Multifactor Customer Classification model for IP Transit product

Abstract

Customer Classification has important role in Customer Relationship Management (CRM) and has been applied in many industries, such as retail and manufacturing. However, there is no single model purposely created only for telecommunication wholesale segment, especially IP Transit. This research develops a model for customer classification with consideration of all aspects of customer - company relationship. These aspects are customer value, customer loyalty, and customer risk. The main point is the suitability with real industry. To achieve this objective, we used real transactional data and appropriate method for processing data. Customer lifetime value analysis is done to measure customer value, while Artificial Neural Network is done for measuring customer loyalty, and also Ordinal Regression is done for measuring customer risk. The outputs from these three measurements become the input for clustering using K-Means. The optimal cluster is four clusters which can be retrieved from Elbow Rule on Ward's Method. From the value of distance to zero point, there is one customer in “Platinum” cluster, twenty two customers in “Gold” cluster, and ten customers in “Silver” and twenty three in “Bronze” clusters.