Sizing Up ZIP, Household Data
BECAUSE CIRCULATION DIRECTORS can’t live by list rentals alone, Analytical Technology, the data services division of list firm Paradysz Matera, has released its 2001 modeling case study. The report offers solid arguments for both ZIP-code-level data modeling and household-level segmentation.
Minneapolis-based Analytical Technology built its study from mailings done by 66 of its clients, including publishers, nonprofits and catalogers. Thirty used ZIP code data models, 30 used household information models, and six employed both.
The results of applying ZIP modeling varied widely depending on how deeply marketers had to mail into their available universe. Those that were able to satisfy their mailing quantity needs from the top decile saw an average 34 percent rise in response rate. But even those that had to mail to 80 percent of their universe due to limitations in the number of names available experienced an 8 percent lift.
Marketers that went with household-level modeling saw even greater increases: Those targeting the top decile achieved a 47 percent jump in response. Meanwhile, offers made to eight out of every 10 names modeled realized a 13 percent gain.
ZIP code models aggregate whatever a marketer wants to use as its success criteria, such as response rates, at the ZIP code level. Marketers can then make future list rental decisions based on high-yield ZIP codes, which can be submitted to list managers as if they were any other select.
For household models, Analytical takes a sample of the names from recent mailings and overlays more than 250 attributes from compilers.
By identifying common attributes of both those who responded and those who didn’t, Analytical isolates which overlaid data indicates a household’s propensity to respond. Each relevant bit of data is then assigned a weight, or level of importance. The resulting formula of attributes and weights is unique to each marketer.
While running the household model may be expensive, it’s often more stable. ZIP codes can change as frequently as every two years, but according to Analytical’s president Jeff Clement, the household model is more likely to be stable if the offer or target market remains the same.