Customer churn analysis through Artificial Neural Networks in Turkish telecommunications market

Abstract

As they maintain their competition with the other players in the sector, the companies should thoroughly analyze the customers that may have been or are likely to be lost in order to prevent their existing customers from being oriented to another competitor. Given the fact that, in numerous sectors, the cost of the acquisition of new customers is fairly higher than that of retaining the existing customers; the customer churn analysis comes into an even further prominence. In the telecommunications market, characterized as one, where the competition is most intensive and severe; the prevention of customer churn becomes even more important now that number porting is allowable. Therefore; the capability to predict the potential customers, who pose churn risk, and to determine the strategies to detract them from the said inclination implies a considerable advantage in the sector. This paper intends to predict the customer churn in the telecommunications market, which is characterized with rapid customer churns, through Artificial Neural Networks (ANN), to identify the variable effective in customer churn, and to introduce the pattern of events in the churn of customers. To that end, the customer churn prediction via ANN was analyzed through MATLAB 11 software with the use of the variables of demographic information, call detail records, billing details, income and education levels, extra service details and satisfaction/ complaint details of 400 mobile phone subscribers in Turkish telecommunications market, gathered by survey. The findings so obtained were then assessed through a comparative approach. © International Economic Society.