Turning gym member data into an early-warning system for retention
Working MVP across gyms in Finland and Mexico · private beta now scaling
The challenge
Independent gyms sit on a lot of behavioral data, but most of it is scattered across different systems and gut-feel conversations between coaches and owners. By the time retention risk shows up clearly, the member has usually already drifted too far to bring back. The opportunity was to build a lean retention product that could help gyms see risk earlier and act on it with confidence — not as another dashboard, but as a working tool grounded in member behavior. The harder problem was not simply "showing data." It was identifying the leading behavioral indicators that actually predict retention, and turning them into something a gym team can use day-to-day — without asking coaches or owners to interpret analytics.
Our approach
We built the first version around a behavioral model: identify the leading indicators of retention from the data gyms already generate, and design the product around those signals rather than around raw reporting. The system was designed around a few product principles: keep the underlying model rigorous, make the outputs action-oriented, and hide the analytical complexity behind plain-language surfaces. The gym team doesn't need to understand the model to benefit from it — they get a clear read on their member base and a confident answer to "who should we focus on this week?" The MVP was built and tested with gyms in Finland and Mexico under real operational constraints: messy historical data, incomplete exports, different gym systems, different membership structures, and the need to stay useful even when the underlying data was imperfect. Along the way, the product was hardened around data quality, member identity, privacy, and coach-friendly workflows.
The result
The work produced a working retention product that turns behavioral data into a practical decision-making tool for gym teams. Operators can now move from "who hasn't been here recently?" to a more useful question: "where is each member heading, and what should we do next?" The product is now being piloted with more gyms and scaling its private beta. The first version has established the core behavioral model, the operating surfaces, and the data foundation needed to keep learning from real gyms as the product expands. The retention gains will belong to the gyms using it. Our role was to turn a behavioral hypothesis into a working product system: one that helps teams see retention risk earlier, act with more confidence, and build retention around member behavior rather than hindsight.
Want a similar engagement?
Tell us about the behavior you want to move. We'll come back with how we'd approach it.