The following is an excerpt from the Direct Marketing Association’s 2005 Customer Prospecting and Retention Report. To learn more about the report, visit www.the-dma.org.
The idea behind both household models and zip code models is that demographics matter and that there are almost certainly prospects on vertical lists that “work” for you but would work even better if prospects with low probabilities, based on their demographics, were excluded from your promotions.
According to a DMA survey, 83% of respondents use vertical or rented lists for prospecting. About 60% of them use either zip code models or household models to select or suppress names from the lists. Usage of models goes up as the size of the firm increases.
The common-sense expectation is that models based on individual-level household demographics should perform better than models based on zip code averages, and in general this is true in terms of lift. But when the costs and complexities of scoring are taken into account the zip code model can sometimes turn out to be the better choice. For that reason, David Shepard, president of Dix Hills, NY-based David Shepard Associates, who developed and analyzed the results for the DMA, suggests performing a gains analysis when it comes to using small-area models, such as five-digit, nine-digit, or block groups, vs. household models.
That said, the decision to use small-area models or household models should not preclude the use of commercial segmentation schemes as part of the mailing process.