Southern Progress’ Steve Crowe on Leveraging Co-op Data
In the quest to mail more efficiently, building a best customer profile is one way to achieve your company’s ROI goals. And using purchase history from a data cooperative to develop this profile can more precisely target prospects and boost response. Here, Steve Crowe, vice president of consumer marketing for Birmingham, Ala.-based Southern Progress Corp., explains how his company leverages third-party data. Southern Progress publishes magazines including Southern Living, Health and Sunset magazines, as well as books through its Oxmoor House division.
Target Marketing: How do you use third-party data to build customer profiles?
Steve Crowe: We purchase publicly available data. Working with Alliant [Cooperative Data Solutions], we leverage their sources of aggregated consumer transactional data to create custom models that predict the likelihood that a consumer will respond to and pay for an offer to buy from us.
TM: How do you determine the attributes of the Southern Progress best customer, then use this insight to guide list rental decisions?
SC: At its very basic level, the best customer is someone who has recently purchased a similar product. However, at the volume of products we are selling, we have to go deeper than that, and that’s where modeling can be helpful. We use our model solutions to suppress the worst performers from our mailings and also to identify more of the highest net responders in marginal list sources. We also use these models to identify the best names for reuse.
TM: What results have you seen since applying this database marketing strategy?
SC: To help us with additional modeling resources as well as expertise, we use Alliant as a marketing partner in addition to our in-house modeling group. Alliant uses our purchase history, combined with the purchase history of their other clients involved in the cooperative and their own data, to build models specific to each of our magazine properties. All the outside purchase history is blind to us, but the models perform very well. We are able to eliminate under-performing segments from our worst and best lists using these models. The fact that we can find good-performing names in our bad lists and under-performing names in our good lists tells me we are on the right track with our model building.