Think about the last time you searched for something important. Did you type two words into Google and scroll through ten blue links? Or did you ask a question — maybe to ChatGPT, maybe to Google's AI Overview — and wait for a direct answer?
Most people, increasingly, are doing the second thing. And that shift — from keyword search to conversational AI search — has quietly changed the rules for every local business trying to be found online. The businesses that understand this now have a significant and growing advantage over the ones still thinking about SEO the old way.
How people used to search — and how they search now
The old search rewarded businesses that had the right keywords on their website. The new search rewards businesses whose customers have used the right language in their reviews — because that's what the AI is reading to form its answer.
Your website is still important. But for local search, your reviews are now the content that AI reads, summarises, and uses to decide whether to recommend you. If your reviews are thin, generic, or old, the AI has nothing useful to say about you — and it will find someone else to recommend instead.
You used to optimise your website for Google's algorithm. Now you need to help your customers write reviews that AI can actually use.
What Google's AI Overview actually does with your reviews
When someone searches with a question — "best family dentist in Cedar Park who's good with nervous patients" — Google's AI Overview doesn't just look at your website. It reads your Google reviews and extracts semantic meaning from them. It's looking for patterns in the language customers use: what services you provide, what you're particularly good at, what kind of customers you serve, what the experience of working with you actually feels like.
It then synthesises those patterns into a recommendation. If three of your reviews mention "great with anxious patients" and two mention "gentle with kids," the AI has enough signal to confidently recommend you for that search. If your reviews all say "great service, highly recommend" — it has nothing specific to work with, and you won't feature in the answer.
Specific services mentioned. Location signals ("came out to our place in Pflugerville"). Experience descriptors ("quick," "tidy," "explained everything"). Customer type ("elderly mum," "rental property," "commercial kitchen"). Problem solved ("fixed a leak," "emergency callout," "wouldn't start"). Every one of these is a signal the AI uses to match you to the right search query.
The difference between a useless review and a powerful one
Here's what this looks like in practice. Both of these are genuine 5-star reviews. Only one of them does anything for your AI visibility.
The second review contains the business name, the type of job, the location, the property type, the specific problem, a time signal, and a trust descriptor. That's the raw material AI needs to confidently match this business to relevant searches. The first review gives AI nothing to work with.
You can't write reviews for your customers. But as we explored in The five-star culture, the way you ask shapes what they write. A specific prompt produces a specific review. "Could you mention what job we did and where?" gets you something the AI can use. "Can you leave us a review?" gets you "great service, thanks."
Why velocity is now more urgent than ever
AI doesn't just read your reviews — it weighs them by recency. A review from two years ago carries significantly less signal than one from last month, for two reasons.
First, AI models are trained to prioritise freshness. Stale reviews suggest a business that may have changed — different staff, different quality, possibly closed. The AI is trying to give the searcher accurate, current information. Old reviews make it uncertain. Recent reviews make it confident.
Second, the language in older reviews may no longer match how people search today. The conversational queries people use with AI are different from the keywords they typed two years ago. A review from 2022 that says "good plumber, fair price" doesn't contain the kind of natural language a 2025 AI query is looking for. Newer reviews, written by customers who themselves use AI and search conversationally, tend to be more naturally descriptive — which makes them more useful as AI training signal.
This is why the velocity argument has become more urgent, not less. A business collecting 3–4 genuine, specific reviews every month is continuously feeding the AI fresh, relevant signal. A business that did a review push two years ago and stopped is working with a fading, increasingly irrelevant dataset.
Every month you collect consistent reviews, you pull further ahead. Every month a competitor goes quiet, their AI signal degrades. This gap doesn't close quickly — it takes months of consistent collection to rebuild. The businesses starting that system now are building an advantage that will be very difficult to replicate in 12 months' time.
ChatGPT and the new discovery layer
Google's AI Overview is the most immediate concern — it appears at the very top of search results and can recommend specific businesses before a user ever sees the traditional results. But it's not the only AI surface that matters.
A growing number of people now ask ChatGPT directly for local business recommendations. "Find me a reliable electrician in Manchester who's good with older properties." ChatGPT pulls from web data — including Google reviews, review aggregators, and business listings — to form its answer. The same principles apply: specific, recent, language-rich reviews give the AI something to work with. Generic or absent reviews mean you don't feature.
This is a new discovery layer that didn't exist two years ago and is growing fast. The businesses being recommended by AI today are not necessarily the ones with the best websites or the highest ad spend — they're the ones with the richest, most recent review content. That's an unusual situation where a small local business with a great review system can genuinely outcompete a larger competitor who hasn't caught on yet.
Most local businesses haven't adapted to this yet. The ones who build consistent, specific review collection now will own this channel before their competitors realise it exists. That window won't stay open indefinitely.
What to do about it — practically
- Stop accepting generic reviews. "Great service" is not enough anymore. When you ask for a review, prompt specificity: "Could you mention the job we did, where you are, and what made the difference for you?" Most customers are happy to — they just need the prompt.
- Make collection consistent, not occasional. One review a week beats ten reviews in a month followed by silence. AI weighs recency heavily. Your review stream needs to look like a healthy, active business — not a burst of activity followed by nothing. See our breakdown of review velocity for exactly what cadence to aim for.
- Respond to reviews with specific language. Your responses are indexed too. When you reply "Thanks Sarah — glad we could sort the boiler in your Victorian terrace so quickly," you've added "boiler," "Victorian terrace," and a location signal to your profile's semantic footprint. Every response is a small SEO act.
- Check your Google AI Overview right now. Search your main service in your area and see if an AI Overview appears. Does it mention you? Does it mention your competitors? That tells you exactly where you stand and what you're up against.
- Don't wait for this to feel urgent. The businesses already winning AI search started their review systems before AI search was mainstream. By the time it feels urgent to most businesses, the gap will already be significant.
The underlying shift
What's happening isn't really about AI — it's about a fundamental change in how trust is established before a purchase. Customers have always wanted to know "can I trust this business?" They used to answer that by asking friends. Then by reading reviews. Now by asking AI, which reads reviews on their behalf and distils an answer.
The mechanism has changed. The underlying need hasn't. Your reviews are your reputation — and your reputation is now being read, processed, and broadcast by AI at a scale and speed that wasn't possible before. A strong, consistent, specific review profile isn't just good for SEO anymore. It's the core asset your business needs to be discoverable in the way customers are increasingly searching.
Keywords on your website got you found in 2015. Backlinks got you found in 2019. In 2025, it's the language your customers use in their reviews — and how recently they wrote them. That's the lever. Pull it consistently and you build an asset that compounds every single month.