Put simply, modeling means finding like characteristics about people that, when combined, enable you to predict future behavior. There are several inexpensive ways to model your data yourself (and some more expensive ways to use external data sources to help you) that can yield more revenue at lower cost from your direct-response program.
Modeling is best used when you have a specific question that you’re trying to answer, such as: “Who are my best donors to approach for an upgrade?” “Which petition signers in my database have the best potential for contributing?”
If you do not have a specific question in mind, it still can be useful to see if there are characteristics influencing the outcome of a mailing that were not in your selection criteria. For example, if donors were selected for a mailing based on recency, frequency and gift size, you might want to look at the results by region, the acquisition package through which donors were acquired or by cumulative donor value over their lifetimes.
Before going into specific cases where modeling can benefit your program, it is important to remember that there are things modeling will not do:
- Modeling does not mean you can ignore direct-response best practices. For example, if you are modeling to eliminate donors from a mailing, test a sample of people you would have included had the model not eliminated them.
- Modeling does not make up for a weak case for support. When times are tough, your case for support must be stronger as you are competing with many groups for limited dollars.
- Modeling cannot capture the emotional element that leads to a contribution. The model can tell you who is more likely to give, but you still need to make a connection between that donor and your cause to motivate him to give.
These specific cases help illustrate some good uses of modeling. Each case begins with a simple question that an organization was trying to answer.