Optimizing Customer Lifetime Value (CLV) Prediction Models in Retail Banking Using Deep Learning and Behavioral Segmentation

ABSTRACT : In a customer-focused financial world, predicting Customer Lifetime Value (CLV) has become crucial for gaining an edge in retail banking. Traditional CLV models rely on simple rules or basic machine learning. These methods often miss the complex relationship between customer behavior and long-term profit. This study introduces a new hybrid modeling approach that combines deep learning (DL) techniques with behavioral segmentation to improve the accuracy, context awareness, and practical use of CLV predictions. By incorporating meaningful customer segments into an LSTM-based time model, we examine whether segmentation-informed AI performs better than traditional methods. The study uses real-world anonymized banking data, assesses models with both statistical and business metrics, and looks at the effects on strategic choices for customer retention, acquisition, and personalized marketing. The results indicate that deep learning models enhanced with behavioral clustering provide better prediction performance, greater clarity, and useful insights for CRM teams. This marks a significant step forward in AI-led customer intelligence.