Articles · Fitness

277 new memberships in a quarter, and the gym is still not growing

12 October 2026

New people join every month, and the number of members stays where it was. The owner's explanation runs like this: enquiries come in overnight, they get answered in the morning, and by then the person has gone to a competitor. The vendor of an AI assistant agrees, and promises 20% more sign-ups in return for round-the-clock replies.

We would start not with an assistant but with some counting, done in a particular order. We will show it on Ostrel Gym & Studio. The club is invented: it is one of the companies we test our audit agent on, and the agent wrote the report on the club's data for one quarter. The four-step order itself suits any club that collects its fees by monthly direct debit.

First, count what the overnight wait costs. It turned out to cost nothing

The owner's complaint is half right. The club took 802 enquiries in the quarter, and 41% of them do arrive when no consultant is at work and wait for the next shift — typically 781 minutes, about thirteen hours. When a consultant is at work, the first reply goes out in 5 minutes.

What matters, though, is not how long someone waited but whether they joined afterwards. Among online enquiries, 35% of those answered within an hour joined, against 32% of those answered later. No real difference. The assistant, at 390 EUR a month and 1,800 EUR to set up, would speed up something that nothing depends on. And "20% more sign-ups" is the vendor's own figure, untested on the club's data.

Second, how many joined and how many left. Almost the same number

The club signed up 277 new members in the quarter. Over the same quarter, 270 memberships ended.

277 new memberships, 270 ended. Sign-ups are fine: the club is losing at the exit.

The owner had seen this number, but in a distorted form. The dashboard counted members who had only frozen their membership as lost: 274 where 270 actually ended. The gap is small and says a good deal: nobody had ever broken the "lost" column into its parts.

Third, who ends a membership. Not only the member

The club collects its monthly fee by direct debit. About 80 debits come back each month on the first run. The system takes it from there: it sends one automatic email and tries again on the fifteenth. If that also comes back, the system ends the membership itself on the last day of the month. Nobody phones the member at any point.

The owner was sure the front desk rang these members, from a Friday list. The desk has had no such list since last year. The calls stopped when the second administrator post was left unfilled.

In the owner's mind the routine had two halves: the system warns and retries the debit, and a person at the desk phones. The second half disappeared, and the first carried on working without a single error. Nothing failed in a way that anyone could have noticed.

Fourth, whether these people meant to leave. It does not look like it

In the last seven days of their membership, 51% of the members whose membership ended were still visiting the club. Among those who stayed, the figure is 52%. That is not how people behave when they decide to leave.

The gate left a second piece of evidence. Of the 173 members whose membership ended at the end of April or May, 48 tapped their card at the gate afterwards and were refused. That is 28%, and 91 refused attempts in all. The desk says these people mostly "had no idea". What happens to them next, and on what terms they can return, nobody at the club could say.

None of this can be seen from the owner's office

The reason a membership ended is recorded on each member's screen. It appears in no report. The owner sees a single "lost" number and explains it with what is in plain view: enquiries and reply times.

That is also why the report stops sooner than one would like. It does not say how many of the 270 memberships were ended by the system and how many by the members themselves: that needs a report with end reasons, and the club has none. What is known is that a good part of those who left never meant to. What that comes to in money, nobody yet knows.

There is almost nothing to buy here

The first step is free: name one person who owns returned debits, and give that person back the weekly phone list. By the administrator's estimate it is about 80 calls a month at about 6 minutes each — roughly 8 hours. The list adds work, it does not take any away, and it does not cancel a single payment the club makes. Its value lies elsewhere: a member reached in time keeps paying.

The second step is a decision for the owner: whether the system should end a membership before anyone has spoken to the member.

One number checks all of it — how many memberships the system ends each month. Compare it after two debit cycles with the phone list, and along the way write down how many members pay after being phoned.

The one paid item the report names is a failed-payment follow-up module: 85 EUR a month and 250 EUR to set up. Come back to it at that same point, and only if the calls prove too much for staff. It reaches the same members as the phone list, so the two benefits cannot be added together.

If it later comes to a module or an agent, three things stay with people: reinstating a membership, a dispute over a payment, and the conversation with a member refused at the gate.

Three numbers a club can pull today

1. New and ended memberships over three months, freezes excluded. The membership movement report in your membership system. Check that freezes are not being counted as losses. The warning sign is two numbers that are nearly equal, like 277 and 270. Then sign-ups are not the problem, and replies to enquiries are not the place to start.
2. How many memberships the system itself ended in a month. If there is no report with end reasons, open the member screen of everyone whose membership ended last month and sort them into three piles: ended by notice, ended for non-payment, frozen. Next to them put the number of returned debits from your payment provider's export. The warning sign is a second pile of any noticeable size — and the answer "nobody" to the question of who phoned these people.
3. How many people with an ended membership came to the gate afterwards. The entry log in your access system: refusals in the first weeks of the month, checked against the list of ended memberships. In the example, 48 people out of 173. Each of those refusals is a person who came to train and found out at the door that they were no longer a member.

The order matters: the exit first, then the entrance. The club in the example was looking for its loss at the entrance, where there was none. If yours is the other way round — few leave, and even fewer join — then enquiries and reply times are worth going back to.

All articles