Year after year, I see the results of the propensity model for phonathon and I become increasingly convinced that calling priority must be based solidly on the predictive score. Call your high scorers first, regardless of their giving history.
Does it not make sense to use giving history to determine strategy, and stop using it to assign call priority? What is so magical about LYBUNTs? Why are we determined to contact every one of last year’s donors, including the ones least likely to renew, when we should be moving on to broadening the base by targeting the high-scoring never-donors who are most ripe for conversion?
(When I say “we”, I’m not referring to my own institution. I’m talking about all phonathon people out there.)
The high scorers in my model have the highest rates of participation, give the largest gifts, are the most likely to convert or renew or return (if lapsed), and are the most likely to increase their pledges year over year. Those high scorers come from every donor category, from loyal donors to first-time donors, to people who’ve never given us a dime before.
The evidence tells me this: We make a huge mistake whenever we call low scorers ahead of high scorers, simply for the sake of squeezing dollars out of a calling group we have composed based on our assumptions.
Not entirely false assumption, of course. We are mostly right to assume that last year’s donors are more likely to give this year than any other group. But calling ALL last year’s donors first assumes that they are uniformly likely to renew, which they most certainly are not.
A constituent’s giving history determines your goals and strategy. For non-donors, the goal is conversion. For first-time donors, the goal is renewal. For loyal donors, the goal is upgrading. There is no reason why these goals can’t be persued simultaneously, or nearly so. In my opinion, it is never wrong to make it a priority to work a segment of never-donors, IF they have a high propensity score.
These days, when it takes multiple calls to get even a loyal donor on the line, call priority is very important. If your model is trained on the most phone-receptive people in your database, then use it.