Customer Risk Identification And Measurement In Semi-Contract Context

At the age of customer-centric relationship marketing, customer assets become increasing important to enterprises. Because of the vulnerability and volatility of cash flow income, customer risk exists in enterprises with objectivity and inevitability. Compared with the non-contract and contract context, the uncertainty of the customer behavior in semi-contract context is complex, and it affects the type and factors of risk largely.Therefore, the research of the formation mechanism and measurement of customer risk is a challenging and novel direction in the research field of customer relationship management. The main results of this research include the following three points.(1) Base on the customer behavior, this research identified and analyzed the customer risk in semi-contract context systematically. Existing literature does not studied behavioral characteristics of semi-contractual customer. But in this paper, we analyzes the customer specific scenarios half contract buying behavior and churn, and then degree of promise breach is identified to be the risk in semi-contract context. The risk factors are Customer revenue uncertainty and customer churn uncertainty. In addition, we also analyze the characteristics of customer risk of different semi-contractual type.(2) Based on P coefficient, we propose the measurement ways of customer risk and factors. Existing quantification methods of customer risk are data mining and comprehensive evaluation. In order to overcome these two shortcomings, this research constructs a quantitative model to measure customer risk and factors, and the result means the level of a customer’s risk relative to the average of groups.(3) Through an improved Bayesian network, build measure models of customer risk. The theory mechanism cannot be determinate in the traditional Bayesian networks, and it cannot handle all types of data. In this paper, the network is built whit theory and the original data is used to calculate value of the node. So, the model is suitable for using historical data to predict future customer risk in semi-contract context.