Identifying And Analysing Consumption Abnormal Decline Customers Based On Stochastic Model

Under the increasingly fierce competitive economic environment, the most basisprofit for enterprises is from their customers. While customers today are given moreand more consumption choices, enterprises normally have plenty of customers.However, different customers make varying degrees of consumption contribution tothe same enterprise. In other words, for customers, different purchase amountrepresents different contribution rates for enterprises. Generally, the rates of oldcustomers are larger than that of new ones. According to the theory of businessmanagement, enterprises ought to focus on maintaining the existing customers whilein reality, they tend to pay more attention to explore the news, and neglect the loss ofthe olds. As a result, profit of enterprises is continuously decreasing, although theyhave taken enormous actions. With more and more failures alike in this area,enterprises begin to shift their attention to the loss of their existing customers. Forcustomers who are leaving enterprises, they can be identified by abnormally decliningconsumption. Customers in the process of leaving are different from those already leftin that the former ones are still consuming but actually intend to be lost. Nevertheless,few enterprises are capable of concerning and identifying these customers, nomention to take control of this pragmatically and effectively. It is obvious thatenterprises need to rely on their resources to identify those customers to prevent themfrom further decreasing consumption. Therefore, how to utilize database which storescustomers’consumption histories to identify leaving customers is of greatsignificance. To remain invincible in the competitive marketing environment, enterprises must know the future consumption trend in the first place, so as to predictcustomers’future consumption amount; by doing this, enterprises make commercialand marketing strategies. In this era, information is everywhere. For most enterprises,they collect customers’consumption histories by member card or online feedback,and thus form their own database. The database is filled with lots of historicalconsumption records, such as consumption frequency, consumption type, andconsumption amount. It is this database that lays a solid foundation for enterprises tomanage customers. To predict customers’future consumption behavior is verycomplicated. So far, most papers related adopt stochastic models; first to predictcustomers’consumption frequency for the time to come with NBD model, and then topredict customers’average consumption amount with Gamma-Gamma model; theproduct of these two terms equals predicts customers’consumption amount. Theabove method plays an important role for enterprises in customers’value evaluationand customer management.This paper shows that, in a given period of time, consumption amount obeysGamma distribution. Assume that parameter p is fixed in different time periods, and qobeys another Gamma distribution, a simpler stochastic model to predict futureconsumption amount will be concluded, among which, p is a ratio parameter and q isa shape parameter. The sum of future consumption amount for all customers canreflect future economic profit for enterprises. In particular, while the model is used topredict a singular customer’s future consumption amount, it can help identifycustomers with abnormal declining consumption and lay a solid foundation in thefollowing customers’differentiation management.In this essay, a new Gamma-Gamma model is developed to predict future profit.When the stochastic model is used to predict individual consumption, an error isinevitable. However, considering the model’s effect in predicting the overall futureconsumption, which can reflect economic profit for enterprises, the error is relatively smaller. What’s more, compared with other models, the new model put up in thispaper is easier. This essay analyses a certain enterprise’s consumption data with atime span of two year and a half with empirical analysis method, and it turns out thatthe accuracy in predicting economic profit is extremely high. On the basis ofpredicting individual consumption amount, a method in identifying customers withabnormal declining consumption is also developed. As what is said above, this sort ofidentification is the basis of effective management of individual customer. In the end,the evolvement of consumption decrease is deeply discussed, and the phenomenon ofthis is amply analyzed. By elaborating the ostensible and potential reasons ofcustomers’leaving, targeted marketing advices are provided for enterprises.