Customer retention has emerged as a critical determinant of sustainability for small and medium enterprises (SMEs), particularly in competitive digital marketplaces where customer acquisition costs continue to rise and switching barriers remain minimal. Despite the widespread adoption of machine learning techniques for customer churn prediction, many existing solutions function as opaque predictive engines, offering limited transparency into the underlying factors influencing customer attrition. This lack of interpretability constrains managerial trust and restricts the practical translation of predictive insights into targeted retention actions, thereby revealing a persistent gap between analytical capability and operational decision-making in SME environments. To address this limitation, this study proposes an Explainable Artificial Intelligence (XAI)-driven framework that integrates predictive modeling with interpretable analytics and retention optimization mechanisms. The framework employs supervised learning algorithms, including Random Forest and Extreme Gradient Boosting (XGBoost), to estimate churn probability using structured customer interaction data derived from publicly available datasets such as the IBM Telco Customer Churn dataset and domain-specific transactional records. Model interpretability is achieved through SHapley Additive exPlanations (SHAP), enabling the identification of influential behavioral and financial indicators associated with churn risk. The effectiveness of the proposed system is evaluated using stratified cross-validation and performance metrics including precision, recall, F1-score, and Area Under the Receiver Operating Characteristic Curve (AUC-ROC). Experimental findings demonstrate that the integration of explainability mechanisms enhances decision transparency while enabling the formulation of context-aware retention strategies for high-risk customer segments. The principal contribution of this work lies in the development of a unified, interpretable churn prediction and retention optimization framework tailored to the operational realities of SMEs, thereby bridging the gap between predictive analytics and actionable business intelligence.

Keywords: Customer Churn Prediction, Explainable Artificial Intelligence (XAI), Customer Retention, Machine Learning, Small and Medium Enterprises, Business Intelligence, Predictive Analytics