Negative Reviews: What 73,142 Reviews Show in 11 Industries

In 73,142 written one- and two-star Google reviews of 2,805 businesses in 11 industries, a language model classified the main reason for each low rating. It chose how the business is run in 65.0 percent (moving) to 86.9 percent (towing and electrical) of the reviews it labelled. No sample was checked against human coding.

Sales Roadmaps has published 11 negative-review studies, from dental practices to gyms, and this analysis reads them together. Each study sampled up to 40 high-review-count businesses in each of eight US metros. It pulled up to 60 of each business’s lowest-rated reviews, then kept the one- and two-star reviews with at least five words of text, dated from 2008 to September 2026.

Every review-content figure below is a share of negative reviews, never a share of customers. The 2,936 businesses sampled after exclusions hold 2,487,470 lifetime Google reviews. Their one- and two-star share ranges from 2.6 percent (electrical) to 20.5 percent (property management).

One question, asked the same way in every industry

Each study ran two model passes. The first assigned theme labels, and every industry had its own label set, so those labels do not compare from one industry to the next. The second pass asked one question of every review: what is the single main reason for the low rating?

Three answers were allowed: the core service itself, how the business is run, or unclear. Every industry used the same prompt, the same model (Claude Haiku 4.5) and temperature zero. Only the business type and the description of the core service changed, for example the dentistry itself or the tow itself.

That second pass is the only review-content measure this analysis compares between industries. It labelled 73,106 of the 73,142 reviews. Pooled, 74.7 percent were classified as mainly about how the business is run.

Industry	Reviews labelled	Main reason: how the business is run	Main reason: the core service	Main-reason agreement (kappa)
Towing	5,916	86.9%	11.6%	0.80
Electrical	1,270	86.9%	11.5%	0.76
Property management	7,554	82.6%	15.6%	0.74
Plumbing	3,706	80.9%	18.3%	0.69
Gyms	5,491	80.1%	19.0%	0.84
Pest control	5,451	76.7%	21.8%	0.85
Dental	9,786	75.6%	22.4%	0.88
Legal	5,068	74.6%	22.0%	0.69
Veterinary	9,149	70.6%	28.1%	0.83
Auto repair	12,878	65.9%	33.2%	0.79
Moving	6,837	65.0%	34.1%	0.68

The agreement column compares the main reason from Claude Haiku 4.5 with Claude Sonnet 4.5 on 200 reviews per industry. Auto repair’s figure was measured on November 3, 2026, when its main-reason pass was completed. Both models come from the same family, so agreement between them is not validation. The shares that remain, between 0.8 and 3.4 percent, were classified as unclear.

Moving and auto repair have the lowest process shares, and even there about two in three negative reviews were classified as mainly about how the business is run. One possible reading, not tested here, is that a damaged sofa or a repair that failed leaves evidence a reviewer can point to.

How the 11 samples were built

Every study drew Google listings in the same eight metros: Phoenix, Dallas, Atlanta, Charlotte, Columbus, Denver, Tampa and Kansas City. Each kept 40 high-review-count businesses per metro. Reviews were pulled lowest rating first, up to 60 per business, and only one- and two-star reviews with at least five words of text were kept. Reviewer names were never collected.

Each sample was then cleaned of listings that did not belong to the industry. Plumbing and electrical kept only companies whose names carried their own trade and no other, because the same multi-trade home-service companies appeared in both pulls. Other studies dropped, for example, a debt collection agency from legal, junk-car buyers from towing, and homeowner-association managers from property management.

Among other exclusions, veterinary dropped emergency and specialty hospitals, pest control dropped wildlife removal and retail, moving dropped junk haulers and container rental, and gyms dropped climbing and children’s programs. The six newest studies also capped each brand at two locations and counted duplicate review texts once. Named exclusions carry a recorded reason for each business.

These choices describe established, high-review-count businesses in those metros. They do not describe every business in the trade, and small or new businesses are not represented.

It holds in 97.6 percent of businesses with enough reviews

An industry share can hide a split, with a few businesses pulling the number up. The data does not show that. Of the 1,617 businesses with at least 20 labelled negative reviews, 39 had fewer than half of those reviews classified as mainly about how the business is run. That leaves 97.6 percent at half or above.

Businesses do differ. After allowing for sampling noise, business-level shares around each industry’s level have a standard deviation of roughly 4 to 10 points. Moving has the widest spread, and there 8.8 percent of the 148 qualifying businesses fall below half. So a business-level figure is worth computing, because it can sit well above or below its industry.

Best-rated and worst-rated businesses show a similar main-reason split

Within each industry, businesses with at least 200 lifetime reviews and 20 labelled negative reviews were split into thirds by their lifetime one- and two-star share. The worst-rated third has 1.8 to 8.1 times the median negative share of the best-rated third. Its main-reason process share differs from the best-rated third by 5.1 points lower to 9.0 points higher, and it is higher in 7 of the 11 industries.

No significance test was run, and electrical has only 7 businesses in each third. The comparison covers the main-reason split only, not the mix of topics. In towing, for example, the topics differ sharply between thirds, as the towing section shows.

Disclosure: this analysis was produced by Sales Roadmaps, which sells operations consulting services. About the operations consulting service

Money complaints rose clearly in one industry

Each industry’s money labels cover its own pricing, fee, billing, deposit, cancellation or upsell themes, so the levels do not compare between industries. The trend within each industry does. A linear trend test from 2018 to 2026, with reviews grouped by business, was run on all businesses and again on businesses whose negative reviews were fully sampled.

With 11 industries tested, a result had to clear a Bonferroni-corrected cutoff, a z score of 2.84, in both runs. Only moving did. In the full data, the share of negative moving reviews carrying a money label rose from 34.8 percent in 2018 to 48.6 percent in 2026. The 2026 reviews run only through September.

Property management, veterinary and legal rose in the full data but failed the fully sampled check. Gym money labels fell in the full data and also failed that check. In 2020, 63.6 percent of negative gym reviews carried a billing or cancellation label, and unauthorized billing alone appeared in 53.7 percent. The data does not say why any of these moved.

In towing, private-property tows concentrate in the worst-rated companies

Towing has the second-highest negative share of lifetime reviews, at 12.1 percent. Inside towing, the worst-rated third of companies carry far more complaints about private-property and impound tows. Those tows appear in 44.7 percent of the worst-rated third’s negative reviews, against 6.6 percent in the best-rated third, a rank correlation of 0.59 across 116 companies.

Of 243 towing companies with at least 200 reviews, 25 have more one- and two-star reviews than all other ratings combined. Private-property tow reviews raise pricing in 52.1 percent of cases, against 31.8 percent of other towing reviews. This is an association, and the data cannot separate the type of tow from the way a given company runs it.

Questions this raises for an owner

These are questions for an owner reviewing their own operation. They are not findings, and the studies did not test any of them. They follow from where the negative reviews in the data concentrate.

Start with the business’s own split. The same main-reason question can be asked of a single business’s one- and two-star reviews, and the answer can sit well above or below the industry level. A business that runs well below its industry on process complaints has a different priority from one that runs well above it.

Then look at how often the familiar complaints arise. Does a final bill differ from the first number given, and how often? When is an arrival time missed without a call, and what does a cancellation take from request to confirmation?

Finally, separate the two kinds of review. Where the main reason is the core work, the review reads as a quality problem first. Where the main reason is how the business is run, it reads as a problem with how the customer was handled, and each needs its own owner and measure. Related reading on Sales Roadmaps covers pricing discipline and sales management.

What this does not show

The studies classify what reviewers wrote. They do not verify whether a complaint was true, and they do not measure what a fix would change. The samples are high-review-count businesses in eight metros, not every business in each trade.

Six of the studies used tighter sampling rules than the first five, including a cap on locations per brand. Owner reply rates to these reviews run from 33.8 percent (towing) to 78.1 percent (plumbing). Among businesses with at least 20 labelled reviews, a business’s reply rate and its main-reason process share correlate at minus 0.14 to plus 0.27. In 10 of the 11 industries, the figure falls between minus 0.14 and plus 0.14.

All labels and main reasons were assigned by language models, and no sample was checked against human coding. The figures above are model classifications of negative reviews, stated with their denominators, and nothing more.

Contact: Get in touch with Sales Roadmaps

Frequently Asked Questions

What are negative reviews mostly about in these 11 industries?

A language model classified the main reason as how the business is run in 65.0 percent (moving) to 86.9 percent (towing and electrical) of the labelled one- and two-star reviews. Pooled over all 73,106 labelled reviews, the figure is 74.7 percent.

Do worse-rated businesses get different kinds of negative reviews?

Only slightly by main reason. The worst-rated third has 1.8 to 8.1 times the negative share of the best-rated third. Its process share differs by 5.1 points lower to 9.0 points higher, with no test run.

Which industry had the highest share of negative reviews?

Property management, where 20.5 percent of lifetime Google reviews of the sampled businesses are one or two stars. Towing follows at 12.1 percent. Electrical is lowest at 2.6 percent.

Are money complaints in negative reviews rising?

Clearly only in moving, where the share of negative reviews with a money label rose from 34.8 percent in 2018 to 48.6 percent in 2026. Money-label levels do not compare between industries, because each industry has its own labels.

How reliable is the model classification?

Two models from the same family agreed with a kappa of 0.68 to 0.88 per industry on 200 reviews each. No sample was checked against human coding, so the figures are model classifications.

Does the analysis show that fixing operations prevents bad reviews?

No. It measures what negative reviews were classified as being about. It does not test any change a business made or what that change would do to its rating.

Related studies

Kamyar Shah Fractional COO, Fractional CMO & Business Consultant

Kamyar Shah is a Fractional COO, Fractional CMO, and Executive Coach, and the founder of World Consulting Group, with over 25 years of experience helping organizations achieve operational excellence and sustainable growth. He has led 650+ consulting engagements producing more than $300M in measurable results.

See Full Bio