Why the star average does so little
Almost every established local business sits between four-point-five and five stars. A number that nearly everyone shares cannot separate anyone. When an assistant has to choose three businesses out of thirty that all have great ratings, it falls back on what it can actually read and repeat.
Review text is the richest source of that. It is written by third parties, it is specific, it is dated, and it sits on pages that get retrieved constantly. In practice it functions less like a rating and more like a body of evidence about what you do.
The three properties that matter
- Specificity. "Replaced our 40-gallon water heater the same afternoon" is usable. "Great service, highly recommend" is not.
- Recency. A steady trickle reads as an operating business. A cluster from two years ago reads as one that may have closed.
- Coverage across services. If ninety percent of your reviews mention one service, assistants will describe you as that one thing and skip you for the rest.
How to ask without breaking the rules
You may ask customers for reviews. You may not offer anything in exchange, write them yourself, or tell the customer what to say. The line that keeps you on the right side of Google's policies and still produces useful text is to ask a question rather than request a rating: "if you leave a review, it helps other people if you mention which job we did".
Your replies count too, and most owners waste them. A reply that names the service performed adds readable, service-specific text to a page assistants already read. Ten replies written that way are worth more than a hundred that say thank you.
Fake and unfair reviews
A single unfair one-star review rarely changes an AI answer. A pattern of them can, because assistants pick up on repeated language. Google has a reporting path for reviews that break its policies. Use it for genuine violations, then get back to generating real text, which is the only durable answer.
Review-text audit worksheet
Open your last twenty reviews. Fill this in. It takes about fifteen minutes and it usually explains the whole problem.
| What to count | How | What the result means |
|---|---|---|
| Reviews naming a specific job | Of the last twenty, how many name a service, a part, or a problem? | Under a quarter and assistants have almost nothing to quote about what you do. |
| Distinct services mentioned | List every distinct service named across those twenty. | One or two means you will only be recommended for one or two things. |
| Services you sell that appear zero times | Compare that list against your actual service list. | These are the questions you are invisible for. Each is a specific ask on your next job. |
| Reviews from the last ninety days | Count them. | Zero is a live problem. Recency is one of the cheapest signals for a machine to check. |
| Towns or neighborhoods named | Count how many reviews name a place. | Place plus service in one sentence is the highest-value review text a local business can receive. |
| Your replies that name the service | Count your own replies containing the service name. | This is free text you control on a page assistants read. Most businesses score zero. |
| Reviews on platforms other than Google | Count them across the two next-biggest platforms in your trade. | One platform is a single point of failure. Assistants retrieve from many. |
Re-run this in ninety days with the same twenty-review window. The change in row one is the clearest measure of whether your asking habit actually changed.
You are looking for the term
Sentiment and mention rate
How favorably an assistant describes you is tracked as sentiment, and how often it names you at all is your mention rate. Review text feeds both, which is why it is usually the single highest-return hour in this whole subject for a business that already does good work.
- Is there a review count that gets you recommended?
- No threshold exists, and anyone quoting one is guessing. Text quality and recency separate businesses far more than raw count once you are past the point of looking established.
- Should I pay for reviews?
- No. It violates Google's policies, it is detectable, and generic bought reviews contain none of the specific language that actually helps an AI answer.
- 01Tips to get more reviewsGoogle Business Profile Help
- 02Report inappropriate reviews on your Business ProfileGoogle Business Profile Help
- 03Guidelines for representing your business on GoogleGoogle Business Profile Help
- 04AI features and your websiteGoogle Search Central
Every link above was opened and confirmed reachable on September 1, 2026.
The automated version
See what ChatGPT, Claude, Gemini and Perplexity say about your business
The free scan runs the buyer questions for your trade and city across all four assistants, shows who gets named instead of you, and lists the sources those answers came from. No signup, about thirty seconds.
Run my free scan