Abstract:
In online retail, the cost to attract new customers is higher than retaining existing customers. As a result, businesses invest in predictive analytics for customer churn. However, modern churn prediction algorithms are black boxes that lack the transparency required to identify what actually causes attrition, tenure-based retention, and customer segments with a high risk of churn. Therefore, the next generation of churn analytics focus must shift from prediction to personalized retention. This study proposes a threefold approach that combines explainable AI to describe feature importance insights, survival analysis to model time-to-event data, and finally RFM (recency, frequency, and monetary) approach to segment customers based on transactional behavior. With this explainability, risk modeling, and segmentation-based approach, personalized retention strategies were proposed to reduce attrition and increase customer loyalty. We propose that by following this approach, organizations can determine why churn is likely, when intervention is most effective, and which customer groups to prioritize.
Published in: 6th International Conference on Advanced Research in Computing (ICARC 2026)
Date of Conference: 18th and 19th of February 2026
Date Added to IEEE Xplore: pending
ISBN Information:
Electronic ISBN: pending
Print on Demand(PoD) ISBN: pending
DOI: 10.48550/arXiv.2510.11604.
Publisher: IEEE
Conference Location: Belihuloya, SriLanka