How do you get reviews AI search actually trusts?

Why reviews matter more to AI than they used to

Reviews always mattered for reputation, but for AI search they’ve become something closer to a credit score. When an assistant decides whether to recommend your plumbing firm in Stockport over three others, it isn’t just reading star ratings, it’s looking at volume, do enough people have an opinion to trust, recency, are people still using this business now, and consistency, do the reviews actually match what your website says you do. A five-star average built on four reviews from 2021 tells an AI system almost nothing useful. Forty reviews averaging 4.6, with a steady trickle of new ones every month, mentioning the actual jobs you do, boiler installs, emergency callouts, bathroom refits, tells it a great deal. That’s also why review platforms like Google and Trustpilot carry more weight than a testimonials page you wrote yourself: they’re third-party, timestamped, and harder to fake convincingly at scale. If you want AI search to trust you, the fastest lever isn’t better copywriting, it’s a genuine, ongoing flow of real customer reviews on the platforms that matter.

The pattern that gives away fake reviews

Fake and bought reviews are easier to spot than most people think, and that goes for AI systems as much as sceptical customers. The tell-tale pattern is a sudden cluster: twelve five-star reviews landing in the same week after months of nothing, all posted by accounts with no other review history. The language gives it away too. Genuine reviews are specific and a bit scrappy, “Dave came out same day and fixed the leak under the sink, sorted, would use again.” Bought reviews tend to be vague and oddly polished, “Excellent service, highly professional, would recommend to anyone,” the kind of sentence that could describe literally any business on earth. AI models, trained on enormous amounts of real human writing, are increasingly good at spotting that generic pattern, and so are review platforms themselves, which suspend accounts and strip reviews once they catch it. The real damage isn’t just the wasted money on a review-buying service, it’s what happens once a platform flags your listing as suspicious: the genuine reviews you did earn can get buried alongside the fake ones.

How to actually earn a steady stream of genuine reviews

Earning real reviews is less about clever tactics and more about making it easy at the right moment. Ask straight after the job’s done and the customer’s pleased, not three weeks later when the memory’s faded. A text message with a direct link to your Google review page gets far more responses than a vague “please review us” mentioned in passing. For trades, that might mean a follow-up text the same evening; for a café or salon, a small printed card at the till with a QR code works surprisingly well. Never offer a discount or freebie in exchange for a review, that crosses into the same territory as buying them, and most platforms’ terms explicitly ban it. What you can do is simply ask more consistently: if one in five happy customers leaves a review when asked, and you go from asking occasionally to asking every time, your review count grows steadily and naturally, exactly the pattern AI systems look for. It’s slow, but no fake-review shortcut can match it for long.

Handling the bad review honestly

A bad review isn’t the disaster it feels like, and it might even help you. A profile of nothing but five-star reviews looks slightly suspicious to a sceptical human, and increasingly to AI systems too, real businesses have the odd off day. What matters more is how you respond. A calm, specific reply, “Sorry this happened, we’ve since changed our booking process to stop this, please call us and we’ll put it right,” reads as a business that takes feedback seriously. Ignoring negative reviews, or replying defensively, does the opposite: it’s a pattern that shows up clearly to anyone, human or machine, reading the full thread. AI assistants summarising your reputation aren’t just counting stars, they’re increasingly reading the substance of what’s said, including your responses. So don’t chase perfect. Chase honest, and handled well.