The Founder of Predictive Donor Modeling
When I started working for UC Berkeley in 2001, UC Berkeley had just closed its Millennium Campaign and experienced a significant inter-campaign drop in donors and revenue. The UC Berkeley Foundation Trustees created a sub-committee focused on annual giving. One of the four recommendations that came out of this sub-committee was to mine our data and develop predictive models to help identify new annual giving donors. With this charge in hand, off I went, and I hired the only person at that time that was developing predictive models for the purposes of donor development, my now 20+ year friend, colleague, and mentor, Peter Wylie.
Peter Wylie was a pioneer in bringing predictive modeling to higher education development. I not only hired Peter to develop UC Berkeley’s first predictive model to acquire new annual fund donors, but I hired Peter to train me in how to mine my own data and develop my own predictive models. This was a fascinating journey, a process I enjoyed tremendously, and then I used my new found knowledge to develop other predictive models for UC Berkeley — such as a leadership giving model to identify new $1K+ donors — and to develop predictive models for other universities and schools, such as Tulane University, University of Melbourne, and the Haas School of Business to name just a few. Each model I have developed is unique to the purpose and institution it was developed to serve, and each of these models has been put into practice to help more strategically target precious resources to drive greater results. And for this, on behalf of myself and the entire fundraising profession, I (we) have Peter Wylie to thank!
Thank you Peter Wylie for sharing your expertise and passion for statistics, and enlightening us all on how predictive modeling can advance our fundraising programs and results.
Over the next few months I plan to post a number of blogs on data mining and modeling, but I wanted to first acknowledge Peter Wylie for his pioneering role in developing predictive models to advance development programs, and to thanks Peter for training me in the practice.
And when I think about what Peter has instilled upon me, I realize that there are a number of key things that Peter and I agree upon in regards to data mining and modeling:
1) Data Mining is Insightful — The process of mining your data is an insightful process in and of itself, so there is tremendous value is mining your data yourself versus paying others to do it for you. One small example is: do you know what percentage of your constituents are on Gmail vs. Yahoo vs. Comcast vs. still on Hotmail/AOL/etc.? A simple mining of your data will tell you this, and will thus tell your direct response marketers what filters they are mostly working up against in getting your email appeals into people’s inboxes.
2) No Black Boxes — Peter and I also agree that institutions should own their predictive donor models, and not be sold a “black box” that they do not know that internal and/or external factors are being considered into the predictive power of their model. By owning your own predictive model, you are assured that your internal data is being amply applied to make your model truly unique to your own business objectives and institutions. All institutions have their unique characteristics — urban vs. rural, public vs. private, large vs. small, professional vs. arts/humanities, or gender/racial mix. So don’t settle for a black box, own your predictive donor model so you know what it entails and you can refresh it at your own frequency for free!
3) Hire a Modeler — Finally, Peter and I also agree that institutions should train their own staff on the arts and science of predictive modeling. The basic statistics behind predictive modeling does not make it out of reach for someone who is reasonably analytical to master, at which point your institution now how a resident expert on developing predictive donor models to develop more targeted ones for specific business purposes, such as: new donor acquisition, leadership $1K+ donors, planned giving donors, and major $100K+ donors. This is why both Peter and I enjoy training others in the practice of mining your data to develop predictive donor models for your institutions.
Here’s a bit more about Peter Wylie, the founder of predictive modeling in higher education development…
Profile: Co-Founder of Affordable Advancement Analytics
https://a3giving.org/peter-wylie/
Book: Data Mining for Fundraiser
https://www.amazon.com/Data-Mining-Raisers-Peter-Wylie/dp/0899643809
Blog: Cool Data
https://cooldata.wordpress.com/category/peter-wylie/
Data Desk
https://datadescription.com/peter-wylie/
Academia.edu White-paper: Making a Case for Modeling
https://www.academia.edu/41551923/MAKING_A_CASE_FOR_MODELING
