It’s time to explore the two variables we created in Part 1. The first was ‘Years of Giving’, and the second was ‘Frequency of Giving’. Both of these things are generally assumed to be predictive of Planned Giving potential.
The key word is ‘assumed’. Based on assumption alone, you can go into your database right now and skim off the top alumni by years and frequency of giving, and call them your top Planned Giving prospects. That would fall into the category of data mining, and you might have some success doing this.
But why not kick it up a notch? If you can do data mining, you can do predictive modeling.
In modeling, characteristics such as years or frequency of giving are regarded as variables, just like ‘Class Year’, ‘Homecomings Attended’, ‘Business Phone Present’, and all the rest of them. And like any other variable, their relative power to predict Planned Giving potential is demonstrable.
Let’s explore ‘Years of Giving’ first.
For ease of visualizing this variable, I chopped it into ranges, as in the chart below. This chart shows how our Planned Giving expectancies (on the right) differ from all other alumni (on the left), with regards to the number of years they’ve made any gift.
Look at the blue parts (no giving) for both stacks: Our Planned Giving expectancies are far less likely than other alumni to be non-donors. That should not be a surprise.
Look at the purple parts (15 to 21 years of giving): Our expectancies are much more likely than all other alumni to give every year. Again, that’s perfectly in line with conventional wisdom. So far so good.
Our next chart shows the same side-by-side comparison for ‘Frequency’ of giving. This variable is quite closely related to ‘Years of Giving’, and we see the same dramatic differences. Our existing expectancies tend to be more frequent donors than the general alumni population.
There you have it, two solid characteristics associated with alumni who choose to enter Planned Giving commitments. These characteristics, and similar ones that might result from a standard RFM analysis (Recency, Frequency, Monetary value), might be enough to satisfy some.
But let me show you something else.
Here’s another side-by-side comparison. Now we’re looking at Homecoming Attendance. Have a look at this.
This is based on attendance data going back more than ten years. I am so glad we have that data, because as it turns out, Homecoming attendance is the second most powerful predictor of Planned Giving potential for our institution – after ‘Lifetime Giving’, but before any other variable related to giving history. Even more so: MULTIPLE Homecoming attendance!
I concede there is a significant age difference between these two groups – I did not take the extra step of limiting the population to older alumni when I made these charts. But the observation of difference between the groups still holds valid. (Only 16.4% of our living alumni from the Class of 1979 and earlier have ever attended Homecoming, but almost 56% of expectancies have.)
Prospecting for new Planned Giving commitments is hard enough. We make our jobs that much harder when we fail to add up the combined power of predictors such as event attendance and a dozen other things sitting in our databases.
If you remain unconvinced, if you still think that past giving behaviour is the only true predictor of future potential, then let me leave you with a final observation from my analysis of our own data: If all of our current, known Planned Giving expectancies were hidden in the database like needles in a haystack, and we were only allowed to use past giving patterns to find them again, we would miss two-thirds of them!
In Part 3, I will finally reveal my top 15 predictors of Planned Giving potential. I promise.