Rec Connect: An AI solution for forecasting recreation membership cancellations

The Problem

Community Recreation Facilities were able to establish a foundational understanding of their members at the aggregate level using descriptive statistics. However, an enhanced understanding of members, both current and potential, is critical to maximizing revenue and retention, specifically predicting active members having a high probability of cancellation within the next quarter.

The Solution

A predictive model using demographics, contract details and usage history to predict cancellations. The model was built using open source tools within existing budget.

The Impact

Rec Connect captures 76-86% of cancellations, enabling targeted email, text or phone campaigns for the at-risk group of 12,000-15,000 active members.

Lessons Learned

Seasonality impacts model performance. Tolerating false positives is acceptable to maximize capturing potential cancellations.

Launch year: 2024