This study explores the application of the Length, Frequency, Monetary, Claim, and Profit (LFMCP) customer value model to predict customer churn in a Thai general insurance company with a primary focus on B2B clients. The research provides valuable insights into the insurance industry, particularly for organizations aiming to enhance customer retention strategies. The churn prediction model is developed in two phases. In the first phase, LFMCP variables are used to distinguish between valued and non-valued customers. In the second phase, the classification performance of five predictive techniques—k-Nearest Neighbor (kNN), Support Vector Machine (SVM), Decision Tree, Random Forest (RF), and XGBoost (XGB)—is compared. Results show that LFMCP significantly improves churn classification, with XGBoost emerging as the best-performing method. The study also investigates the importance of LFMCP variables using the Analytical Hierarchy Process (AHP). While the analysis explores varying weight configurations, findings suggest that equal weighting of the LFMCP variables results in superior predictive performance for churn classification in the insurance context. By incorporating additional predictors beyond the traditional RFM (Recency-Frequency-Monetary) model, the study introduces a novel and effective framework for churn prediction in the insurance industry. This expanded model, combined with insights from the AHP evaluation, forms the key contribution of the research, offering practical value for enhancing decision-making and customer retention strategies.