<p>In increasingly competitive markets, maintaining customer loyalty—particularly during the post-warranty period—has become a critical challenge for service-oriented organizations, especially in the automotive industry. Once formal manufacturer obligations expire, customers’ decisions are shaped more strongly by their actual service experience and perceived value rather than contractual commitments. This study aims to identify and explain the key determinants of customer loyalty in the post-warranty stage using explainable data mining techniques. First, potential loyalty-related indicators were identified through a fuzzy Delphi process involving industry experts. Subsequently, survey data collected from customers of Iran Khodro dealerships were integrated with their actual service records to construct a comprehensive analytical database. To predict and analyze customer loyalty, three machine learning algorithms-Decision Tree, K-Nearest Neighbors (KNN), and Naïve Bayes-were implemented. Model performance was evaluated and compared using standard classification metrics and cross-validation techniques. The findings indicate that the Decision Tree algorithm based on the Gain Ratio criterion outperforms the alternative models in terms of both predictive accuracy and interpretability. Feature importance analysis further reveals that transparency in the repair process, positive service history, technical expertise of personnel, customer awareness of dealership service benefits, and knowledge of warranty conditions are the most influential factors affecting post-warranty loyalty. Beyond demonstrating the practical value of explainable data mining in customer behavior analysis, the results provide actionable managerial insights for enhancing after-sales service strategies and strengthening long-term customer relationships.</p>