A comparison of two multivariate analysis methods for segmenting users of alternative payment means

Purpose: Alternative payment means have been expanding rapidly in recent years. The need to identify the segments of customers that are targetable for both financial and nonfinancial institutions is growing. The purpose of this paper is to use two different methods, discriminant analysis and decision trees, in order to compare the effectiveness of the two methods for segmentation and identify critical consumer characteristics which determine behavior and preference in relation to the use of payment means. Design/methodology/approach: Using data from 321 bank customers, decision tree and discriminant analysis methods are used, first to test the same set of variables differentiating the customers and then to compare the respective results and prediction ability of the two methods. Findings: Results show that discriminant analysis has a better model fit and segments the customers in a more effective way than the decision tree method. In addition, each method shows different variables to differentiate the customer groups. Research limitations/implications: The findings are limited to the sector and country of the study, as well as the convenience sample that has been used. Practical implications: Suggestions for financial managers to better understand their customers’ behavior and target the right group are discussed. Originality/value: This is the first attempt to compare decision trees and discriminant analysis as alternative segmentation methods for payment means.