After a cursory search of Google and Amazon, it appears there has been no publication of “Chicken Soup for the Data Analyst Soul.” I shouldn’t be surprised. Dashboards, tables, and charts do not typically evoke the same feelings as curling up by the fire with a favorite book. And yet, there are times when you can find a bit of Charles Dickens or Harper Lee in a spreadsheet. This is a tale of one of those occurrences.
First some background: We have a predictive model for donor value at TrueSense called GPS (Giving Potential Scores) that is remarkably powerful. It utilizes artificial intelligence to calculate every donor’s unique value potential to the cent. As opposed to RFM, which looks at just three values, GPS can look at hundreds of data points for every single donor. And unlike RFM and other segmentation models that categorize donors based on what they have done in the past, GPS organizes donors based on what they are going to do next.
Many of my coworkers have heard me say, “Don’t try to outthink the model — you’ll lose.” I have a confession: I try sometimes, too. But, mostly as quality control after a fresh build. My partner-in-prediction, TSM Data Scientist Allison Johnson, and I were doing just that recently as we dug into major gift projections for a number of clients. That is where the “Chicken Soup” part of the story starts.
Most organizations start with giving history (largest gift, annual value) to prioritize the mid and major donors on their file. It is usually a good place to start. I would expect to see major gift prospects with a gift of $1,000 or more making it into Level 1 on such a file. But this time, we were surprised to see this was not always the case. We discovered gifts as high as $2,000 were found among our Level 2 prospects (valuable, of course, but unlikely to ever hit $5,000 or more). We cracked our knuckles and started to consider data input problems.
But it turns out there was no problem at all. Some readers might already be seeing through the lines here. Remember that GPS predicts what donors will do next, and a $5,000 gift was certainly not in the cards for them.
Allison and I realized at the same moment what we were seeing. These lower level/big gift prospects all had the same traits: They’d been giving for a while, small gifts here and there for years. Their incomes were below average. Many of them had indicators of their own financial struggles. Most were older.
But these folks, when the chips were down last year, went all in on their neighbor.
Some of these large gifts you can imagine were stimulus payments passed on to organizations who served impacted causes. Others were likely opening their rainy-day fund for someone else that never had one.