The May 2007 postal rate increase sent every direct retailer scrambling. It’s hard to argue the hike’s effectiveness as a catalyst for renewed analytical vigor.
Our clients have been analyzing everything from the impact of page count reductions and co-mailing programs to the most appropriate tools to optimize circulation. And for one, preliminary research indicated that a new zip model might be the right solution at the right time.
Zip modeling is not new. It remains a data-based tool that requires in-the-mail validation, but the postal rate increase was as good a time as any for many retailers to test it.
In one such case, our prospecting circulation was dominated by the cooperative databases (as it often is) and we extracted interesting, if not unfamiliar data from our testing.
First, we used a zip model to identify a list of “super” zips. These zip codes were the most highly penetrated by existing customers, and fit a demographic profile consistent with our best buyers. Then we analyzed our sources of prospecting names to see how effective they were at finding names in those zip codes.
What we found is that even the weakest co-op model names had more prospects in our best zip codes than any individual list that was available for rental. While that augured well for cooperative database performance compared to those outside lists, we also noted that an overwhelming majority of names even in the best cooperative database segment were not in our top zip string (62%).
We did not change our front-end co-op database selection strategy based on these findings. Instead, when we put the catalogs in the mail, we tracked the co-op prospects two ways: Super zips vs. non-super zips (geography) and uniques vs. multis in the merge (direct-to-consumer propensity).
What we found in this case, as we often do, is that a prospect’s direct shopping habits are more meaningful than where they live. The first chart below shows that among the cooperative database names we mailed, those in the super zip range behaved no differently than those not located in those zips.
On the other hand, if we received a name from multiple sources and flagged that name in the merge, we created an extremely meaningful segmentation variable.
We did not unveil the results of those low-penetration marginal rental lists, but we can report that they were extremely weak. Demography drove the selection of a few test lists, and even in our best zip codes those lists delivered unacceptable results.
The notion that customer behavior can be more predictive than customer location is not new to us, but we were impressed by the clarity of this example. There are several conclusions we draw from these results in addition to those outlined above:
- To a degree, geography is a self-fulfilling prophecy when statistical modeling based on buying behavior drives name selection.
- The model does not need to mirror your customer file exactly in terms of its geographic profile (indeed, if you think about it, you don’t want it to).
- You must validate tactical circulation tools such as zip models in the mail by testing it against your current selection strategy.
That last point is critical. The beauty of database marketing is the ability to measure. To not test is to miss an opportunity to become more informed. More importantly, the experience of one company won’t necessarily be shared by the next.
Every customer file is unique; therefore every zip model is unique to that file. It logically follows that the effectiveness of a zip model should be unique as well. Only testing will quantify the impact of a given circulation tactic.
Jude Hoffner is a director at San Rafael, CA-based catalog consultancy Lenser.