While most magazine promotions are spread out over time, it is possible for a subscriber to get two promotions within one week. However, that rarely occurs, he affirms.
Profiling for Profitability
As Meredith Corp. develops subscriber profiles, each magazine group works with the data analysis team to create models based on this insight and past profitability data. The models allow the groups to prospect efficiently from the corporate housefile. Each group tests its models, analyzes which of them worked well and uses that data to determine which offers or combination of offers to send.
Showalter says Meredith Corp. uses a combination of decision trees, linear and logistic regression, and other data-mining techniques to segment customers on the database. “Historically for magazines, we have optimized the models based on net response, but we are now building models to maximize profitability. With profitability models, we take into account not only response and payment information, but also whether the customer responded to an upsell or cross-sell offer at the time of payment, as well as cost factors such as how many efforts it took to collect payment,” he explains.
According to Ball, the company often begins with best-customer models or decision-tree models after it has a mailed a sample to use for regression analysis. He says, “Ultimately, we prefer to complete regression analyses for any programs that have sufficient universes.”
Analysts then apply lessons learned from one publication group to another. For example, elements of a promotion that worked for one cookbook or decorating magazine will probably work well for a similar product in another group. Ball asserts, “The data services group is a corporate asset available to everyone. Learning from one profit center is shared with all of the analysts so we can leverage learning as quickly as possible.”
Contact Strategy and Cross-promotions: A Departmental Decision