When I had the honor of leading direct marketing for Mothers Against Drunk Driving, I would train about the leaky bucket. You know the one. You have a bucket full of donors. The bucket has holes in it. If you want your water level to rise (i.e., get more revenue), you must fill the bucket faster than water is leaking out. Simple and easily understood.
The challenge we face is that our buckets are very leaky. The last annual Fundraising Effectiveness Project found that 80% of first-time donors will not become second-time donors. Thirty-nine percent of multi-year donors won’t give again this year. And the recapture rate of these donors has fallen almost 40% in four years.
At the current rate, our donor files will be halved in the next decade on average. Our buckets have too many holes.
You know the buckets. Lapsed versus active. Single versus multi. Digital versus mail (versus phone versus canvass versus… etc.). Godzilla versus Megalon. Matures versus Boomers versus Gen Xers (and occasionally Millennials).
Each bucket does a job, trying to pick the things that are most like each other and put them in the bucket and exclude the things that are most unlike the things in the bucket.
But they do that job so poorly. Let’s take the classic active versus lapsed dichotomy and say that your organization draws that line at two years. Do you believe in your heart of hearts that there is a substantial difference between someone who donated 23 months and 28 days ago and another person who donated 24 months and 2 days ago if every other thing about them is the same? And are two people who donated 24 months and 2 days ago the same if everything else about them is different?
This is one of the better, more predictive segments. And yet it fails as a segment because the people in the bucket want different things and messages from other people who are allegedly in the same bucket.
When someone last gave is a poor predictor of when someone will give again alone. As part of a larger predictive model, it is a great addition that helps make a model more predictive. So is all other transaction data. So are consumer data points. So is behavioral data. So is the person’s history with your organization.
Once in this model, you should be able to line up your donors from 1 to n in exact order of how likely they are to make money for you if you communicate with them.
Further, you can alter the communication to make them even more likely to give. If one person gives to you because they adopted their cat from one of your shelters and another gives because they are a dog breeder who can’t stand animal cruelty, the fact that these two are almost equally likely to give obscures the fact that they want to give to different things for different reasons. Thus, different appeals are called for.
This takes us out of the comfort of the bucket. We used to have a bucket. That bucket had a label on it. And we could test what happened with that labeled bucket.
In order to get to the promise of one-on-one marketing, we need to let go of bucket thinking. We need to embrace that we will no longer “know” everyone who is getting an appeal; that knowledge was illusory anyway. We need to embrace that we will no longer have full control over who gets what messaging. Our human-based segmentation John Henry is no match for the machine-learning-based steel-drivin’ machine.
Consumer marketers are doing this already to make us buy things. Only when we get rid of the bucket will we be able to grasp the relevance that makes our buckets less leaky and puts us on a par with those who want to sell things, not impacts.