Jennifer Schulties, Senior Marketing Manager at Martha Stewart Living and Keith Bergendorff of Publishers Clearing House Analytical Services talked about predictive modeling solutions in a single model in a DMA 2011 session I attended today.
They call it lifecycle modeling across channels for acquisition and retention.
My take? They are in a difficult place.
Schulties laid out the problems: poor payment on sweeps, deteriorating bottom tier response on direct marketing, shifting renewal behavior, unfavorable insert response. Ouch.
She tried a daily giveaway and got more traffic and subscriptions, but at lower value. So, she created a model around what products customers paid for and how often they paid and learned that 45% of sweeps players paid 2x more than bad payers.
It’s always helpful to learn who not to market to.
Then she tried a 32-page bookalog, with a very compelling offer: free trial with a premium. Of course, with a free trial, payment is delayed. Could a “megamodel” help determine what offers will make low responders respond? The answer is “yes, but we still have problems.” They can get a 25% lift over the control, but the premium is more trouble than it’s worth. So they may cut marketing to this tier altogether.
It’s also helpful to learn what offers not to make.
What about testing those reply cards that flutter out of magazines you buy at the newsstand? Unfortunately, it takes a year to find out what worked, because it all really depends on annual renewals.
Is there any good news?
Both Schulties and Bergendorff spend substantially more on acquisition than renewals. Schulties targets primarily magazine subscriber lists, but what with all the magazine closures, the list universe is diminishing. So now she must find new list sources and score them in advance to help improve results.
Bergendorff evaluates list sources by customer lifetime value – so he knows a list is a success after 24 month period. Yikes!!!
My perception: these publishers have to wait too long for results and haven’t really mined the data the way they could.