Identifying Customer Segments with Highest Potential value
The client already had comprehensive RFM analytics, and overall market penetration information. However, this was at the aggregate level and no attempt had been made to identify high potential segments based on current sales data. With a fixed margin and steady purchase frequency, the requirement was to define the largest segments with affinity to the product.
The customer records although large in number were not attribute rich, and several overlays of census and survey data were required to develop a fuller picture of the customers. To zero-in on the most active group, a combination of K-means and hierarchical clustering techniques were used to partition customer attributes based on non-consumptive metrics. The results were definitive: seniors, regardless of affluence or activity were the dominant purchasers. Also, Hispanic families were a significant group purchasing the products. Both segments were surprising: while seniors were not big spenders, they made up for it in numbers, and the latter was a known segment, but the large associated value of this segment was surprising to the client.