We reviewed nearly 30,000 Google local businesses, and this is what we found
Our collector, Axio-CRED, read the public Google listings of 28,814 local businesses in 11 trades across 50 US cities: the star counts behind 10 million reviews, and the 190,953 reviews Google shows first, with the owners' replies. Here is what they say about how local reputations are won and lost.
Nine in ten reviews are five stars
Of the 10,026,521 reviews counted, 91.6% are five stars and 4.4% are one star. Two, three and four stars together make up just 4.1%. Customers rarely leave a middling review: they give five stars, or now and then one.
Property managers are the exception. 15.3% of their reviews are one star, likely because many of their reviewers are tenants rather than the property owners who hire them.
With ratings this compressed, a tenth of a star is a big move. The next two findings show how big.
A 4.7 is not what it used to be
Because almost every business sits between 4.7 and 5.0, the same rating ranks very differently by trade. A 4.7-star AC repair company is out-rated by 80% of the AC repair companies in the same search results. A 4.7-star property manager is out-rated by only 22%.
Judge your rating against your own trade in your own city, not against a general sense that 4.7 is good.
Six one-star reviews can knock a business below 4.5
For every business rated 4.5 or above (23,703 of them), we worked out how many new one-star reviews it would take to bring the displayed rating down to 4.4. The median across all trades is 15. For 27% of businesses it is five or fewer.
Trades with fewer reviews are the most exposed. A typical funeral home or property manager falls below 4.5 after 6 one-star reviews, and almost half of them after five. A typical emergency plumber can absorb 23.
That is the scale review extortion works at: a handful of one-star reviews posted together, then an offer to remove them for a fee. Google has a form for exactly this: report negative review extortion scams on your Business Profile.
Know your number: how many one-star reviews stand between you and the next step down. If a burst arrives with a demand for money, don't pay and don't reply to it. Keep screenshots and dates, and report it through Google's form.
Owners answer praise more often than criticism
Owners replied to 58% of the reviews we read, but only 48% of the one- and two-star ones. Among businesses whose first-shown reviews include a negative one, 45% leave at least one of those unanswered. Funeral homes answer one negative review in four.
When owners do reply, they are quick: 49% of replies arrive within a day. But many are templates. 6.8% of replies repeat another word for word, and 239 longer replies appear, word for word, on three or more different businesses.
An unanswered one-star review is often the first thing a new customer reads. Answer every negative review, in your own words, and don't reuse the same reply.
What customers complain about depends on the trade
We sorted the 16,927 negative reviews we read by what they mention. Each trade has its own weak spot. A review can mention several things, so the figures overlap.
| Trade | Mentioned most | Next |
|---|---|---|
| Moving companies | damage 53% | price or quote 40% |
| Emergency plumbers | price or quote 47% | communication 30% |
| Personal injury lawyers | communication 45% | insurance or claims 19% |
| Property managers | communication 44% | deposits or refunds 22% |
| Roofers | communication 39% | price or quote 30% |
| Funeral homes | communication 37% | no-shows or waiting 12% |
| AC repair | price or quote 35% | communication 34% |
| Auto body shops | damage 34% | price or quote 34% |
| Medical spas | communication 33% | price or quote 24% |
| Locksmiths | price or quote 31% | communication 18% |
Most complaints are not about the work itself. Communication comes up in a third or more of negative reviews in 8 of the 11 trades: slow answers, missed calls, not being told what is happening.
The same reviewers turn up on competing businesses
Among the 189,484 reviewer accounts behind the reviews we read, 202 reviewed two or more competing businesses, meaning the same trade in the same city, within a single week. 13 reviewed three or more competitors on the same day, almost always with five stars.
Some of this has ordinary explanations, such as a supplier reviewing the businesses it delivers to. Some of it deserves a closer look. None of it shows who wrote a review or why. We read at most eight reviews per business, so the real numbers are higher.
If a competitor's praise looks coordinated, look for patterns rather than single reviews: the same accounts, the same days, the same wording. That is what the Fake Review Audit reads for, and what Google's reporting tools ask you to show.
A few businesses hold most of the reviews
In AC repair, the top 10% of businesses hold 67% of all reviews. To match a city's top three, a typical AC company would need about 30 times its reviews: 4,729 against 159. Emergency plumbers (22 times) and moving companies (15 times) are close behind.
Location matters too. The typical business in Houston has 230 reviews; in Long Beach, 46. In some trades the results are crowded with chains: one national key-copying kiosk brand makes up one in five locksmith results.
In a review-heavy market, volume is a moat. Ask every customer for a review, not only the happy ones, and keep the pace steady.
How we did this
Axio-CRED searched Google Maps for 11 trades (for example, “locksmith in Denver CO”) in 50 US cities on 24 and 25 September 2026, and recorded the listings each search returned, typically 60 to 120. That is the visible part of each market, not every business in it. One of the 11 trades is included in the totals but not named in the charts. For each business we read the star counts Google shows and up to eight of the reviews Google shows first, with any owner replies. Ratings and counts are as displayed on those days. Complaint themes come from keyword matching in English.
Patterns such as shared reviewers are signals worth checking, not verdicts about any business or reviewer, and we name none here.