Accurate production prediction is essential in product sales efforts, especially food products whose raw materials have a short shelf life. This paper aims to present a system application model based on the Neural Network algorithm to predict the number of Siomay sales in the future, as a reference for preparing raw materials appropriately. The prediction uses historical data as system training data. The Neural Network trial used 357 historical sales data, 7 initial data used as references, 315 data as training data, and 35 latest data as test data. The neural network input variables were the average sales of the previous 7 days, sales value 1 to 3 days before, the end of the month, identification of discount/benefit days, and weekends. This research methodology includes data collection, pre-processing through data normalization to a scale of [0, 1], and designing a neural network architecture consisting of an input layer, a hidden layer, and an output layer. The Backpropagation algorithm was used to train the network by iteratively updating weights to minimize error values using the Mean Squared Error (MSE). Test results show that the BPNN model is capable of recognizing demand patterns with a high degree of accuracy. Optimal parameters such as learning rate, number of epochs, and number of neurons in the hidden layer significantly influence convergence speed and prediction accuracy. This system is expected to be a management tool for making more accurate and efficient inventory procurement decisions.