Client: A retailer with catalog and store operations

Project:  Implement an a priori customer segmentation scheme designed by marketing department

Challenge:  Keep development costs reasonable while grappling with very complex requirements.

Achievement: The marketing department defined a set of customer segments.  Raw transaction data was available, so the segmentation schema took place in a new customer data mart.  The segments were an attribute of the customer dimension, but were derived from 3 other dimensions of the transactions: Recency, Brands and Channel.   While it was intuitive for marketers to designate a particular segment as someone who has transacted with one brand but has with another, capturing the absence of information added more complexity to the derived segments.

The solution consisted of becoming intimate with the requirements and refining them.  The data mart ultimately contained a history table of every segment to which any customer had belonged to over the past 2 decades.

By clearing understanding the problem and capturing the requirements in the application logic vs. the data model, a maximum amount of information could be derived in one pass through the data.  This meant the project was completed in 1/3 the time and 1/4 of the cost than was proposed by a large vendor of typical star-schema/BI applications.