Only 10.1% of beginners were still training a year after they started. Half had stopped by week 19.
Those numbers come from a study published in August 2026 in Frontiers in Sports and Active Living. Researchers followed 389,481 adults using a strength training app, including 100,709 complete beginners, for twelve months each.
The interesting part is not the dropout. Everyone in this industry knows about dropout. It is what predicted it.
What actually decides whether a new member stays?
The first 28 days. Not how hard they trained, not what they paid. How consistently they showed up in the first month. Each step up in active training days during those four weeks was linked to a 27% lower chance of quitting by week 11. The effect faded as the year went on, to 13% by week 26 and 9% by week 40, but it never reversed. The first month keeps paying out long after it ends.
One caveat before you build anything on this. These are app users logging strength workouts, not studio members booking classes. Adherence (the share of people still training at a given point) is measured the same way in both, but the setting is different and nobody has run this study in a boutique studio. Treat it as a strong signal about the shape of the first month, not as a measured result for your business.
The first month keeps paying out long after it ends.
The Run RateWe already know price does not carry that month
This is the third time we have arrived at the same four-week window from a different direction.
Yale researchers tracked nearly 12,000 fitness app users and found that paying for a premium tier lifted engagement for about four weeks, then the gap closed completely. The money bought attention, not habit.
South Korea halved the reward in its national exercise payment scheme, and participation doubled anyway. The cash was never the thing bringing people back.
So price does not hold the first month, and neither does paying people. That leaves an awkward question. If money is not what gets a new member through week three, what is?
Who is responsible for week three?
Look at what a class pack actually asks of someone. They buy ten classes. Then, every week, they have to decide again. Pick a day, pick a time, book it, show up. Ten separate decisions, made by a person who is new at this, during the exact month that decides whether they stay.
A course asks for one decision. They enrol in a six-week programme, the sessions are already in the calendar, and the same people are in the room each week.
That is a scheduling difference, not a pricing one, and it lands directly on the window the research points at.
| Class pack | Course | |
|---|---|---|
| Decisions the member makes | One per visit | One, at purchase |
| Who schedules week three | The member | You did, at signup |
| Who else is in the room | Whoever booked | The same group each week |
| What a quiet week looks like | Invisible until the pack expires | An empty named spot |
| What you can act on | Usage, after the fact | A missed session, that week |
That last row is the one operators underrate. In a pack, a member drifting away looks like nothing at all until the credits expire months later. In a course, week three has a register, and an absence has a name on it.
The software is catching up to this. Wellyx ships a Courses module on 28 September that sells a multi-week programme as a single booking, and the same company dropped its salon, spa and clinic products in August to build for gyms only. Useful, but the mechanism is older than the feature. Course-based teaching has worked this way for a century.
What we are not telling you
We looked for evidence that courses retain better than packs in real studios. Everything we found came from companies selling booking software, with no sample size and no method attached. None of it is in this article, and you should treat those numbers the same way when you meet them.
What we have is a well-powered study saying the first month predicts the year, two pieces of our own reporting saying price does not hold that month, and a plausible mechanism nobody has properly tested.
How to test it yourself
Run one six-week course against your normal class pack. Same slot, same instructor, same price per session. Then compare attendance in week eight, after both have ended and nothing is obliging anyone to be there.
You are not measuring what sold better. You are measuring who is still in the room when the thing they bought has run out. That is the number that decides whether a member reaches year two, and it is knowable in about ten weeks.