The next frontier of predictive modeling is not about predicting a possible outcome for each constituent (such as identifying constituents who are most likely to lapse, renew, make a planned gift, etc.). The next frontier of predictive modeling is known as persuasion modeling.
The objective is to predict the type of contact method your organization should apply to each constituent in the data base to maximize that constituent’s chance of a successful outcome (such as renewing, or not lapsing). I just happened to find a link to the keynote address video from the 2013 Predictive Analytics World conference where the speaker, Eric Siegel, described the method. It is a truly informative presentation and completely easy to understand.
Here’s an article from HP’s Chris Surdak on why persuasion modeling is the next big thing. He writes it from a for-profit point of view, but the same principles apply to our sector too. We not only want to know which communication channel will likely have the greatest positive effect on each constituent, but we also want to know whether we should be using that communication channel at all! Reaching out could possibly have an unintended negative effect, so we only want to employ more expensive levels of communication treatment (like call centers) to those constituents likely to be persuaded to give again.
Definitely take a look at the video, it is illuminating and will make you feel like you’ve got the inside scoop on analytics best practices.