This study predicted declining customers for Regal Rexnord and recommended when a sales associates should take action to improve the declining customers performance. The purpose of this study was to use historical customer data (sales orders) to provide insight into customers with a potential to have a decline in sales. Three techniques for predicting declining customers were compared (logistic regression, k-nearest neighbor, and random forest). Random forest outperformed the other two techniques and was used to develop the prediction model. The random forest model was trained and tested on a dataset of 7,418 customer. The resulting model had an AUC score of 0.762. The recommendation system was not able to provide a recommendation on the specific action that a sales association should take to improve customer performance because the data used for this model was not clean. The resulting model recommends any type of action or contact with the declining customer to help improve their sales performance.