Telecom providers are known to offer value-added services as a subscription add-on to the standard core services. In the competitive environment today, a telecommunication company feels that the best approach to measure and improve the business is to be as data-driven as possible. This research aims to develop a model to automate the prediction of average revenue income per user (ARPU). Work done in this paper uses information from the sample business databases which yield about 114,080 rows of data. Several models were built and studied to compare their performances. Four important target variables were considered to appraise the performance of the models. The paper presents two potential strategies to raise the model acceptance rate and considerably increase the model’s accuracy.