The world’s rapid pace of development has led to an increase in competition among businesses of all aspects. Hence, besides growing to keep up with the innovation, enterprises need to take customer retention as a strategic priority. Importance– performance analysis (IPA) is a primary tool that has been widely used for customer satisfaction management. It is an effective means of assisting practitioners in prioritizing service attributes when enhancing service quality and customer loyalty. Typically, IPA treats the sample as a homogenous group, which may cause the result low validity and reliability. Therefore, in this study, IPA is proposed along with Density-based spatial clustering of application with noise (DBSCAN) and Back-propagation neural network (BPNN) to overcome the limitations of conventional IPA and improve the results obtained. A case study of a telecommunication company is presented to demonstrate the implementation and application of the proposed framework. Compared with a conventional IPA approach, the results indicated an increase in reliability and applicability in devising customer retention strategies.