What question did this study set out to answer?

The aim is to develop a predictive framework that accurately forecasts customer lifetime value and future spending.

April 19, 2026

Predictive Customer Lifetime Value: Transitioning from Segment-Based Classification to Probabilistic Revenue Forecasting

Key Points

The aim is to develop a predictive framework that accurately forecasts customer lifetime value and future spending.
Developed a hybrid model combining BG/NBD for customer activity estimation and XGBoost for spending predictions.
Utilized probability estimation to improve upon traditional segmentation techniques.
Implemented an analysis of customer behavior to enhance future revenue forecasting.
Demonstrated increased accuracy in predicting customer spending compared to traditional segmentation methods.
Provided a more dynamic approach for businesses to understand and forecast customer value.
Indicated that businesses can benefit from moving towards probabilistic models for revenue management.

Abstract

Earlier approaches to Customer Lifetime Value (CLV) have largely focused on segmentation techniques such as RFM analysis combined with clustering methods like K-Means 1. While these methods are useful for identifying customer groups, they do not directly estimate how much value a customer will generate in the future. In this paper, we propose a predictive framework that moves beyond descriptive segmentation toward revenue forecasting. The approach combines the BG/NBD probabilistic model to estimate whether a customer is still active 2 with an XGBoost regression model to predict future spending 3. This hybrid setup allows businesses to move from static customer grouping to a more dynamic and practical forecasting system.

Mark Helpful

Bookmark

Relay