Analisis Segmentasi Pelanggan pada Bisnis dengan Menggunakan Metode K-Means Clustering pada Model Data RFM

Abstract

The development of business strategies, particularly in the marketing of SMEs, requires the utilization of business intelligence as the foundation for objective decision-making. This research aims to develop a business intelligence scheme for SMEs and design targeted assistance strategies for SME support institutions. The implementation of business intelligence involves leveraging transactional data from SMEs to ascertain customer segmentation and correlating it with Customer Relationship Management (CRM) strategies. Transactional data is processed into a Recency, Frequency, Monetary (RFM) data model. Customer segmentation is achieved through a clustering process using the K-Means algorithm, and the results yield distinct profiles for SME customers. Evaluation processes are conducted to determine the optimal solution for the number of customer segments. Evaluation methods, including the Elbow Method, Silhouette Scores, and Davies–Bouldin Index, are employed to determine the optimum cluster. The evaluation results indicate that the optimum cluster is 3, with the best Silhouette Score being 0.548 and Davies–Bouldin Index at 0.76. The first customer segment exhibits the highest shopping frequency and monetary value, categorizing them as active and profitable customers. Special loyalty services are recommended for this segment. The second segment, despite having the largest number of customers, exhibits a shopping frequency of only 1-2 times, with an average recency of approximately the last 2 months. These customers require effective after-sales service. The third segment consists of customers who last shopped more than 6 months ago, making them a low-priority segment. Re-engagement strategies, such as email marketing, are suggested for this segment. Support institutions can focus on CRM assistance targeting these three identified segments.