Best-Practices Marketing Database Content is defined by the Ten Commandments listed in the May 5, 2007 e-Letter. The focus this month is on The Tenth Commandment: Overlay data must be included, as appropriate.
We define overlay data as the demographics and "firmographics" available for purchase or lease from third parties. Examples of demographics are: age, estimated income, presence of children and marital status. Examples of "firmographics" are: industry sector and number of employees. This contrasts with behavioral data about your own customers, such as order, item and promotional detail.
Overlay data can be applied to the two types of database marketing analytics: 1) determining whom to promote, including when and how often; and 2) determining what to promote. Generally, overlay data plays a limited role in the first, but a very important role in the second.
The foundation for the first type of analytics is statistics-based predictive models such as up-sell and cross-sell. Typically, the output of these models is hierarchies of customers based on their expected near-term performance. By crossing model output with financial metrics, it is possible to determine the precise contact intensity that is appropriate for each customer.
Generally, this model output is heterogeneous. For example, the inhabitants of a model's Decile 1 might be a mix of young vs. old, and self-purchasers vs. gift-givers, with very different historical merchandise purchase characteristics. In fact, the only guaranteed common characteristic will be that - by definition - their expected future performance is stronger than Deciles 2 through 10.
Typically in the presence of robust customer behavioral data, overlay data provides no cost-effective incremental predictive power in determining whom to promote. In contrast, overlay data often plays a seminal role in the second type of database marketing analytics; that is, determining what to promote. Generally, such output is homogeneous; that is, groups of customers are defined that display some form of commonality. Next month's e-Letter will explore the importance of this homogeneity, and the role that overlay data plays in solving the what-to-promote problem.
