The competition in product sales between the pharmaceutical industry in Indonesia is getting tougher. Market conditions and demands are also increasingly complicated and unpredictable. Therefore, the pharmaceutical industry must have strategic planning in the marketing field, where the offer is available. Hitherto, PT. Metiska Farma has applied a prediction method for the needs of production plan. However, the results of applying such forecasting method are inaccurate, because, besides being less effective, this method is done manually. In the study presented in this paper, a prediction test based on Machine Learning techniques was conducted, which was the Long Short Term Memory (LSTM) method. To test the proposed technique, the product dataset “X” was used with performance parameters of Mean Absolute Percentage Error (MAPE) and Root Mean Squared Error (RMSE). The results of the study, through evaluating the model performance of data training on data testing, showed that the LSTM value in predicting sales was 13,762,154.00 for RMSE in rupiah values, and the MAPE was 12%.