'Customer churn' is a term used by telecom service industry to denote the customer movement from one service provider to another or quit existing service without switching the provider. The customers are the main source of the telecommunication companies to make profit and the customers can easily change their operators which happens frequently, especially in the recent years. Therefore, there is an increasing competition in the market to retain and acquire customers. With this purpose, churn analysis is useful to predict the behavior of the customers and to create appropriate marketing strategies for them.

 

To prevent the churn of these customers, Turk Telekom launches new products and retention campaigns which are marketed through sales channels and call centers. Turk Telekom adjusts the budget and scope of the campaigns according to the churn probability calculated by the churn rate analysis of the previous year. The calculated churn rate is used for future campaigns for all of the customers without any classification. 

 

Current simplistic methodology for estimating the churn probability of customers for a defined period does not include distinctive raw data to analyze the customer behavior differences but the campaigns target all the customers as a whole. Several problems occur due to this targeting issue. 

 

The projects scope consists of clustering the customers and determining the right churn probability for each cluster which will optimize the retention campaign processes and result in optimized revenue and expenditure plan. Briefly, the main aspect of the project is to determine if the customers would still churn when there would be no campaigns prepared for these customers; i.e does Turk Telekom give unnecessary reductions to customers who were not likely to churn? What would be the churn rate if there were no campaigns made?