Application of Survival Analysis Methods in Analysing Customer Churn at Nigerian Telecommunications Companies

Customer churn represents a major revenue risk for Nigerian telecommunications companies competing in a saturated South West market, and survival analysis provides a methodologically superior framework for modelling churn timing and identifying predictors compared to cross-sectional logistic regression. This study applies survival analysis methods to analyse customer churn at a leading Nigerian telecommunications operator using six months of anonymised subscriber event data covering 50,000 customers in Lagos and Abuja operational zones. Kaplan-Meier survival curves were constructed for churn by customer segment, network type, and contract duration. Cox proportional hazards regression identified independent predictors of churn hazard. The proportional hazards assumption was verified by Schoenfeld residual tests. Median time-to-churn was 4.2 months for prepaid customers compared to 11.8 months for postpaid customers. Cox regression identified call drop frequency (hazard ratio 2.8), data speed complaints (HR 2.1), and competitor promotional exposure (HR 1.9) as the three strongest churn predictors. Customers who contacted customer service more than three times without resolution had a 3.4-fold elevated churn hazard. Monthly data consumption above 10 GB was significantly protective against churn (HR 0.41). The study provides a statistically rigorous churn prediction framework and recommends proactive retention intervention targeting high-hazard customer segments identified at 60-day post-subscription. Keywords: survival analysis, customer churn, Kaplan-Meier, Cox regression, telecommunications