<p>Reasonable inventory management is important for a company’s development, and accurately forecasting retail demand is a key component of inventory planning. Using a cross-border e-commerce company as a case study, this paper analyzed its current inventory management state. Then, several quantitative forecasting methods commonly utilized today were presented. Building upon this foundation, a back-propagation neural network (BPNN) model, integrated with the sparrow search algorithm (SSA), was developed to predict retail demand for Product S. Among the various quantitative forecasting methods, the weighted moving average method performed best in prediction, although it was not as effective as BPNN. The SSA-BPNN model achieved a mean absolute error (MAE) of 2.56%, a root mean square error (RMSE) of 3.21, a mean absolute percentage error (MAPE) of 1.89%, and an R² value of 0.97 for retail demand forecasting of Product S. When the model was applied to inventory optimization for Product S, the inventory turnover speed was improved, and the inventory costs were reduced. The results verify the effectiveness of the SSA-BPNN-based retail demand forecasting approach in inventory management. It can be applied in real-world cross-border e-commerce.</p>