The Customer Churn Analysis Of Automobile Sales Service Shop

With the rise and sustainable development of China ’s economy after the reform and opening up,civilian vehicles have entered the vision of ordinary families.With the increase of family car ownership in China,after-sales service has ushered in new opportunities.Automobile after-sales service includes a series of contents such as quality assurance,claim,maintenance,maintenance service,automobile spare parts supply,maintenance technical training,technical consultation and guidance,market information feedback and so on.Under the condition that the profit space of the whole vehicle is squeezed,the automobile after-sales service will become a new profit growth point in the automobile industry chain.Along with the opportunities,there are also challenges.On the one hand,in order to promote new cars,the automobile sales field often gives very large discounts,and even makes " loss-making sales." Automobile dealers are bound to try to recover profits from after-sales.On the other hand,4S stores have limited management capabilities and lack of targeted services.Two factors cause4 S stores to face serious customer churn problems.Therefore,it is particularly necessary to use more accurate means to analyze the after-sales data of 4S customers.With the development of statistics and the improvement of computer computing power,corresponding to the expansion of customer data,more and more research focuses on customer churn.However,in previous studies,more subjective methods are often used as the standard for customer churn.At the same time,the phenomenon that customers repeatedly come to the store and the same customer has multiple records is ignored.In order to avoid the subjectivity in the analysis and effectively use the data characteristics of customers ’ repeated visits to the store,this paper uses recurrent event analysis,a method derived from survival analysis,to model and analyze the after-sales data of automobile 4S store customers.The proportional hazard models of two styles are established respectively.The first is the fixed effect Andersen-Gill model.After that,the author added random effects to the model and expanded it into a frailty model.In the setting of frailty distribution,the author used three PVF distributions and lognormal distribution,and then compared multiple models.The modeling results show that among the 25 variables used for modeling,7 variables have the greatest impact,namely,whether the visit to the store is regular maintenance,the number of times the customer purchases insurance,the average payment price for maintenance,the average discount enjoyed by the customer,the average time interval of the customer ’s arrival time,and the longest time and average time consumed by maintenance.At the same time,it is best to introduce positive steady-state random effects into the model.The results of the analysis show that the four variables of purchase frequency,average payment price,average discount and maximum maintenance time are negatively correlated with the possibility of loss,while regular maintenance,average time interval and average maintenance time are positively correlated.Therefore,4S stores should maintain more regular maintenance,and customers with longer time intervals should maintain higher attention.At the same time,4S stores should moderately speed up maintenance,especially the pace of small maintenance,and give customers appropriate discounts to reduce the possibility of customer churn.