<![CDATA[Indonesian water utilities face persistent challenges in managing payment delinquencies due to diverse customer characteristics, geographic limitations, and inadequate analytical capabilities. Addressing this issue is essential to optimizing revenue collection and supporting sustainable operations. This study aims to develop a data-driven customer segmentation framework using K-means clustering to enhance delinquency management. The framework incorporates six engineered features—Debt Efficiency, Payment Behavior Score, Category Risk Score, Geographic Risk Score, Consumption Intensity, and Financial Risk Score—designed to capture customer payment behavior, consumption patterns, and geographic risk. We applied the model to 1,500 anonymized customer records from PT Air Minum Giri Menang, focusing on those with delinquencies exceeding four months. Risk scoring was based on quintile distribution, and optimal clustering was determined through the elbow method combined with silhouette coefficient analysis. The results produced a two-cluster solution (silhouette score = 0.538), showing statistically significant differences across features (p ¡ 0.001) and medium-to-large effect sizes (Cohen’s d = 0.52–2.12). The segmentation identified medium-risk customers (86.7%) who require preventive management and high-risk customers (13.3%) who need billing intervention. Urban areas exhibited higher delinquency risk (18.4%) than rural areas (2.5%), indicating the need for geographically targeted strategies. All customer data was anonymized following Indonesian data protection protocols. In conclusion, the proposed framework transforms manual billing supervision into an adaptive, data-driven management system, contributing to segmentation research by introducing utility-specific engineered features for Indonesian water utilities.]]>
