This paper explores the prediction by applying machine learning models and performance measurement of algorithms on the Telco Customer Churn dataset. The ultimate aim would be to examine the application of machine learning to forecast the high propensity to churn and allow companies to implement a retention policy. Three machine learning models such as Logistic Regression, Decision Trees, and Random Forest have been utilized and compared against accuracy, precision, recall, and F1 score. The outcome shows that the best was 80.5% accuracy, the best precision was 79.1%, the best recall was 81.3%, and the best F1 score was 80.2% using Random Forest due to its superior handling of non-linear relationships and feature interactions. The outcome shows that the most effective approach for modeling non-linear and complex relations between data and consumer behavior would be to employ ensemble approaches utilizing the application of Random Forest. This study offers practical insights for CRM-related predictive modeling and provides an application of the use of machine learning to accomplish maximum retention and maximize campaign best. Future research can explore applications to use when applying to real-time applications of CRM, extend the research to various business areas, and explore the application using advanced algorithms such as XGBoost, SVM, and neural networks for real-time customer behavior prediction systems when accomplishing maximum predictions.