An Algebraic Approach to Quality Metrics for Customer Recognition Systems

Success in implementing a Customer Relationship Management (CRM) system requires close attention to data quality issues. However, most of the literature focuses on the quality of the input streams rather than the quality of the customer data integration (CDI) and customer recognition outcomes. This paper describes some preliminary research into the creation and validation of quality metrics for customer data integration and customer recognition systems. The approach is based on an algebraic view of the system as producing a partition of the set of customer transactions it processes. Comparing one system to another, or even changes to the same system, becomes a matter of quantifying the similarity between the partitions they produce. The authors discuss three methods for measuring the similarity between partitions, suggest the use of these measurements in creating metrics for customer recognition accuracy and consistency, and report on early experimental results.