Reducing Customer Churn for XL Axiata Prepaid: Factors and Strategies

Abstract

Understanding and mitigating customer churn is pivotal for companies, especially those operating with subscription models like telecommunications firms. This study focuses on XL Axiata Company, delving into the factors driving customer churn. The data used is from a questionnaire and focuses on consumers who have used prepaid cards and are considered churn customers, while customers who continue to use cards are termed non-churn customers. Using machine learning algorithms such as logistic regression, ANN, and XGBoost, the data is applied in the prediction step of customer churn classification. The best machine learning method's coefficient results will be followed by strategy analysis using the QSPM method and risk analysis of the loss distribution using the CVaR method. According to the findings of this study, ANN is the most accurate machine learning method, and network factors are the most important factors in customer churn, followed by level of interest in VAS products, company services, failed calls, customers who have made calls to call service, package prices that are fairly expensive, and ads that are less attractive. The QSPM strategy study indicated an AI/ML approach to examine the bundling promo with VAS products, taking into account the impact on customer churn and implementation costs. The CVaR risk analysis results reveal that the VAS products that can be prioritized in the VAS promo plan are bundling products in the form of primary quotas and online games, which are more profitable than bundling with video streaming or chatting.