Empowering Fundraisers with the Results of Predictive Analytics; 5 Questions with Rich Horne
Recently, The Chronicle of Philanthropy released their results of a study on America’s giving habits, both state by state and by income levels.
In the study, The Chronicle study found that “Americans give, on average, about 3 percent of their income to charity, a figure that has not budged significantly for decades. However, that figure belies big differences in giving patterns between the rich and the poor.
The wealthiest Americans—those who earned $200,000 or more—reduced the share of income they gave to charity by 4.6 percent from 2006 to 2012. Meanwhile, Americans who earned less than $100,000 chipped in 4.5 percent more of their income during the same time period. Middle- and lower-income Americans increased the share of income they donated to charity, even as they earned less, on average, than they did six years earlier.”
There was also a counterpoint article that called in to question the methods and how to interpret the results.
We thought this was a great opportunity to do a Five Questions piece with one of our fundraising customers about the use of data and how it is used and interpreted. So we contacted Rich Horne, Associate Director, Research, Development at Yale University, to discuss the articles, interpreting data, and how predictive modeling is utilized in Fundraising at Yale.
The study noted the results, state by state, of itemized deductions of donors, using IRS data. Since we love data, we were intrigued by the approach. As someone who uses data all of the time for different fundraising campaigns, what did you think of their methods?
Taking both of the articles for context, regarding the approach taken by the Chronicle, well, we love government data but, it’s a very particular thing. As the National Journal article pointed out, the comparable but not identical Gallup Poll that bore on analyzing charitable giving data for folks, you have to be careful of what you are asking and what sort of thing you expect to glean from the data you are interrogating. Our experience with IRS data has always been great. I think that data.gov is a wonderful thing and IRS data is accurate but sometimes it’s “bucket-ized” which can dilute the granularity of it somewhat. I think in situations comparing information across the country you are fine. But state-by state comparisons, I know we don’t have much of an appetite for that sort of thing because it masks very important sub-trends and other compartmentalized and pocketed behavior.
So, generally speaking, “IRS data good!” It doesn’t intersect that well with Yale’s particular world of prospects because they tend to eclipse specific areas. I personally find this sort of research gratifying to engage with intellectually but it doesn’t deal too directly with the kind of varied people that Yale is mainly interested in.
Both articles were very interesting as they call to the front the kinds of ways that you can get mired in potentially troublesome conclusions related to what the data is saying, especially how you couch it and whatnot. This concern resonates in me and my partner, Deepti, quite a bit because we have to be very careful in how we communicate what it is we think we’re seeing when we are doing our own analyses. A lot of times, that includes specifically addressing what it is we sought to find and what kind of data we were looking at—be it IRS data or any of our own data. Then, we ensure that the people who we communicate those analyses to know exactly what we are attempting to do so they can actually help change the strategy if we are not asking specifically for what it is that they requested and so we can answer the questions that they find pressing.
The fact they only included itemized deductions – do you think that skews the data? Is there anything that could be done to compensate for that?
The second part of that question is excellent, I think. I don’t know how I would approach compensating for the fact that we have this type of information on folks who are itemizing their deductions. That to me at least connotes a level of sophistication. Obviously people are doing what they can to reduce their adjusted gross income as much as they can—according to what the law and the tax code will allow. That just makes good sense—rational acting. But yes, it does skew data in that you can’t speak too broadly about the implications of that without constantly drawing attention to the fact that this is the group that you are dealing with.
I think if I were to put together such an article or analysis, I think I would have on more than one point reminded the reader that the data that we’re looking at is a subset. It is not the universe whereas the Chronicle article pointed out that this accounted for about 80 percent of the Giving USA stats—that being what I assume is generally considered to be the most comprehensive stab at collecting giving information in the United States. It means something that someone is an “itemized deductor”. How much that would skew the data I don’t know that I would be able to say. I certainly would want to look into that because I can tell you that if we were to employ such data, one of the first questions would be “How do you think this affects the analysis you just performed on this group of people?”
We might use similar IRS data, for instance, to conceive of new marketing campaigns- direct mail, solicitations and that kind of thing. We certainly would be interested in knowing where we might want to increase the number of people that travel through a particular area if it’s known broadly to be an area where people are generous by nature.
Obviously, the Utah example that both articles highlighted is a good one and it is something that you should be aware of. Data does reveal these kinds of things fairly reliably. What you might lose is the kinds of pockets that would be identified where you would have folks who would be particularly philanthropic by nature but maybe they don’t itemize deductions. That clearly is not a tranche in terms of wealth or income to say that it is not significant. Not being able to count for fully 20 percent of giving numbers is something I would want to investigate further. Exactly how much it is skewed I can’t say but I think anybody would agree that it would.
On a more general level, what are the biggest hurdles facing fundraisers and how is data being used to tackle them?
The biggest hurdles facing fundraisers today are the same ones they have always been. We do want to be better at getting the most out of our fundraisers. Any fundraiser’s day is the same length as anyone else’s and in that day we want to ensure that they are able to help to discharge their duties as effectively as possible. What used to be just art and anecdote is now almost uniformly accompanied by and complimented by what data can provide in terms of informing their decisions. We can essentially use our resources more efficiently.
This still, of course, all rests on connecting with people in the right way. Fundraisers are the crux of that relationship and, when they do their jobs well, good things happen. Great data usage can make doing the same things we’ve always been doing well even better and help us achieve better results more easily by providing us with both narrower and broader lenses. Where should a fundraiser be on any given day across a particular month or year? What should we be expected to pull in terms of donations for selected causes?
Do we keep relying on the same kinds of projections? We don’t. Our projections are fairly robustly enabled by more helpful data and just more data in general.
So the biggest hurdles are the same. Data now allows us to assist fundraisers in coming up with their short term and longer term strategic goals. We are talking about not just the gift officers that go out and meet people but all the way up to the top—the chief fundraiser for your university or non-profit organization. It connects all of that and it makes it one thing. It’s an organic relationship and it is one that is highly informed by data in the kinds of ways that are important. For example, identifying pockets of relatively philanthropic people might be helpful.
How has predictive modeling changed the conversation between prospect researchers and gift officers?
Here at Yale, that has been one where you have an old Augustine institution where you have folks in fundraising where there is a natural affinity with the organization and Yale’s mission and how they can help.
Our conversation with Yale’s gift officers now includes predictive modeling as a component in their tool belt. We are comfortable with our general strategy. We know that we are going to do pretty well and we are very grateful for that. The departments such as Annual Fund and Planned Giving who have partnered with us to use predictive modeling value it greatly for its ability to help verify their longstanding wisdom on what kind of prospect is the best kind of prospect. Also then, they use it to help identify the next best group of donors. These are not the folks who are going to jump out intuitively to anyone who has been in the business of, for instance, Annual Fund Fundraising or Planned Giving Fundraising for years. Rather this tells them the next best group to target. That can be a very cumbersome and not too intuitive process if you are just doing basic data mining.
Our predictive modeling initiatives early on were basic- direct mail solicitation or even including people as part of an informative campaign to start them getting used to receiving information from us for very specific kind of charitable gifts. This is communication they may not have received otherwise for the Annual Fund- especially for large reunion years coming up. The fundraisers who are working with us think that it is a wonderful compliment to their own well-developed impression of what kind of person makes a great donor. They’re going to be able to look at the first 500 people on the list that we provide and say, “Yes, these names are not surprising to me. They all make sense.”
So then, the next 500 names (in descending order of probability) that are just below the ones they recognize, the fundraisers say “I now have confidence in them as well.” They don’t necessarily recognize them but they can see that what you are saying is they have a lot of the same characteristics and that feel comfortable to them. “I am going to go and implement those as part of my strategy. I will talk to the first 100 of those 500 and see how it goes.”
That’s the best thing that has happened here. How we have grown as a group to work together and complement each other’s information. We are starting to work with more schools as well—environmental sciences and the school of forestry.
We know that fundraising management is constantly being asked to increase their results. What do you see as the most critical assets in meeting this continuous growth expectation that your office can provide?
The single most critical asset from where we can contribute something is in helping to segment a very, very large donor or potential donor population into branches based on the likelihood of achieving what it is we are hoping folks will be able to do. We can understand better how to devote our limited resources and compete successfully in a quite competitive space.
Fundraising definitely borrows heavily from people’s good experience at the institution. There is some gratitude there and warm feelings. But we’re not the only thing that someone may have in their annual philanthropic considerations. They may also be quite interested in assisting non-profits whose missions are complimentary to Yale’s but are different. Some folks think it makes sense to give more money to smaller organizations who have very distinct missions that, for instance, may help with international crises or poverty alleviation. They may also want to give to religious organizations or hospitals who have helped them. So even though we may be their lone alma mater, it doesn’t mean that we should expect that folks should give us whatever bit they have set aside for us.
I think it’s important for us to be able to distinguish which donors are best for which matches here, with their own philanthropic interests that we can identify. We need to pursue them not just in a general way. I think those days are long gone. They are past us now. Many organizations still do that but I don’t think that’s the trend.
I think the trend now is, by using various sets of data we can bring all of that data into the fold here. We analyze those sets of data points that we can cross large enough populations and then be much better about identifying not just who a potential prospect might be but also base it on more information that we can access responsibly, and what they might be interested in here.
So, if you want to see if something is exciting to a fundraiser, you not just suggest that there is an addition to their portfolio who is a fairly wealthy individual but also that we have some evidence of what that person’s philanthropic interests have been. That gives us not just a new potential donor but also a bit of information about what they might find most interesting at Yale based on the fact that they have given to other organizations in a particular area. Their most critical assets center not just on identifying wealth because that is just a given. Being able to identify specific interests and then catering to those prospects and interests is key.
Harness the power of many.
Create and share reports and datasets across the enterprise, and put analytical power in the hands of everyone.
Veera creates a truly data-driven culture. Try it for yourself today.