Using contextual information in transactional segmentation: an empirical study in e-commerce

The growing complexity and variability characterizing markets have induced scholars and marketers to propose new segmentation approaches. Recent research has shown that including the context in which a transaction occurs in customer behavior models, improves the ability of predicting their behavior. However, no systematic research has studied whether contextual information really matters in market segmentation. To this aim we conducted an empirical study in an e-commerce application across a wide range of experimental conditions. The results show that context strongly affects the composition of segments. Moreover, including context in the segmentation approach can improve both the homogeneity of segments and the ability of predicting customer behavior. Finally in some experimental conditions, the finer contextual information is, the better segmentation results are. Some managerial implications related to the benefits and complexity of a contextual segmentation are discussed.