Written one- and two-star Google reviews of at least five words, up to 60 per business, from high-review-count gyms and fitness clubs in eight US metros. A language model, with no human-coded check, classified the main reason in 5,491 of them: 80.1 percent how the business is run, 19.0 percent the equipment and facility themselves, 0.9 percent unclear.
The study covers 5,497 written one- and two-star Google reviews of at least five words, drawn from the up-to-60 lowest-rated reviews retrieved per business, from 256 high-review-count gyms and fitness clubs in Phoenix, Dallas, Atlanta, Charlotte, Columbus, Denver, Tampa and Kansas City. In each metro, up to 200 listings were screened and, after filters including a cap of two locations per brand, the 40 businesses with the most Google reviews were sampled. Exclusions then left 282 businesses, of which 256 have reviews in the corpus. Review dates run from 2010-01-25 to 2026-09-30. A separate multi-label model pass assigned at least one business-process label to 83.3 percent of the 5,497 reviews.
Every figure below states its denominator. None is a share of members or customers.
The model-classified main reason
The main-reason pass was run by a language model, Claude Haiku 4.5, which chose between how the business is run, the equipment and facility themselves, and unclear. It classified 5,491 of the 5,497 reviews.
| Model-classified main reason | Share of the 5,491 reviews labelled by the main-reason pass |
|---|---|
| How the business is run | 80.1% |
| The equipment and facility themselves | 19.0% |
| Unclear | 0.9% |
A second model, Claude Sonnet 4.5, classified 200 of the same reviews. On the binary question of how the business is run against every other main reason, the two models agreed on 94.5 percent of the 200, with a Cohen’s kappa of 0.841 (95 percent confidence interval 0.74 to 0.93). It is not validation against human coding, and no sample was checked against human coding.
Process and the facility, counted label by label
A separate pass assigned theme labels, and a review can carry more than one. Some describe the business process, such as billing, cancellation or staff conduct. Others describe the equipment and the cleanliness of the facility. A single review can carry both kinds.
| What the labels cover | Share of the 5,497 reviews |
|---|---|
| A business-process label only | 63.6% |
| Both a process label and a core-service label | 19.7% |
| A core-service label only | 10.3% |
| Neither a process nor a core label | 6.4% |
The model assigned at least one core-service label to 30.0 percent of the 5,497 reviews. The multi-label pass and the main-reason pass answer different questions. One asks what a review mentions, the other asks what the model classified as the main reason, and the two are not reconciled review by review.
The model-assigned themes
The table lists every label in the taxonomy. These are themes a model assigned to what reviewers wrote, not verified events. A review can carry several labels. The shares overlap and are not summed. The label Other carries 17.2 percent of the reviews, and its contents were not characterized.
| Model-assigned theme | Share of the 5,497 reviews |
|---|---|
| Staff conduct | 43.5% |
| Cancellation process | 29.3% |
| Billing the reviewer describes as unauthorized | 28.0% |
| Sales and contract terms | 24.2% |
| Communication failure | 24.2% |
| Facility cleanliness | 20.2% |
| Equipment condition | 19.2% |
| Other | 17.2% |
| Hours and access | 10.6% |
Facility cleanliness, at 20.2 percent, and equipment condition, at 19.2 percent, are the two core-service labels. Hours and access, at 10.6 percent, is the smallest named label.
Cancellation process
The model assigned the cancellation process label to 29.3 percent of the 5,497 reviews. Both quoted excerpts describe reviewers’ accounts of cancellation steps they say they were required or told to follow.
“I’ve been trying to cancel my membership for 3 months. They won’t let you come in and cancel. You have to go through an email process.”
“he said that I had to physically sign a paper to cancel the contract. That isn’t that unusual, but he told me I couldn’t sign that day”
No contract or cancellation was checked against company records, and the study makes no finding about what any terms required.
Billing the reviewer describes as unauthorized
The model assigned this label to 28.0 percent of the 5,497 reviews. These are reviewers’ accounts, labelled by a model. Whether any charge was authorized was not checked.
“After my membership ended they kept charging me after saying there would be no renewals.”
“Canceled my membership months ago, still being charged and just noticed.”
In both quoted excerpts, the charge the reviewer objects to came after the membership had ended or been cancelled, in the reviewer’s account.
Sales and contract terms
The model assigned the sales and contract terms label to 24.2 percent of the 5,497 reviews. The quoted excerpts describe information the reviewer says was not disclosed at joining and a denied request to suspend an account.
“they failed to tell me when I joined that they would be moving their location”
“My last request for a temporary suspension of account while I travel for a couple of months , which you still have to pay for by the way, was denied”
Staff conduct
The model assigned the staff conduct label to the largest share, 43.5 percent of the 5,497 reviews. It is also one of the two catch-all process labels in the taxonomy. The quoted excerpts describe reviewers’ accounts of on-site staff conduct and instructions.
“Staff don’t greet or acknowledge you upon entering or leaving.”
“Staff gave unclear and inconsistent instructions, and when I finally came on the exact end date, there was no one available”
Disclosure: this study was produced by Sales Roadmaps, which sells operations consulting services. About the operations consulting service
The number that keeps this honest
The 282 businesses sampled after exclusions hold 136,703 lifetime Google reviews between them. Of those, 11,587, or 8.5 percent, are one or two stars, and 84.2 percent are five stars. Of the 282, 256 have at least one review in the corpus studied here.
Percentages in this study use the denominator stated with each figure. Main-reason shares are of the 5,491 reviews the main-reason pass labelled. Label shares are of all 5,497 reviews. The base rate and the five-star share are of all 136,703 lifetime reviews. None of them is a share of members or customers.
Operational questions these reviews raise
These are questions for an owner reviewing their own operation. They are not findings. The study reports what reviewers wrote and how models labelled those reviews. Related reading on Sales Roadmaps covers membership programs, sales playbooks and business efficiency.
What does it take to cancel?
The model assigned the cancellation process label to 29.3 percent of the 5,497 reviews. The quoted examples describe a reviewer saying they had to use an email process and another saying they could not sign on the day, situations the study did not count separately. An owner reviewing their own operation can compare the steps staff are instructed to describe with the written terms.
What happens in billing when a membership ends?
The model assigned the label for billing the reviewer describes as unauthorized to 28.0 percent of the 5,497 reviews. The quoted examples describe charges after a membership ended or was cancelled. An owner can ask what the billing system does on the day a membership ends.
What are staff instructed to tell new sign-ups?
The model assigned the sales and contract terms label to 24.2 percent of the 5,497 reviews. The quoted examples describe information the reviewer says was not disclosed and a denied request to suspend an account. For an internal checklist, an owner can ask what staff are instructed to tell new sign-ups about freezes, moves and renewals.
How the study was done
Google reviews were retrieved through the DataForSEO business data API, sorted by lowest rating, up to 60 per business. In each of the eight metros, up to 200 listings were screened. After filters, the 40 businesses with the most reviews per metro were sampled, with at most two locations per brand.
Thirty-eight businesses were excluded after their names were read: 13 climbing gyms, 13 children’s programs or adventure parks, 10 institutional facilities (school, university, military, hospital, non-profit and community center), 1 stretching service and 1 martial-arts school.
The corpus consists of the retrieved reviews rated one or two stars with at least five words, after exclusions and deduplication. Identical review texts posted under two listings were counted once, which removed 5 reviews. Candidate review-theme codes were proposed by a language model, Google Gemini 3.1 Flash-Lite, on a seeded sample of 200 reviews, and the research team consolidated them into a fixed label set.
Labels were then assigned by the same model. A second model, DeepSeek V3.2, relabelled 200 reviews and agreed on process against not process in 92.5 percent of them, with a kappa of 0.768. Each label was asked to carry a verbatim span from the review, and 96.1 percent of labels carry one. The rest are counted without a span.
As a label-based sensitivity test, which is not a main-reason result, the two catch-all process labels, staff conduct and communication failure, were ignored. On that basis, 58.8 percent of the 5,497 reviews still carry at least one process label. Reweighted by each business’s lifetime one- and two-star count, the any-label process share is 84.4 percent. With equal weight per business it is 80.8 percent.
The stored review data carry a business key, the star rating, the date, the text and an owner-reply flag only. No reviewer names were stored. No reviewer or reviewed business is named in this article.
Limitations
Google reviews are self-selected. The sample covers high-review-count businesses in eight metros: after filters including a cap of two locations per brand, the 40 businesses per metro with the most Google reviews among up to 200 listings screened, before exclusions. Review dates run from 2010-01-25 to 2026-09-30, and the sample is sorted by rating, not by date, so the study makes no claim about trends over time.
All labels and main reasons were assigned by language models, and no sample was checked against human coding. The label Other holds 17.2 percent of the reviews, and its contents are not characterized here.
Contact: Get in touch with Sales Roadmaps
Frequently Asked Questions
In this study, which model-assigned themes are most common?
Staff conduct, which the model assigned to 43.5 percent of the 5,497 reviews studied. Cancellation process follows at 29.3 percent and billing the reviewer describes as unauthorized at 28.0 percent. A review could receive more than one label, so these shares should not be summed.
In this study, what share of reviews was classified as mainly about the equipment and facility themselves?
In the main-reason classification, 19.0 percent of the 5,491 labelled reviews were assigned to the equipment and facility themselves and 80.1 percent to how the business is run. At least one core-service label was assigned to 30.0 percent of the 5,497.
In this study, how often did the model assign the cancellation process label?
The model assigned the cancellation process label to 29.3 percent of the 5,497 reviews.
Among the businesses sampled, what share of all reviews are one or two stars?
Among the 136,703 lifetime reviews held by the 282 businesses sampled, 8.5 percent are one or two stars and 84.2 percent are five stars. That is a share of reviews, not of members.
How closely did the models agree?
On the binary main-reason question, how the business is run against every other main reason, a second model agreed on 94.5 percent of 200 reviews, kappa 0.841. For the theme labels, process against not process, a separate second model agreed on 92.5 percent of 200 reviews, kappa 0.768. No sample was checked against human coding.
Does the study verify whether what reviewers wrote is true?
No. It reports what reviewers wrote and how often the model assigned each theme. No charge, contract or cancellation was checked against company records.
Related studies
- The negative dental reviews study
- The negative plumbing reviews study
- The negative electrician reviews study
- The negative law firm reviews study
- The negative auto repair reviews study
- The HVAC profit benchmark review
- The negative veterinary reviews study
- The negative pest control reviews study
- The negative moving company reviews study
- The negative towing company reviews study
- The negative property management reviews study
