Customer retention in the financial industry: An application of survival analysis

Hong Zhang, Purdue University


Recently, in both marketing theory (academia) and practice (industry), the emphasis in relationship marketing has shifted to long term customer relationship management emphasizing customer retention or loyalty. This study has two main purposes: (1) to investigate the impacts of selected firm-customer interaction behavior and demographic characteristics on customer retention behavior in the financial industry, (2) to compare the results of a static method of analysis (logistic regression model) and a dynamic method (Cox’s hazard method) for customer retention. The statistical analyses conducted employing two cohorts showed that the Cox model was stable across cohorts. Using Cox’s hazard method, all of the independent variables were found to be significantly related to customer retention, while some of demographic factors (for example, tenure, marital status and gender, and channel usage) were not found to significantly affect customer retention in the logistic regression. The results show that with the enhancement of interaction between a bank and its customers by increasing customers' service usage, cross-buying, tenure experience, and complicated product usage, customers were more likely to stay longer with the bank. Additionally, age, education, and income were found to be positively related to customer retention, while customers with higher time pressure were less likely to defect. Single females were found to be least likely to attrite, while married customers were most likely to default or switch to an alternative. Single males were more likely to stay than married ones, but more likely to churn than single female customers. Finally, managerial implications were drawn, and limitations and potential for further study were discussed.




Widdows, Purdue University.

Subject Area


Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server