<![CDATA[Customer segmentation is a crucial strategy for understanding consumer behavior and enhancing customer retention, particularly for companies operating in the Internet Service Provider (ISP) industry. One of the most common challenges faced by ISP companies is customer-related issues, especially the difficulty of retaining subscribers to continue using internet services. This study aims to segment ISP customers based on patterns of digital service usage and monthly billing using the K-Means Clustering algorithm, with a specific focus on a churn data case study. The dataset used in this research was obtained from Kaggle, namely customer_churn_prediction_dataset.csv. Eight variables representing customer behavior were selected for analysis: InternetService, OnlineSecurity, OnlineBackup, DeviceProtection, TechSupport, StreamingTV, StreamingMovies, and MonthlyCharges. The preprocessing stage involved one-hot encoding to transform categorical variables and data normalization using the StandardScaler technique. Cluster evaluation was conducted using the Silhouette Score and the Davies–Bouldin Index to determine the optimal number of clusters. The results indicate that the optimal configuration was achieved with k = 3 clusters, yielding a Silhouette Score of 0.46 and a Davies–Bouldin Index of 0.82. The resulting clusters exhibit distinct characteristics, namely passive customer clusters, low-cost customer clusters, and premium customer clusters. This segmentation provides strategic insights for ISPs in designing more targeted promotional strategies, such as focusing marketing efforts on specific customer clusters, determining retention priorities, and developing more personalized service offerings. The findings demonstrate that the combination of service usage and cost variables serves as an effective parameter for differentiating customer preferences. ]]>
