Customer Segmentation of E-Wallet Top-Up Users Based on RFM and K-Means Clustering

Abstract

The rapid growth of digital payment systems has significantly increased the use of e-wallet services, creating challenges for banks in understanding customer transaction behavior and developing targeted marketing strategies. This study aims to segment e-wallet top-up customers using the Recency, Frequency, and Monetary (RFM) model integrated with the K-Means clustering algorithm. The research follows the CRISP-DM framework, covering business understanding, data preparation, modeling, and evaluation stages. The dataset consists of 143,836 bill payment transaction records collected from a government bank in North Sumatra over a two-month period. RFM values were calculated to measure customer engagement and transaction value, followed by clustering analysis to group customers based on behavioral similarity. The results identified three distinct customer segments: Silver, Gold, and Platinum. Evaluation metrics indicate that the clustering model produced stable, meaningful segmentation, providing strategic insights to support personalized marketing initiatives and to improve customer retention and service optimization.