Lifting the Veil: A Fresh Look at Churn Management for Season Ticket Holders Using Machine Learning

Churn management is a critical challenge for organizations seeking to retain season ticket holders (STHs). Traditional approaches have primarily relied on binary logistic regression to classify STHs as likely to renew or churn, subsequently targeting churners with retention incentives. While effective to some extent, this strategy overlooks advances in data science and the potential to optimize managerial decision-making. Leveraging proprietary data ( n  = 23,565) from an National Collegiate Athletic Association (NCAA) collegiate athletic program, we employ random forest and support vector machine algorithms to demonstrate that machine learning methods outperform the traditional logit model in predicting STH renewal outcomes. Building on this, we apply an uplift machine learning algorithm to move beyond conventional churn identification by assessing STHs’ sensitivity to retention interventions. Findings highlight that targeting individuals who are both likely to churn and receptive to retention efforts is a more efficient and economically beneficial strategy, offering actionable heuristics for enhancing the effectiveness of churn management programs.