Using RFM Model for Customer Segmentation and CLV
Abstract - Ecommerce companies in China are facing a limited amount of customer demographical information provided by eCommerce platforms, which leads to insufficient data for customer segmentation. This research proposed a solution for eCommerce companies to segment their existing clients. From historical transactions, RFM model provides three-criteria analysis results of individual customers. Using the results of RFM analysis and K-means clustering, customer segmentation clusters can be obtained. Then, the fuzzy AHP method provides evaluations of weightings of the three criteria in RFM model, and calculations cluster CLV provides valuable cluster information based on each company’s related situations.The case company provides historical transactions from 2017 to 2018, with 2776 customers’ information included. The proposed approach identifies four clusters, and each cluster differs significantly from each other in terms of RFM scores. Then, calculations of fuzzy AHP on RFM criterion conclude that monetary is the most essential and vital for the case company criterion, followed by recency and frequency. Cluster CLV results indicate that customers from the most valuable cluster (cluster number 3) are more likely to purchase multiple product lines and categories.
Keywords - RFM, K-means clustering, fuzzy AHP, customer segmentation, eCommerce, eCommerce platform, CLV.