How Do Fitness Brands Find Out Why Members Leave?
Learn how to combine check-in and CRM data with surveys, exit surveys and at-risk member interviews to find why gym members leave and test fixes before they cancel.
Marketing and member experience leads at gyms and boutique studios see cancellations every month, and the count rarely comes with a reason. The early warning usually sits in the club’s own records. Attendance thins out and class bookings stop well before a member asks to cancel. Those records show who is drifting, but only asking members explains why. This guide sets out a five-step approach to fitness member retention research that pairs check-in and CRM data with surveys, exit surveys and interviews with at-risk members. With it, a team can find the drivers of churn while members are still paying and test fixes before rolling them out.
How fitness brands research why members leave
The method runs as a loop: measure, look for warning signs, ask, compare groups, then test. Each step feeds the next:
- Define retention, churn and a baseline for each join-month cohort.
- Mine CRM and check-in data for the attendance, booking and payment changes that come before cancellation.
- Ask current members why, using surveys and NPS to size issues and interviews to explain them.
- Run exit surveys and cancellation interviews with members who leave anyway, and segment every finding by member type, tier and join season.
- Test retention fixes against a holdout group of members who don’t receive them.
Step 1: Define retention, churn and the cohort baseline
Before any member is asked anything, the team needs one agreed way to count who stayed, because every later step measures change against it. Three definitions cover most needs:
- Annual retention: Members on the books at the start of the year who are still members at the end, divided by the starting count, times 100. Members who joined during the year are left out so they can’t hide losses.
- Monthly churn: Cancellations during the month divided by active members at the start of that month.
- Cohort retention: The share of members who joined in a given month who are still active at 30, 60 and 90 days, six months and twelve months.
Monthly churn compounds. A club losing 3% of members a month keeps about 69% of its starting members after a year, and one losing 2% keeps about 78%. The team also has to decide what counts as active. A member who pays but hasn’t visited since spring counts as retained in the billing system and as lost in practice, so it pays to track paying members and attending members separately.
For an outside reference point, the Health & Fitness Association’s 2025 Fitness Industry Benchmarking Report found member retention averaged 66.4% in 2024. A club comparing its own number should treat the match as approximate.
Many clubs expect losses to cluster in a member’s first three months, and cohort curves by join month are how a team tests that expectation. That is also why those curves matter more than a single annual figure.
Step 2: Mine CRM and check-in data for early warning signs
Recent absence, a fall from a member’s own usual visit or booking pattern, and changes in payments or freezes are the signals to investigate, used directly or through churn scores. Members may stop attending well before they cancel: in an older study of three US clubs, DellaVigna and Malmendier found members cancelled 2.31 months after their last visit on average, as summarized by Membership Quest, leaving a window in which outreach and research can still reach a member.
Attendance signals that predict cancellation
The most useful signals come from comparing a member against their own routine rather than a club-wide average. The usual signals to watch for are:
- Short absences: HFA guidance identifies the members at highest risk of quitting as those who haven’t attended for more than a week but did attend within the last two weeks.
- Frequency drops: A member who goes from three visits a week to one is worth treating as a possible routine break, even if they still show up.
- Booking declines: Track fewer advance class bookings and more late cancellations alongside check-in changes.
- Payment trouble: Failed card payments cause some cancellations outright and can signal that a member is already halfway out.
- Early-tenure frequency: Treat the target of 20 visits in a member’s first 60 days, which club software vendors widely repeat, as a threshold to test against the club’s own cohorts.
What churn scores and predictive models can and cannot show
Club management software now turns these signals into risk scores. Typical inputs are days since the most recent visit, average visits over recent months, days since the last purchase and length of membership, and the output sorts members into high, medium or low risk.
A score answers who is at risk, and it stops there. Two members can show the same drop in check-ins for unrelated reasons. One changed shifts at work and can’t make the 6 a.m. class anymore; the other stopped coming after a trainer they liked left. The score treats them identically, and the fix for each is different. The model’s job is to pick the members worth asking. Asking them is Step 3.
Step 3: Ask members why with surveys, NPS and interviews
Teams use behavior data to identify at-risk members and estimate when they might cancel. Turning that into reasons takes two methods with different jobs: surveys measure how common an issue is across the membership, and interviews explain why it drives people away.
Member surveys and NPS
Survey quality starts with question wording. AAPOR’s best-practice guidance says questions should:
- Avoid language that pushes respondents toward an answer.
- Cover one concept at a time.
- Stay short.
In a gym survey, a question asking how much a member enjoyed the newly renovated locker rooms assumes the answer; asking how the locker rooms compare with what the member expected does not.
Timing follows the member journey. Bain separates relationship NPS, asked once or twice a year with no trigger, from transactional feedback asked after a specific experience. For a gym, the natural triggers are the first visit, the end of the first month and day 90, the stretch where losses cluster. Bain also asks the NPS question first, before specific questions can anchor the overall rating.
Sample size matters more for NPS than many teams expect. A Quirk’s article reports that NPS needs roughly two to four times the sample of a simple top-two-box satisfaction score to reach the same precision. A club that wants to compare January joiners with long-tenured members needs that larger sample in each group, along with the incentives and cleanup that come with it.
Interviews with at-risk members
Interviews reach the members the churn flags from Step 2 already caught: still paying, attendance falling, not yet gone. The club can invite them directly with a personal message, which keeps the request in the club’s own voice and doesn’t depend on a panel.
Strella runs AI-moderated interviews with real members, each of whom talks voice-to-voice with the AI moderator at a time that suits them. The AI moderator follows the team’s discussion guide and asks follow-up questions based on what the member just said, so a vague answer like “I got busy” can turn into the specific week the routine broke. On Strella, a team can have the guide drafted from its research objectives or bring its own, and a built-in check flags leading questions before the study goes live. Hundreds of these interviews can run in parallel, and results synthesize as interviews finish. To hear how the moderator follows up, try an interview.
This kind of listening can also be ongoing rather than a single study. In fitness, [solidcore] uses Strella for continuous client understanding. A continuous read lets a studio notice when the reasons members stay or drift away start to change, rather than learning about them after cancellations rise.
When an interview touches a member’s personal health, some teams want a person in the room, since they cannot predict where the conversation will go. For those studies, Strella supports human-moderated interviews run by the team’s own moderators. Either way, the split between methods follows the question: a survey answers how many members share a problem, and the answer to why attendance broke comes from talking with the members it broke for.
Step 4: Run exit surveys and cancellation interviews
Some members will cancel before anyone reaches them, and an exit survey is the last chance to learn why. A short set of questions, each mapped to a likely driver, keeps the survey answerable in a couple of minutes:
- All drivers: What is the main reason you’re leaving? (Multiple choice, with an open “other” field.)
- Price and value: What would the membership have needed to include to feel worth what you paid?
- Location and time: How did getting to the club fit into your week, and did that change recently?
- Onboarding: In your first month, what helped you build a routine here, and what got in the way?
- Staff: Tell us about the last conversation you had with someone on our team.
- Facility: How did cleanliness and equipment compare with what you expected when you joined?
- Alternatives: Where, if anywhere, do you plan to work out next?
- Save opportunity: Is there anything we could have done that would have kept you as a member?
Price answers need a second question. A Quirk’s article on why subscribers cancel warns that “too expensive” often means “too expensive relative to the value,” and advises: “Dig deeper.” Question 2 does that work in a survey, and a follow-up interview does it better.
Timing is a design choice. A survey at the point of cancellation captures the reason a member gives at the moment of departure and is short enough to fit inside the cancellation flow. A follow-up around 30 days later allows a fuller conversation after the member has had time away from the club. A club can run the short survey at cancellation and test a follow-up interview with a subset of leavers. AI-moderated interviews suit these follow-ups well, because a member who has just cancelled may not want to talk to an employee of the club they left. An AI moderator runs the conversation voice-to-voice with no one from the club across the table, and members tend to be more candid about what went wrong.
Exit surveys also miss a large group. In a UK study by Rand and colleagues of 1,726 new members on twelve-month contracts, only 22% were still attending in month twelve, though all were still paying, as summarized by Membership Quest. Those members never reach a cancellation form, which is why the at-risk interviews in Step 3 matter.
Common churn drivers the research should test
Exit surveys and interviews work best when the team goes in with hypotheses to confirm or rule out. A Sogolytics survey of 1,004 US participants, run online in December 2024 by the survey software company, found 19% had left or switched gyms in the past year. Among respondents who had left or switched, 28% cited expensive membership fees, the leading reason. Other reasons in the survey’s attrition results:
- Switching to home workouts or online fitness: 22%
- Inconvenient location: 20%
- No longer meeting fitness goals: 16%
- Poor customer service or staff: 12%
- Limited equipment: 11%
- Cleanliness or safety concerns: 11%
Community and social connection belong on the list as a hypothesis too: in a UK study of 10,000 members, 88% of group exercise members had retained their membership as of January 31, 2014, against 82% of gym-only members, as HFA reports.
Each of these is a starting hypothesis rather than a finding for any one club. Onboarding doesn’t appear on the list, yet the heaviest losses fall in the first 60 to 90 days, when a new member either builds a routine or doesn’t. Members rarely name onboarding or community as their reason for leaving, even when attendance and class data point toward them. That gap between what members say and what they do is the reason interviews follow surveys.
Segment member retention findings by member type, tier and join season
Drivers rarely apply evenly, so each finding should be cut by segment before anyone acts on it. IHRSA’s One Million Strong study, covering about 1.5 million members in the US and Canada from 2012 to 2015, found gender made little difference to retention but age did. At 24 months, 37% of members aged 16 to 24 still belonged, against 50% of members aged 55 and older.
Members paying month to month were four times more likely to cancel than those on a 12-month minimum term. In the same study, members of independent clubs were 47% more likely to cancel than those belonging to a chain, even after age, gender and type of payment were taken into account.
Build tier, join-season and location comparisons from the club’s own records. The useful cuts are:
- Age band: Compare reported drivers and retention curves by age band.
- Membership tier and contract: Compare month-to-month with annual, and base with premium tiers.
- Join season: Plot the January cohort’s retention curve against cohorts from spring and autumn.
- Usage pattern: Separate class-goers, gym-floor members and personal training clients.
- Location: Published data compares independent and chain clubs; compare clubs within the same brand to see whether a driver is local or systemic.
Step 5: Test retention fixes with holdout groups
Segmented findings turn into hypotheses, and a fix that sounds right still needs proof before a club rolls it out. A holdout group supplies that proof: researchers randomly split eligible members, such as every member the churn signals flagged in a given month, so that most receive the new intervention and a smaller group receives nothing new. The difference between the two groups is the effect.
Intuition is a poor guide to which fixes will work. In a 2021 megastudy published in Nature and summarized by The Decision Lab, Katherine Milkman and colleagues tested 54 four-week programs on 61,293 members of a US gym chain. Under half raised weekly visits over the control group, by 9% to 27%, and impartial forecasters couldn’t predict in advance which ones would. The study measured visits rather than cancellations, so a club running its own test should track both. The metrics to watch:
- Cancellation rate at 30, 60 and 90 days after the intervention
- Weekly visit frequency against each member’s prior baseline
- Freezes and failed payments
- Dues retained per member in each group
The payoff calculation uses the club’s own numbers. The often-repeated claim that acquiring a member costs five times more than keeping one isn’t fitness-specific, so the club calculates its own acquisition cost by dividing marketing and sales spend by new joins. The value of a fix is the number of extra members retained, multiplied by monthly dues and expected remaining tenure, minus the cost of the program. Payback is the program cost divided by the monthly dues it keeps. If that payback beats what the club would spend to replace the same members, the fix is worth scaling.
Tools for fitness member retention research compared
Teams use a different kind of tool at each step above to answer a different part of the question:
| Tool type | What it reveals | Typical turnaround | Cost range |
|---|---|---|---|
| Built-in churn scoring in club software | Which members are at risk and how soon, not why; the score uses the software’s preset inputs, such as visit recency, recent visit averages and tenure | Uses accumulating club data | Included in some club management software plans |
| Survey and NPS platforms | How widespread an issue is and how sentiment varies by segment | Set by the response window the team chooses | Subscription fees, plus incentives, cleanup and the large sample needed to project results |
| Custom churn model built on the club’s data | Which behaviors in the club’s own records carry the most weight in predicting dropout, still not why | Depends on data cleanup and model build | Mostly custom-quoted |
| AI-moderated interviews | The reasons behind behavior, in members’ own words, with follow-up questions | Interviews run in parallel around the clock; results synthesize as interviews finish | Varies by study and level of service |
Getting started with a 30-day member retention research plan
The five steps fit into a month for a team that already has access to its club management data. A workable sequence:
- Week 1: Agree on definitions of retention, churn and active member, then build retention curves for the last 12 join-month cohorts.
- Week 2: Pull check-in, booking, payment and tenure data, set absence and frequency-drop thresholds, and produce the first list of at-risk members.
- Week 3: Send a short NPS and satisfaction survey to the broad membership, add an exit survey at cancellation, and interview flagged members.
- Week 4: Segment the findings, write one or two hypotheses with the driver and the cohort each targets, and launch the first intervention against a holdout group.
Strella runs the qualitative interviews in week 3 as AI-moderated conversations, and every finding links back to the member’s own words, so a member experience lead can bring the exact quote behind a hypothesis to the team that has to act on it. Teams can run studies self-serve or with hands-on help. Strella also offers full service for teams that need hands-on research support rather than another study to run.
FAQ
Why do gym members cancel their memberships?
Members most often state price as their reason for cancelling, followed by switching to home workouts, inconvenient location and no longer meeting fitness goals. Weak onboarding in the first 60 to 90 days and a lack of social ties also drive churn, though members name them less often.
What retention rate is typical for gyms?
Published benchmarks run from about two-thirds to three-quarters of members a year. HFA’s 66.4% average for 2024 is one comparison point. Match retention definitions before comparing, and set any benchmark against the club’s own join-month cohort curves.
How many members should a gym survey or interview?
Surveys need enough responses in every segment the club wants to compare, and interviews need far fewer. For interviews, a systematic review by Hennink and Kaiser found that studies of similar groups with focused questions reached saturation within 9 to 17 interviews. Interview findings explain reasons but can’t be projected to the whole membership as percentages.
What data should a gym pull from its CRM to spot at-risk members?
A gym should pull check-in history, class bookings and late cancellations, payment failures, freezes and membership tenure. The most predictive measures compare each member with their own routine, such as days since last visit and the change in weekly visits over the past month. Members who haven’t attended for more than a week but did attend within the last two weeks sit in the highest-risk group.
How do retention drivers differ by age, gender or membership type?
Age and contract type shaped retention while gender made little difference. In IHRSA’s One Million Strong data on about 1.5 million members, 37% of 16 to 24-year-olds were still members at 24 months against 50% of members aged 55 and older, and month-to-month members were four times more likely to cancel than those on 12-month terms. Join-season patterns need the club’s own cohort comparison.

