An upcoming presentation I’m giving for advancement professionals working in Annual Fund, Planned Giving and other fields has me asking myself: What is the pure essence of the goal of predictive modeling? Those of us who actually create the models might not stop to think about it very often. For this audience, though, I need to step back and state a problem – for which predictive modeling is the natural solution.
The best answer I’ve come up with is that the problem is The Unknown, and that predictive modeling gives us insight into the unknown.
Sounds mysterious, but in fact “the unknown” is something fundraisers rub up against all the time. We have tons of data, we’re drowning in it, but the raw data fails to shine a light into some dimly-lit corners of our alumni databases. Below are four concrete examples.
1. The majority of your alumni have never given to the annual fund before. Which non-donors are most likely to be converted into donors? And which lapsed donors are most likely to return to the fold? The Unknown: no or little past giving history to guide you, and too-large pools of ‘lybunts’, ‘sybunts’ and undifferentiated non-donors.
2. Among your current donors, who is most likely to be moved to a higher level of giving? The Unknown: Looking at giving history alone, one $50/year donor looks just the same as another – but one is never going to budge, while the other is ready to be asked to give $500, or $5,000.
3. Which alumni are most likely to be responsive to an invitation to discuss Planned Giving? The Unknown: Planned Giving donors fly under the radar. Too many alumni share the same characteristics you’ve been told to look for, i.e. many small gifts over an extended period of time. Alumni are approached too late in life.
4. You have a big alumni event coming up. Email is cheap, but many alumni respond to a mailed invitation. You can’t afford to mail an event invitation to everyone, so which alumni ought to get a mail piece? The Unknown: If they’ve never attended anything before, you have nothing concrete to go on.
These are just four examples of ‘the unknown’; I’m sure you can identify more. You may not be the one doing the actual modeling, but if unknowns like these are factors in your program, you need to know that predictive modeling is ideally suited to shedding some light on them. When we leave these dark corners unlit, we waste time and money chasing people who won’t give, won’t move up, won’t start the conversation about planned giving, won’t come to our event – and we pay too little attention to the ones who will.
Predictive modeling does not give us definite answers; it gives us probable answers. Given a choice between a probable answer and no answer at all, most of us choose the former. In fact, we do so every day!