Your Best Review Has 5 Stars — Google Is Showing Searchers Your One 3-Star Review Instead
A plumbing outfit in Torrance has 187 reviews and a 4.9 average. One of them — a single 3-star review from 2022 — says "a little pricey but they got it done right." Someone in Redondo Beach searches "affordable plumber near me." Google's local pack pulls up the business, and instead of surfacing one of the 160 five-star reviews that mention nothing about cost, it shows that exact 3-star snippet under the listing. The searcher reads "pricey" in a business's own Google listing and clicks the next result down.
The owner never sees this happen. He checks his rating once a month, sees 4.9, and moves on. He has no idea that for a meaningful slice of searches — the ones that use words like "affordable," "cheap," "cost," or "estimate" — Google isn't showing his average at all. It's showing the one review in his entire history that happens to use the same language as the query. Star rating and displayed snippet are two different systems, and most owners have only ever optimized one of them.
Your Star Average Isn't What Google Actually Shows a Searcher
The number at the top of a Google Business Profile — 4.9, 187 reviews — is an aggregate. It's a trust signal, and it does influence whether a business shows up in the pack at all. But it is not the text a searcher reads when they scan results. That text comes from a separate layer: a snippet-selection process that scores individual reviews against the specific words someone typed, then displays whichever review scores highest on relevance to that query — independent of its star rating.
This is why two different searchers, on the same day, looking at the same business, can see completely different reviews. Someone searching "licensed electrician Gardena" might see a five-star review that mentions "licensed" and "professional." Someone searching "cheap electrician" might see a two-star review that says "found someone cheaper afterward." The business didn't change. The query changed, and the retrieval layer picked a different piece of text to match it.
The "Reviews Mentioning" Feature Is Doing More Work Than You Think
Open any Google Business Profile on mobile and tap into the reviews section. You'll often see a row of topic chips — "pricing," "professionalism," "response time," "cleanliness" — generated automatically from the language across all your reviews. Google is running natural-language extraction on every review you've ever received, tagging it by topic and sentiment, and building a searchable index out of it. When a query semantically maps to one of those topics, that index is what gets consulted — not your overall average.
If you have four reviews mentioning "pricing" and three of them are neutral-to-negative because that's simply what people wrote when cost came up, "pricing" as a topic is now underweighted toward negative sentiment in your profile — regardless of the fact that your average across all 187 reviews is excellent. The topic-level corpus and the aggregate rating are calculated from the same raw data but tell two different stories, and Google surfaces whichever one is relevant to the person searching.
Why This Keeps Happening: Review Requests Are Generic, So the Corpus Is Thin on the Terms That Matter
Most trades businesses run one review-request workflow: job finishes, a text goes out, it says "if you had a good experience, please leave us a review," and whatever the customer writes is whatever the customer writes. That produces reviews clustered around vague praise — "great service," "very professional," "would recommend" — which is fine for the aggregate score but does nothing for topic coverage. Nobody is writing about warranty terms, emergency response speed, whether the crew was licensed and insured, or how the quote compared to competitors, because nobody asked them to.
The result is a corpus with deep coverage on generic sentiment and almost no coverage on the specific terms customers actually type into Google before they call anyone. When one of the rare mentions of "cost" or "estimate" happens to be lukewarm, it isn't diluted by four or five positive reviews using the same language — because those don't exist yet. One data point becomes the entire answer Google gives for that query.
This is a volume-and-distribution problem, not a reputation problem. A business can have outstanding actual service and an outstanding average rating and still lose specific searches because its review corpus was never built to cover the terms people search on.
Building a Review Corpus That Wins the Query, Not Just the Average
The fix isn't damage control on one bad review — that review isn't going anywhere, and obsessing over it wastes time. The fix is out-covering it: making sure every commercially important search term has three to five recent, specific, positive reviews using that exact language, so the topic-level index has enough good data that one outlier can't dominate the snippet Google chooses to show.
That means review requests stop being generic and start being targeted to the job that just happened. A water heater install gets a request that references the specific service and invites comment on the quote process. An emergency call at 11 p.m. gets a request that invites comment on response time. A job with a multi-year warranty gets a request that invites comment on the guarantee. The ask doesn't script the customer's words — that's against Google's policy and it reads as fake anyway — but it primes the topic so customers naturally write about the thing that actually differentiated the experience, in their own language.
Review replies matter here too, in a way most owners miss. A thoughtful reply to a negative review that reframes the situation with specific, factual language re-enters the same NLP index Google is scoring — it's additional text tied to that review that can shift how the topic reads. A one-line "Thanks for your feedback" reply does nothing. A reply that says "We quoted the job at $340 before any work began, which reflects licensed labor and a 2-year warranty on parts" gives Google new topical language to weigh against the original complaint.
The Weekly Cadence That Keeps a Bad Review From Ever Winning a Search
This only works as an ongoing system, not a one-time cleanup. A business generating five or six targeted reviews a week, each tied to a specific service and specific language, builds topic coverage fast enough that new negative mentions get outnumbered before they can dominate a query. A business generating one generic review every few weeks never catches up — the index stays thin, and every new negative mention lands as a disproportionately large share of the total signal on that topic.
This is the same operating rhythm Axesris runs for home services clients across the South Bay: weekly review requests tied to job type, weekly monitoring of which search terms are triggering which snippets inside Google's own performance data, and reply copy written to shift topic sentiment, not just to be polite. For plumbing companies specifically, the highest-leverage terms are almost always pricing, emergency availability, and licensing — the three things people search before they'll pick up the phone, and the three things most review corpora are thinnest on.
Pull up your own Business Profile right now, tap into reviews, and read the topic chips the way a stranger searching for your service would read them. If you don't know what shows up when someone searches your trade with the word "affordable" or "reliable" in front of it, you don't actually know what your listing is telling people — you only know what the average says. Those are two different numbers, and only one of them is what's costing you the click.