Almost nobody decides to quit the gym on the day they cancel. They stop coming, then they stop thinking about it, and then a renewal date arrives and they let it pass. By the time you get the message, the decision is weeks old.
Which is good news, in a way. A decision made over weeks leaves weeks of evidence. The evidence is attendance, and most gyms already have it.
Establish what normal looks like, per member
A member who comes twice a week and keeps coming twice a week is not at risk. A member who came five times a week for a month and now comes twice is in trouble, even though twice a week sounds fine.
So the number that matters is not visits per week. It is visits per week compared with that member's own baseline. Take the first four to six weeks after joining as the baseline — that is the period when motivation is highest and the pattern they intended to keep is visible. Everything after that gets measured against it.
The four signals worth watching
1. A sustained drop against baseline
One quiet week means nothing. People travel, get ill, have a deadline at work. Two consecutive weeks at half the baseline or less is different. That is the point where a habit has broken rather than paused, and it is the earliest signal that is reliable enough to act on.
2. A long gap
Count days since the last check-in. For a member whose baseline is three or more visits a week, a gap of ten days is a strong signal. For a once-a-week member the threshold is longer, maybe three weeks — which is exactly why the baseline matters. A single gap threshold applied to every member will flood you with false alarms from the light users and miss the heavy ones entirely.
3. The first ninety days
New members churn at a different rate from established ones. Someone six weeks in who has started drifting is a different problem from someone two years in who has started drifting — the first has not yet built the habit, the second has lost it. Both are worth contacting, but not with the same message, and the new member is the more urgent of the two.
4. Proximity to the renewal date
Attendance decay matters more the closer it sits to a renewal. A member who goes quiet three weeks before their membership expires is the highest-priority conversation in the building. A member who goes quiet the week after renewing has time to recover, and you have time to help them.
Combine the signals rather than ranking by any one of them. Falling attendance, inside the first ninety days, with a renewal approaching, is a different level of risk from any of those alone.
Why cancellation surveys arrive too late
Asking a member why they cancelled tells you about a decision that is already made. The answers also tend to be polite rather than true — “too busy” is the socially easy reason and covers everything from cost to intimidation to a trainer who left.
The attendance data does not have that problem. It does not tell you why, but it tells you when, and when is the part you can act on.
What to do once a member is flagged
The intervention matters as much as the detection. A few things that tend to work:
- Make it a person, not a broadcast. A message from a trainer the member actually knows lands differently from a bulk SMS. If you have to choose between contacting fifty members automatically and ten members personally, choose the ten.
- Ask, don't sell. The opening should be a question about how they are getting on, not an offer. Discounts offered to a member who has lost the habit buy a month and rarely more.
- Offer a concrete next session.“Come back soon” is not actionable. A specific slot, with a specific person, on a specific day, is.
- Check whether something changed. A trainer leaving, a class moving, a shift pattern changing at work — these are fixable, and the member usually will not raise them unprompted.
Watch the cohorts, not the headline number
Total membership hides a great deal. A gym that signs twenty members and loses eighteen looks stable month to month and is not.
Group members by the month they joined and track what share of each group is still active at three, six and twelve months. This tells you whether a change you made actually worked, because you can compare the cohort that joined after it with the ones that joined before. Headline membership can never answer that question.
Start simple
You do not need a model. A weekly look at two lists gets most of the value:
- Members whose attendance over the last fourteen days is below half their baseline.
- Members with a renewal in the next thirty days who appear on the first list.
The second list is short and it is where the conversations should start.
Nexgrit includes a retention report that lists members whose attendance is falling. Whether you use software for it or a spreadsheet and a Monday morning, the principle is the same: the information that predicts a cancellation is already in your check-in records, and it is there weeks before anyone tells you they are leaving.