Churn erased $1.6 billion in a day. The cause was the front desk and billing.

Planet Fitness beat consensus on revenue and on earnings on May 7, 2026, and the stock fell 31% that day. The miss was net member joins.

Net joins is joins minus cancels, the one number the model actually runs on, and it appears on no line the company reports. When it came in light, full-year guidance fell from 4-to-5% same-club sales to about 1%, and roughly $1.6 billion of market value vanished between two closes. Every reported metric was green.

A soft join number is existential because of the shape of the business. The franchise segment earns a 71.9% margin on a royalty levied on member dues, not on workouts. A member is an annuity. Management named the cause as higher-than-expected attrition and could not explain it. The explanation was not missing. It was unread.

Two questions decide churn. The first: what makes a member write a one-or-two-star review. The second, what makes them actually cancel, comes in section 04. They do not have the same answer.

Holding club, period, reviewer type and review length constant, each operational failure was tested as a cause of a one-or-two-star review, estimated twice by two independent methods, checked three ways that a false cause would fail to reproduce, and confirmed by a model that strips out every fixed difference between one club and another.

A rude or absent front desk raises the chance a member turns hostile by 36 points (26 to 52), the largest effect in the data. Billing failures and cleanliness each add about 25. Broken equipment, the complaint members raise most often, adds only 5. These measure anger, not exit. Anger is the front desk. Whether that anger becomes a cancellation is the second question, and its leader is billing.

Rank the six failures two ways: how often members raise them, and how much each one actually causes an unhappy member. The two lists come out nearly upside down.

Broken equipment is the complaint members raise most, and it anchors the capital debate: the extended replacement cycle, the remodels, the recovery amenities. In a club fixed-effects model it also has the smallest effect of the six, an odds ratio of 2.0 against the front desk's 35.3.

The front desk is the reverse: raised less loudly, close to free to fix, and the single largest cause of an unhappy member. The investment is aimed at the noise, not the damage. The highest-return move in this analysis is not a machine. It is a trained front desk and a billing system members can trust.

Here the ranking flips. The front desk causes the most anger, but anger is not exit. Billing is only the third-largest cause of a bad review, and by far the largest cause of a member announcing they are done.

A member who raises a billing or cancellation problem says they will leave or warn others 50% of the time, 36 times the fleet baseline and nearly four times the rate for the front desk (13%). The mechanism is not that cancelling is too easy. It is that the recurring billing is not trusted.

This is the wound the company keeps touching. It made cancellation frictionless and, in its own words, watched attrition rise; the instinct will be to add the friction back. The data points the other way. Members are not leaving because the exit is easy. They are leaving because the billing cannot be trusted. The fix is the trust, not the exit.

The failures are not spread evenly. Strip the inflated ratings away and the fleet is two different companies at the same price.

Same brand, same fifteen-dollar membership, same eighteen months, same metro. The Outer Suburban clubs average 4.84 stars with a 3% one-star share. The Urban clubs average 3.46 with a 34% share, and their broken-equipment, cleanliness and safety rates run ten to twenty times higher. About a fifth of the fleet sits in areas the government calls low income, exactly the members a surprise charge or a rude desk drives away for good.

The reason none of this reached the boardroom is the instrument. The figure leadership watches is part memory and, for nine months, part marketing.

Between February and October 2025 about 4,903 five-star reviews arrived above the normal rate, more than half of everything on record, while the count of unhappy members never moved. It was a review-generation burst, not a change in the gyms: a scraper cannot pick only happy reviews, and January, the busiest sign-up month, is the quietest for reviews. The displayed rating rose on solicited applause, then fell when it stopped, telling leadership nothing true either way.

Underneath, the honest signal is buried. Google shows one all-time average per club built on 36,279 reviews; the last eighteen months are a quarter of that. At the weakest urban club the door still reads 4.1 while recent members give it 2.8. A metric that cannot move cannot warn. The count of who writes, and why, moved months before the rating did.