This paper experimentally investigates the application of machine learning on the demographic and behavioral characteristics of the customers to predict customer churn. The goal is to analyse the predictive strength of the decision trees, ANN, and Logistic regression models to catch potential churn customers, and to assist in providing business insights. The analysis shows machine learning algorithms consistently outperform traditional statistical models and that time in platform and number of orders are the leading predictors of churn. Demographic variables such as gender and location had a comparatively less significant impact on churn prediction. It also emphasizes the correct evaluation of the model through the in-depth understanding of “cross-validation and AUC metrics for an accurate representation” of the results. While the machine learning models exhibited superior predictive capacity to more traditional approaches, it also highlight some of the challenges of “interpretability, specifically referencing increasingly complex approaches like the neural net as an example”. Future work paths include enhancing model interpretability, leveraging real-time churn prediction platforms, and advancing the use of ML models in different sectors. Key features of this study The outcome of this study will provide detailed information on the factors affecting customer retention and the intention to repurchase.