This study aims to provide suggestions for improvements in overcoming stock shortages of soft drink products using a forecasting method. The results of such forecasting will be compared with the forecasting methods used by the company at this time. The Autoregressive Integrated Moving Average (ARIMA) method was used in this study to improve the accuracy of demand forecasting in soft drink products (TBE 350 mL K12 Aseptic). This study used product sales data for the period January 2016 to January 2022. Based on the results of calculation and data processing, it is known that the best model is ARIMA (2,1,0) with a MAPE value of 35,966%. Meanwhile, the method used by the company has a MAPE value of 36.569%. It Shows that the ARIMA method (2,1,0) has better forecasting accuracy compared to the company's forecasting method with a MAPE difference of 0.604%.  The validation results were obtained forecasting in January 2022 with ARIMA (2,1,0) of 22,569 cartons, while the company's method was 21,194 cartons. This shows that the ARIMA method (2,1,0) is more accurate in forecasting because it has a forecast value in the January 2022 period close to the actual demand value, which is 23,193 cartons. The ARIMA model equation (2,1,0) for forecasting soft drink products in the following month is Zt = 0,494Zt-1 + 0,210Zt-2 + 0,297Zt-3