10 Ways to Get the Most from Co-op Databases

Feb 22, 2005 8:18 PM  By

Cooperative databases such as Abacus, I-Behavior, NextAction, Z-24, and Prefer account for more than half of the prospecting done by small and midsize catalogers, according to Stephen R. Lett, a Bethany Beach, DE-based direct marketing consultant. Why? Because these prospect names perform well and represent a good value for the money.

That said, catalogers can get even more value for their money, Lett adds. Among his suggestions:

1) Use more than one co-op. There will be overlap of names, of course, but each co-op will identify prospect names unique to its modeling methodology. Using only one co-op can limit your prospecting universe.

2) Use co-ops for outside list optimization. This entails matching outside names against names from a co-op and assigning a score to them. Especially useful when you’re using compiled or subscriber names, this process can suppress less productive names and subsequently improve the response rate from a particular list. This technique can also be used to optimize the performance of marginal lists by suppressing segments that are performing below acceptable response. Mailers using this technique will typically suppress 10%-15% of a list, thus providing enough of a lift to make a marginal list worth mailing.

3) Try rental singles suppression. “Singles,” or one-time buyers, generally perform at less than the response rate for the average rental overall. Therefore, one of the more common uses of list suppression is optimizing the rental singles coming out of a merge. Better prospect names can be used instead, resulting in an overall increase in the performance of outside prospect lists.

4) Use co-ops to reactivate customers and inquirers. Most catalogers can mail back at least to a portion of their 36-month-old buyers with acceptable results. You can use co-ops to help reactivate the older buyers and inquiries you don’t plan to mail. Mail all the names you can and ship the balance to a co-op. It will run a house file reactivation model against these older names and identify a select group worth mailing again. This has proven to be a most cost-effective way of converting older buyers and inquiries to a more recent RFM (recency/frequency/monetary value) cell.

5) Add a mail drop. It is difficult to overmail your house file (or at least your better customers). The same holds true for the better-performing names from one of the cooperative databases. If holiday is your best season, perhaps you can squeeze in a late mailing to the best segments of your house file including a select of the best modeled names from the co-ops. This additional drop should perform at 70% of a regular house file mailing to like RFM cells.

6) Enhance results with data overlays. Co-op database I-Behavior is able to enhance results by using external data as part of their modeling. It offers attitudinal segmentation through its partnership with Yankelovich and it MindBase segmentation system. I-Behavior is able to link behavioral targeting with the underlying attitudes and motivations driving purchase behavior. These enhanced data overlays can help expand the prospecting universe.

7) Leverage SKU-level data for modeling. I-Behavior and NextAction model at the SKU level. This can provide greater precision in categorizing buyers. SKU-level modeling can be used to build customer models that help match traditional purchase history such as RFM with a cataloger’s merchandising strategies.

8) Update your models as frequently as possible. Critical to the models are the data you supply to the co-operative databases. Be sure to update the co-ops monthly or more often if possible. The “freshness” of your data is key to the results achieved.

9) Work with your marketing representatives. Share results with them. This will help the co-op tweak its models in order to improve performance. Often a few test mailings are needed to find the best models for your offer. Don’t work in a vacuum, and don’t give up after one test of a model.

10) Test new models. All models have limited universes. Once you get beyond the first two or three segments results begin to decline. Therefore, it is important to test new models so that you can identify other prospect names to mail. As a general rule, when Lett selects names from a co-op he likes to mail to 70% proven models while testing 30% of the circulation to new models.