Businesses increasingly rely on Customer Lifetime Value (CLV) metrics to inform strategic customer engagement and retention strategies. However, a systematic literature review reveals a significant gap in the integration of customers’ risk factors into CLV calculations. Despite the large amount of customer data collected by companies, risk-adjusted CLV predictions using Machine Learning (ML) have been largely underexplored. This study addresses this gap by proposing a novel set of Risk-Adjusted Return (RAR) metrics tailored to noncontractual (B2C) settings in the telecommunications industry. Using the Cross-Industry Standard Process for Data Mining (CRISP-DM), ML models are designed to incorporate customer churn probability and beta value into the discount rate for CLV calculations. Four RAR metrics are introduced and validated using eXplainable Artificial Intelligence (XAI) techniques. The results demonstrate the distinctiveness of the RAR approaches, with high accuracy in churn prediction (85%) and strong RAR model performance ([Formula: see text] of 92% and MAPE of 20%). XGBoost shows superior performance in churn prediction, while CatBoost excels in RAR prediction. Key features influencing RAR include loyalty points, revenue metrics, churn probability, and beta value, consistent with traditional RAR calculation factors.