How Reviews Help AI Recommend Your Business | Build Trust That Converts

I had a weird moment last Tuesday. I was testing something for a client, asked ChatGPT to recommend an electrician near Hitchin, and watched it confidently suggest a bloke with fewer reviews, a lower star rating, and a worse website than my client. My client was livid. But once I dug into why, it made perfect sense, and it changed how I think about reviews entirely.

The AI wasn't broken. It was doing exactly what it was designed to do. It was reading reviews like source material, not scorecards.

Your reviews are being read by robots now

Not metaphorically. Literally. ChatGPT, Perplexity, Gemini, all the AI assistants people are using to find local businesses, they're pulling review text apart word by word. They're not scrolling through and thinking "oh that looks like a decent business." They're extracting data. Services mentioned, problems solved, areas covered, how you handled complaints. All of it gets processed and stored as facts about your business.

So when someone in Stevenage asks an AI "who can fix a combi boiler on a Sunday," the AI goes looking for review content that matches. Not star counts. Not how many reviews you've got. The actual words inside them.

And most business owners have no idea this is happening.

Stars don't do what you think they do

Right, I need to rant about this for a second. I keep meeting business owners who are obsessed, properly obsessed, with maintaining a 4.9 or 5.0 rating. They panic about a single 4-star review. They send follow-up texts begging for five stars.

Stop it.

I've tracked this across about 40 local businesses over the past year. Star rating has almost zero correlation with whether AI recommends you. A roofer sitting at 4.4 stars was getting recommended three times more than his competitor at 4.9. Why? Because his reviews were full of specific, descriptive language. "Replaced the lead flashing around our chimney stack after the storm in January." Compare that to "excellent service, 5 stars, would recommend." The second one is lovely. It also tells an AI absolutely nothing.

Stars still matter for the human scrolling Google Maps. For AI recommendation? They're background noise.

What AI actually pulls from a review

OK so I've spent a lot of time (too much, my wife would say) reverse-engineering what AI tools pay attention to in reviews. It breaks down roughly like this:

Specific services and skills. "Fitted a new consumer unit" is gold. "Did a great job" is invisible.
Geographic mentions, even vague ones. "Covers all the villages around Baldock" or "came out to us in Letchworth within the hour", these build a map of your service area inside the AI's understanding
Problem-and-resolution stories. Before and after. What was wrong, what you did about it. AI loves these because they match the way people actually ask questions... "my boiler keeps losing pressure, who can help"
How you respond to negative reviews (more on this in a bit)
Brand names, equipment types, property styles. If someone mentions you worked on their Victorian terrace or their Worcester Bosch boiler, that's a keyword the AI can match to future queries

And here's what it largely ignores: adjectives. "Amazing." "Brilliant." "Fantastic." These are the review equivalent of white noise. Nice for your ego. Useless for AI.

The trick with review responses that nobody talks about

A customer leaves you a review saying "really happy with the work, thanks Dan." Three seconds to write, means well, tells AI nothing.

But your response? That's where the magic is. You can write back something like: "Cheers! Glad the new wet room turned out well. Those Edwardian floors can be tricky to level but it was worth getting right. Hope you enjoy it."

You just told every AI system reading your profile that you do wet rooms, work with Edwardian properties, and handle floor levelling. The customer didn't say any of that. You did. And it counts.

I started doing this with a decorator in Hitchin about six months ago. We rewrote all his review responses to include detail about what he actually did on each job. Within two months his appearances in AI recommendations jumped. We changed nothing else. Same website. Same number of reviews. Just better responses.

Most businesses either don't respond to reviews at all, or they copy-paste "Thanks for the kind words!" every single time. Both are missed opportunities, but that copy-paste thing might actually be worse because AI can see the repetition and it signals... nothing.

Getting better reviews without being weird about it

Bit of an awkward subject. Nobody wants to be that business that hassles customers for reviews. And you shouldn't be. But there's a difference between hassling and guiding.

When a job goes well, most happy customers are perfectly willing to leave a review. They just don't know what to write. So they default to "great service, highly recommend" and move on.

What works better is something dead simple. After the job, in person or in a follow-up message, say something like: "If you fancy leaving us a review, it really helps if you mention what we worked on. Helps other people with similar jobs find us." That's it. You're not asking them to write an essay. You're not scripting their review. You're just giving them a nudge toward being specific.

I've seen the difference this makes dozens of times. One landscaper I work with went from reviews like "lovely garden" to reviews like "completely redesigned our front garden with a new block-paved driveway and low-maintenance planting beds." Same customers, same satisfaction level, wildly different usefulness for AI.

Negative reviews aren't the enemy (mostly)

OK this might be controversial. But a well-handled negative review can actually help your AI visibility. I know. Sounds mad.

But think about what AI sees when it reads a negative review and your response. It sees a specific problem described. It sees how you handled it. It sees professionalism, or lack of it. And if you responded with detail about what happened and what you did to fix it, that's more useful content about your business entering the system.

I'm not saying go collect bad reviews. Obviously. But when you get one, don't just get defensive. Respond properly. Explain what happened. Mention what you've done since. Be a bloody grown-up about it. AI reads the whole thread and the resolution matters.

The AEO angle on all this

If you've heard me talk about AEO (Answer Engine Optimisation) before, this is a big chunk of it. Your website matters, your Google Business profile matters, but your reviews are the part of AEO that most businesses completely overlook.

AI tools trust reviews because they're third-party. You didn't write them (well... you shouldn't have). They carry more weight than the claims on your own website because they're independent verification. So when you're thinking about how to get AI to recommend your business, your review strategy isn't a nice-to-have. It's a core part of the work.

And it compounds. More specific reviews means more AI recommendations means more customers means more specific reviews. Once you get the wheel spinning it builds on itself.

So what do you actually do on Monday morning

Don't overcomplicate this.

Go read your last 20 reviews. Count how many mention a specific service, area, or problem. If the answer is less than half, you've got a gap
Write proper, detailed responses to your recent reviews. Even the old ones, there's no rule against going back
Next time a customer's happy, ask them to mention what you did. Keep it casual
Check what AI actually says about you. Ask ChatGPT or Perplexity to recommend a business like yours in your area. If you're not coming up... now you know why

That's it. No massive overhaul needed. Just start being specific and encourage your customers to be specific too.

If you want someone to actually audit what AI thinks about your business right now, book a call and I'll walk you through it. I do this for businesses across North Hertfordshire and it's always an eye-opener. Or if you want to read more about AEO in North Hertfordshire, that's a good starting point too.

Either way, your reviews are doing more work than you realise. Make sure they're saying the right things.