Analyzing player churn in esports games: a survival analysis of league of legends log data

Purpose Understanding player churn behaviors is essential for the long-term sustainability of esports ecosystems. This study investigates the key factors influencing churn among players of League of Legends (LOL), focusing on behavioral, performance and social predictors. Design/methodology/approach Using a dataset of 295,008 LOL gamers, the study applies Cox regression survival analysis to examine the effects of four dimensions of player churn: game engagement level (GEL), game performance level (GPL), play variety level (PVL) and social play level (SPL). The study also investigates temporal patterns to identify critical churn windows, offering practical insights for retention strategies. Findings The analysis shows that churn probability increases by 8% after 60 days and 20% after 90 days, with role-based variations. Within GEL, only play level reduced churn, while total number of matches and time gap per team match had minimal impact. Surprisingly, higher Game Tier increased churn, likely due to tier-related frustration. All six GPL indicators lowered churn, supporting previous research. PVL findings were mixed – favoring preferred positions increased churn, but champion diversity decreased it. Social connectedness reduced churn overall, though playing with teammates was associated with lower win rates, which may negatively influence player retention. Originality/value This study is original in that it introduces socio-psychological factors into churn analysis in esports by focusing on engagement, performance, variety and social play. In addition, the findings offer recommendations for players, such as maintaining motivation through diversified play and social participation, which may help prevent churn.