI systems are trained to be sceptical of it. Specific, detailed, human reviews carry weight because they are hard to fake at scale.
A business with forty detailed reviews that mention the actual job, the actual street and the actual outcome gives an AI far more to work with than one with four hundred reviews that say “great”.
4. Your replies are part of the evidence
Most owners think of a review reply as a courtesy to the reviewer. It is now also a public statement that machines read. A reply that names the service, acknowledges the specifics and sounds like a real person adds context the review alone does not provide. A blank space, or a copy-and-paste “Thanks for your feedback”, adds nothing.
Be honest about the limits here. Nobody can guarantee a position in ChatGPT or Google AI Mode, and Google itself says review count and rating contribute to local prominence without promising a ranking. But the businesses with rich, current, well-answered profiles are the ones with something to be recommended on.
5. What to do this week
Search for your own business in two or three AI tools, using the kind of long question a customer would ask. Note what comes back and what is wrong. Then read your ten most recent reviews and replies as if you were a stranger. Would a machine, or a person, learn anything specific about what you do and how well you do it?
Fix the gaps in the order that is easiest: answer unanswered reviews, ask recent happy customers to mention what you did for them, and make sure your profile details are accurate everywhere.
Conclusion
Google has been clear that review count and rating contribute to local prominence, and equally clear that nothing is guaranteed. What you can control is the evidence. The first thing an AI knows about your business is what it has read, so make sure what it has read is true, specific and yours.
Read next