Mining data to discover customer segments

Abstract

Increasingly companies realise that consumers differ in their needs, preferences, sensitivities, opinions and behaviours. The transition from mass marketing to target marketing that is currently in progress creates demands for more sophisticated customer segmentation techniques. The key enabler of any segmentation strategy is customer data. Customer data are the raw material that must be captured, integrated and effectively analysed in order to achieve the goal of profiling customers. Effective segmentation requires both static customer profile data (eg demographic, lifestyle, preferences) and customer behaviour data (eg usage, loyalty, profitability). Once these data are integrated they can be interrogated to discover groups of customers sharing the same characteristics and needs. Market segmentation is the process of partitioning the heterogeneous market into separate and distinct homogeneous segments. A segment consists of a group of consumers who react in a similar way to a given set of marketing stimuli. Traditionally, the enterprise would define a segmentation matrix and then, based on the data, would allocate customers to segments. This a priori approach to segmentation defines, in advance, a framework or system that describes characteristics of customers or prospects based on information that is known about those individuals. What data mining makes possible is a different approach to segmentation— namely cluster segmentation. The cluster segmentation approach, in direct contrast to the a priori method, seeks to discover naturally occurring clusters of customers who share common characteristics or behave in the same way.