Incorporating Loyalty and Periodicity into an Data-Driven RFM Customer Segmentation Model: An Empirical Study in the B2B Retail Sector

Machine learning advances have transformed customer relationship management (CRM), especially through data-driven customer segmentation. A commonly used approach is the Recency–Frequency–Monetary (RFM) model, which segments customers based on how recently and frequently they purchase, and how much they spend. Although effective, RFM does not fully capture behavioral patterns critical in specific business contexts. To address this limitation, this study extends the model by introducing two additional behavioral dimensions: Loyalty, which measures the stability of purchase values, and Periodicity, which reflects the regularity of purchase intervals. The resulting RFMLP model was applied to real transaction data from a computer accessories business-to-business (B2B) distribution company. Customer segments were generated using k-means clustering, and segmentation quality was evaluated with silhouette scores. Experimental results indicate that RFMLP enables more nuanced customer segmentation compared to the traditional RFM model, providing practitioners with enhanced insights for targeted marketing and strategic decision-making.