The Importance of New List Placement in the Merge/Purge Process

The Importance of New List Placement in the Merge/Purge Process
By Perry D. Drake

The following article, written by Perry D Drake, appeared in Target Marketing TM Tipline, August 11, 2004. It provides direct marketers with some guidelines regarding common list testing mistakes.



When prospecting for new customers, many marketers make a very important mistake in setting the merge/purge rules, which ultimately can result in a misread of the test and a wrong decision being made.

When accessing a new list in terms of response versus other lists, if the new list is cheaper than the others being promoted, it should have the highest priority in the merge/purge process.

Why? Because if this cheaper list is successful, you will want to promote the names on this less expensive list and not from a more costly list, should the names appear on both.

To determine the priority in the merge/purge process, consider how the list ranks regarding costs (projected rollout cost per thousand, not your testing cost per thousand) versus the lists you currently promote. For example, if the new test list ranks the second lowest in terms of costs, then it should have the second highest priority just behind the cheapest list. This will allow you to assess its true value to the bottom line in a rollout situation. And, likewise, if the new list is the second most expensive list, when testing, make it the second to lowest priority in your merge/purge process.

The same holds true if you are considering the build of a best customer or response model on a new prospecting list. You must determine the modeled lists’ rollout cost and see how it ranks relative to the other lists in your prospecting pool to determine its priority in the merge/purge process. However, keep in mind that it may be hard to properly reconcile the model unless you give it top priority to ensure you are reconciling the model on the exact same types of names the model was built on. This is a very important issue to keep in mind, but not a trivial one to implement by any means.

If a modeled list will not enter the merge/purge in top priority, or near top priority, another option is to consider modeling the names that survive the merge/purge. Doing so will increase the likelihood that the model will be successful and maintain its integrity. Until very recently this wouldn’t have been practical. But new data mining software, which can create and score standard models in hours, now makes this option attractive. When post-merge/purge models are used, the consistency between the universe on which the model was built and the universe on which it was implemented increase the chance of both in-market success and accurate evaluation.

So, remember to pay special attention to the priority of your list tests in the merge/purge process and if testing a model on those new list tests, ensure you will be able to properly reconcile the model. Not doing so will, without a doubt, yield misleading results and you may overlook a potentially profitable prospecting list for consideration.