<![CDATA[Background: Customer segmentation is an important strategy for Micro, Small, and Medium Enterprises (MSMEs) to improve marketing effectiveness and maintain customer loyalty. However, conventional approaches based on static data are often unable to capture the dynamic nature of customer behavior. Objective: This study aims to develop a more comprehensive customer segmentation approach by integrating static analysis based on Recency, Frequency, and Monetary (RFM) with dynamic analysis using Dynamic Time Warping (DTW) through a Hybrid Integration Matrix for Agatek Print customers. Method: This study uses transaction data collected from May 2024 to October 2025, comprising 8,808 transactions from 2,194 customers. The data were analyzed using the K-Means and DBSCAN algorithms for static segmentation and Dynamic Time Warping (DTW) for time-series pattern analysis. Model performance was evaluated using the Silhouette Score. Results: The results show K-Means outperforms DBSCAN, achieving a Silhouette Score of 0.5491 compared with 0.0103, indicating a more stable clustering structure. The DTW analysis successfully identifies dynamic transaction behavior patterns that are not captured by static approaches, as evidenced by four DTW clusters (2,135, 2, 1, and 56 customers) that do not align with the static RFM groups, including project-based and small-scale retail transaction patterns. The integration of both approaches reveals that most customers belong to segments characterized by low customer value and inactive purchasing behavior combined with project-based transaction patterns. Conclusion: The hybrid RFM–DTW approach provides a more comprehensive understanding of customer behavior and can serve as a foundation for developing more adaptive marketing strategies in MSME contexts.]]>
