This research is motivated by the imbalance in stock provision at UMKM Madu Al-Ghifary, which is still managed manually without utilizing historical data, leading to overstock and stock shortages. The objective of this research is to develop a stock prediction system using the Double Exponential Smoothing (DES) method to improve stock planning accuracy. The research method includes collecting historical stock-out data from January 2023 to May 2025, time series data analysis, and web-based system development using the Extreme Programming approach. The prediction process applies the DES Brown model and is evaluated using Mean Absolute Percentage Error (MAPE). The results indicate that the developed system can predict stock requirements with relatively low error rates, thereby supporting more accurate decision-making in stock provision and raw material procurement. The implementation of this system contributes to improving operational efficiency and reducing the risk of lost sales opportunities due to stock shortages. Therefore, the application of the DES method in a prediction system proves to be an effective solution for inventory management in SMEs.