Three Reasons Not to Build a Model
While it’s true that plenty of direct marketers are employing models to improve such factors as response rate, profit per acquisition and customer retention, this data tool does not fit all business situations. And, building a successful model takes more than simply the go-ahead from the corner office.
Maria Marsala Herlihy, senior vice president of strategic consulting and analytics at KnowledgeBase Marketing, a Richardson, Texas-based database marketing solutions provider, defined three common scenarios in which a marketer should not invest in a model during her session, When Good Models Go Bad, at the DM Days NY Conference & Expo held earlier this year.
Change—“The best models are built on stable historical data for businesses that are not planning on changing these dimensions in the near future,” says Herlihy. So any significant alterations to your product mix, marketing approach, messaging, pricing or branding strategies means you need to hold off on developing or applying any models until your business picture stabilizes.
Insufficient data—While this seems like a no-brainer, sometimes marketers are not aware of how significant the gaps in their data can be. Unless you have 1,000 of whatever information you need to build the model you desire, she advises, you are better off investing time or money to fill in the missing information than building what is likely to be a faulty model.
Walk Before You Run—Before you invest considerable time and money in a custom model, make sure you’ve tested the more conventional targeting methods. “RFM, univariate and product-based selections are industry standards because they work as a first step, and probably should be your first step as well,” notes Herlihy.
Herlihy can be reached at (972) 664-3600.