From Routes to Retention: How Adamanta Turned Climber Churn into a Data Driven Growth Engine

When you walk into an Adamanta gym on a busy evening, it feels like the future of climbing in Latin America, walls packed with routes, music, competitions, and a community that keeps growing every month.

But behind that energy, the leadership team was facing a problem shared by almost every fitness brand in the region: people were signing up, climbing for a few monthsand then disappearing.

Adamantadidn’t just want more members. They wanted lasting members. Like many highgrowth fitness operators, Adamantahad expanded faster than its data infrastructure.Multiple locations across Mexico. A growing mix of memberships, drop-ins and events. A legacy ecosystem of CRM, POS and operational tools that didn’t “talk” to each other.

Information lived in different systems and spreadsheets. Leadership could see monthly churn totals, but not the behavioral patterns behind them: Who was slipping away? At what point? Andmost importantlywhy?

Industry data already painted a worrying picture: in Mexico, roughly 7 out of 10 people abandon the gym within the first months of the year. Adamanta could feel that same pattern in their own numbers, but lacked of a reliable way to quantify, predict and act on it.

The question they brought to Aconcagua was simple and ambitious: “Can we predict which climbers are about to churnand act before they leave?”

The Business Challenge: Growth Without Visibility

Like many highgrowth fitness operators, Adamantahad expanded faster than its data infrastructure. Multiple locations across Mexico, a growing mix of memberships, dropins and events, and a legacy ecosystem of CRM, POS and operational tools that didn’t talk to each other.

Step 1 – Building a Unified Data Backbone

Our first move was not to train a model, but to clean up the data. Together with Adamanta, we centralized three core sources inside Google Cloud Platform, using BigQuery as the analytical backbone: customer & membership data, weekly activity data, and membership lifecycle history.

Step 2 – Discovering the Behavior Behind Churn

With the data ready, we moved into exploratory analysis. We confirmed what many gym operators intuitively feel, but rarely quantify: churn is overwhelmingly behavioral, not demographic. The real signals of risk came from recency, frequency, tenure and spending patterns.

Step 3 – Segmenting the Climbing Community

Before predicting churn, we wanted to understand who Adamanta was serving. Using dimensionality reduction (PCA) and clustering techniques, we grouped thousands of climbers into five actionable segments based on real behavior, from Elite Clients and Champions to AtRisk members.

Step 4 – Two Models, One Retention Engine

Rather than relying on a single algorithm, Aconcagua designed a dual-model strategy to see churn from two complementary angles. The first model, built with LightGBM, looks at consolidated variables per customer to produce a global risk score.The second model, powered by Random Forest, tracks week-by-week changes in behavior, acting like an early-warning system.

Step 5 – From Predictions to Playbook

Predictions alone don’t keep anyone on the wall. Actions do. Aconcagua embedded the models directly into Adamanta’s cloud data ecosystem, adding a churn probability field that updates as new weekly data arrives. These scores feed into a suite of Power BI dashboards and are operationalized through a Churn Analytics Playbook that defines segments, plays and responsibilities across teams.

Step 6 – A 12-Week Path from Idea to Pilot

The entire engagement followed a structured 12-week roadmap: kickoff and data ingestion, EDA and business understanding, baseline model and segmentation, dashboard and playbook development, and finally user acceptance testing and pilot rollout.

The Impact: From Guesswork to Predictive Strategy

While churn is an ongoing battle, the early impact of the project is already clear in Adamanta’s strategic planning: projected churn reduction, higher customer lifetime value, more efficient marketing spend, and better facility planning. Most importantly, Adamanta now sees churn as a manageable behavior they can anticipate and influence.

Why This Matters Beyond Climbing

Although this case was built with a climbing chain, the blueprint applies to any membership or subscription-based business: fitness clubs, co-working spaces, education platforms and subscription services. If your business depends on people coming back, you face the same questions Adamantahad,and the same kind of data-driven strategy can help you answer them.

Ready to Climb Your Own Data Wall?

At Aconcagua, we don’t just build models; we build decision systems that empower your team. Whether you’reoperatingone location or thirty, we can help you centralize scattered data, design churn and retention models, deploy dashboards and playbooks, and lay the foundations for a scalable cloud analytics ecosystem. If you’d like to explore how this kind of solution could look for your business, reach out to Aconcagua and let’s start mapping your route.