Your Google rating decides what AI says about you

Ask an AI engine who should sell a house in any town and watch what appears beside the agency names: star scores and review counts, lifted straight from Google Business Profiles and printed into the answer like league positions. Not your website's claims. Not your awards page. The Google number, quoted to the decimal, beside your rivals' numbers. For most agencies this is the single most visible fact the machines publish about them, and most have never once checked what it looks like from the outside.

The number they quote is not always the number you think

Two traps here, both measured rather than imagined. First, branches: the engines often attach your branch profile's rating to local questions, not your firm-wide figure, so an agency proud of its overall score can be represented by its weakest or smallest branch, especially where a profile's location label is vague. Second, staleness: we once watched an engine confidently quote a rating an agency had outgrown months earlier. The source was a forgotten line in the agency's own website code, invisible to every human visitor, dutifully read by every machine. Your rating exists in more places than your Google profile, and the engines will quote whichever copy they find. If those copies disagree, you do not get to choose which one a seller is shown.

Quoted is not the same as weighed

Now the finding that stops this being a simple "get more stars" sermon. In one measured answer, an engine ranked an agency rated 4.2 from 42 reviews above a rival rated 4.7 from more than 400. The lower score led the recommendation. Why? The answer praised the lower-rated firm's branch network and its long opening hours: reach and availability outweighed reputation that day. The lesson is precise: your rating is what the engines quote, but it is not, on its own, what they weigh. It sits inside a bigger evidence file, review volume and recency, how visible and consistent your operation looks, what your own pages say, what third parties say. A strong rating with a thin file loses to a weaker rating with a fat one.

What actually moves it

The unglamorous habits, in the order they pay. A steady ask: a review request built into every completion and every tenancy start beats a yearly panic, because engines weight recent reviews most heavily. Replies: answered reviews, including the bad ones, are evidence a machine can read that somebody is home. Branch hygiene: every profile labelled with the town it is actually in, so the right rating attaches to the right question. And consistency: the same, current figures everywhere your rating appears, on your site as visible text fed from a live source, never hand-typed, because hand-typed numbers rot and machines quote rot.

What does not work

Two dead ends worth naming. Bought, gated or cherry-picked reviews: the platforms detect the patterns, and a suspended profile takes your quoted rating to zero. And stuffing your own website's code with self-declared review scores: Google explicitly ignores self-serving review markup, and in every AI answer we have ever captured, the engines took ratings from Google and independent review platforms, never from a firm's own markup. The rating that gets quoted is the one other people gave you, held somewhere you do not control. That is exactly why it is trusted, and exactly why the only way to change it is to earn it.