By Jason Shults, University of Arizona Foundation
A friend once told me that you can get to the heart of a person’s true nature by asking them “why” about any opinion they hold, and then asking “why” again to the subsequent answers they give. Somewhere at the bottom of the “whys” lies the person’s true beliefs and motivations.
I discovered later that this wasn’t an original idea, or at least not a unique one. Sakichi Toyoda, founder of the company we now know as the Toyota Group, had developed “The Five Whys Technique” as a diagnostic tool nearly a century earlier. If you ask the question often enough, the thinking went, you can get to the root cause of an issue.
The 5 Whys Technique in Action
The same technique can be applied to donor data. Often, when we think about data analysis, we can become overwhelmed by the sheer amount of information available to us. What fields should we consider? What kind of analysis should we do? How do we know where to begin?
Our prospect development team was given this challenge last year when the University of Arizona began to explore a fundraising initiative based around scholarships, which will likely be a cornerstone of our next comprehensive campaign. We had plenty of data in our CRM — some good and some not-so-good — and we knew which donors had contributed to our target funds. But we didn’t have the IT or business intelligence resources at that time to perform an extensive data extraction. We also didn’t have the analytics expertise to perform a comprehensive modeling project.
We decided to start small, with a simple database query to find all the donors to scholarship funds over the previous five years. Using that query, we exported two fields of data at a time. The first field was a constituent record identifier, so we could consult an individual donor’s record when necessary. The second field was an exploratory field: a piece of information we were interested in.
We performed this technique on any field that held demographic data, education data, campus involvement, relationships, employment or any recorded contact with the University. We
fed the output fields — the donor identifier and the single exploratory field — into IBM’s SPSS Statistics software, although this same simple analysis could be handled easily in R, Python or Microsoft Excel. We tallied up the data in the exploratory fields to find the frequency of each result. And then we began to ask “why.”
Sample Frequency Analysis Results
When looking at alumni giving to our target scholarship fund, the relatively higher giving rate of Gamma Omicron fraternity members led us to ask “Why,” and to perform further research on the organization.
- Why, for example, were so many of our alumni donors to scholarships members of a certain fraternity? This led the prospect research team to look more closely at this fraternity and their traditions of giving back.
- Why were so many graduates of our business school giving disproportionately to these scholarship funds? Did they give at a higher rate to scholarship funds than they gave to all funds in general? If so, why?
- Why was the ratio of donors to these funds so heavily weighted to the male gender? Was this ratio the same as the ratio of donors to all funds? (Hint: It was similar, because of historical household gift-recording practices.)
The purpose of this kind of exploratory data analysis is not necessarily to answer questions, but to find avenues for further research and to guide future analytics projects. For instance, the anomalous, higher giving rate of members of one fraternity led us to do three things: (a) perform additional research on the fraternity to find reasons for the higher giving rate, (b) add this information to our ever-growing “Research Bible,” where our facts and findings are kept, and (c) begin to think about additional factors to explore, such as giving rates by members of other clubs and organizations.
As we continue to prepare for the upcoming campaign, these small discoveries have helped to guide and inform our preliminary prospecting efforts, and are mentioned, when relevant, in our referral notes to gift officers. Finally, it should be pointed out that we rarely need to use all of the suggested “Five Whys.” One or two is usually enough to hint at a fruitful way forward.
A few guidelines to help you get started with exploratory frequency analysis:
- Be generous when looking at potential exploratory fields. You never know where some enlightening piece of information might be hiding.
- Define a time period to explore — perhaps the last three, five or ten years.
- When relevant, be sure to look at the initial results for your target funds against the backdrop of donors to all funds.
- Look at the results graphically, with plots or bar charts. This helps to reveal differences and anomalies that a table of numbers might disguise.
- Consider the context of the data. Has it been entered consistently throughout the years? Inconsistently?
When faced with a new campaign or fundraising project, you’ll certainly want to take advantage of every resource available. If you’re able to perform or outsource descriptive or predictive analytics, do so; but don’t forget to think small, as well. A simple frequency analysis could help you find your next “why.”