Modeling might not be able to predict results

November 12, 2013       The NonProfit Times      

Predictive models, which predict future behaviors based on past action, have proven to be very helpful in fundraising. They can isolate predictive response variables and be applied to large datasets, reduce interpretation subjectivity and identify opportunities for optimizing a universe.

During the 2013 New York Nonprofit Conference sponsored by the Direct Marketing Association Nonprofit Federation, Joseph Coakley of the Food Bank for New York City, Britt Fouks of Target Analytics, and Jeff Johnson and Michele Peys of LW Robbins answered several questions about predictive models and their usefulness.

Some models are sexy, but are they appropriate for our organization? Let the numbers be a guide. Each modeling agency will be able to show which of its models is the best. Test the strongest.

What qualities and characteristics should we look for in a modeling partner? The first time, go with an established group/s. An organization that has 100,000 or more 25+-month lapsed names should have the lapsed names modeled immediately.

Are certain models out of your league if you’re a smaller organization with a smaller file? Lapsed models are tough to develop without a large number of names. Co-op models might require minimums, so ask for details. Reach out to see what a modeling agency can do.

In acquisition, what is the best mix of vertical rental and exchange lists vs. modeled lists? This is a question best asked by results. The list broker/agency will make a recommendation. There is no silver bullet.

NonProfit  Times
The Leading Business Publication For Nonprofit Management