This paper presents an optimization model for the strategic localization of cold chain distribution centers, integrating both economic costs and delivery reliability. The proposed approach combines K-Means clustering for territorial segmentation with a Gaussian kernel-based reliability estimation, which subsequently solves a multi-objective optimization problem to minimize costs while maximizing reliability. When applied to the Casablanca region, the model demonstrates a significant improvement in network performance, with delivery reliability increasing from approximately 10% after the territorial reorganization. Certain zones, such as Casa 4, now exceed 66%, whereas others, including Casa 10, require further targeted adjustments. In addition, the approach enables a substantial cost reduction, decreasing from 20,026 MAD to around 10,062 MAD, thereby confirming its economic efficiency. Overall, the methodology provides a simple, data-driven, and reproducible decision-support tool that can be further strengthened through sensitivity analysis and real-time monitoring to support future planning and operational implementation.