Customer retention is a critical challenge for companies in an ever-changing environment. This project focuses on a leading certification services organization in Colombia with an extensive portfolio and an average of 25,000 customers. The organization faces the challenge of anticipating customer needs and behaviors, with certificate renewal being a key indicator of satisfaction and continuity in the business relationship.The main objective of this study is to develop a predictive model to identify customers with a high probability of not renewing their certificates. Using the Cross Industry Standard Process for Data Mining (CRISP-DM) methodology, historical customer data is analyzed, relevant features are selected, and several classification models are compared to predict the probability of renewal. The evaluation of the models includes the optimization of classifiers and the evaluation of their performance using the confusion matrix, a tool that allows measuring the accuracy of the predictions. The classification model with the best performance to identify customers with a potential risk of leakage is decision trees, with an accuracy of 98%. The predictive model provides the organization with an effective tool to anticipate certificate renewals, allowing the implementation of customized proactive retention strategies. These actions may include loyalty campaigns and adjustments to service protocols, tailored to the specific needs of each client.