CARRE Framework Optimizes Churn Retention with Counterfactual Reasoning

CARRE combines retrieval-augmented generation with cost-aware counterfactual scoring to recommend customer retention actions. On the IBM Telco dataset, it achieved a 79.8% greater mean risk reduction compared to the SHAP baseline by grounding explanations in specific feature transformations.