Donor volatility: testing years of non-giving as a predictor for the next big gift

Guest post by Jessica Kostuck, Data Analyst, Annual Giving, Queen’s University

 

During my first few weeks on the job, my AD set me up on several calls with colleagues in similar, data-driven roles, at universities across the country. One such call was with Kevin MacDonell, keeper of CoolData, with whom I had a delightfully geeked-out conversation about predictive modeling. We ran the gamut of weird and wonderful data points, ending on the concept of donor volatility.

 

When a lapsed high-end donor has no discernable annual giving pattern, is it possible to use his or her years of non-giving to predict and influence their next big gift?

 

Our goal for our Annual Giving program was to identify these “volatile” donors (lapsed high-end donors with an erratic giving history), and reactivate (ideally, upgrade) them, through a targeted solicitation with an aggressive ask string.

 

(For more on volatility, see Odd but true findings? Upgrading annual donors are “erratic” and “volatile”, which describes findings that suggest the best prospects for a big upgrade in giving are those who are “erratic”, i.e. have prior giving but are not loyal, every-year donors, and “volatile”, i.e. are inconsistent about the amounts they give.)

 

I did some stock market research (see footnote), decided on a minimum value for the entry-point into our volatility matrix ($500), and, together with Senior Programmer Analyst, Kim Wilkinson, got cracking on writing a program to identify volatile donors.

 

 

 

Our ideal volatile donors had given ≥ $500 at least twice in the last 10 years, without any consecutive (“stable”) periods. Year over year, our ideal volatile donor would act in one of three ways – increase their giving by at least 60%, decrease their giving by at least 60%, or not give at all. Given the capacity level displayed by these volatile donors, we replaced years of very low-end giving <$99) with null values (“throwaway gifts”).

 

We had strict conditions for what would remove a donor from our table. If a donor had two years of consecutive giving within a ±60% differential from their previous highest giving point (v_value), we considered this a natural (or, at least, for this test, not sufficiently irregular) fluctuation in giving, and they were removed from the table. If the donor had two consecutive years of low-end (but not null) giving ($99-$499), this was considered a deliberate decrease, and they, too, were removed. Conversely, if a donor had two consecutive years of greatly increased giving, this was considered a deliberate increase, and they were also removed.

 

At any point, a donor could be admitted, or readmitted into our volatility matrix, by establishing, or re-establishing, a v_value and subsequent valid volatility point.

 

The difference between a lapsed donor and a volatile donor

 

Below is a sample pool of donors we examined.

 

 

Donor 1 is volatile all the way through, with greatly varying levels of giving, culminating in two years of non-giving. Donor 1 is currently volatile, and thus enters our test group.

 

Donor 2 is volatile for two years – FY07-08 and FY08-09 (v_value of $5,000 in FY07-08, followed by a valid volatile point in FY08-09 with a decrease of -80%), but then is removed from the table in FY09-10 with only a -50% decrease in giving. They do not establish a new v_value, even though their FY09-10 giving meets the minimum giving threshold for this test, because of their consecutive, only marginally decreased giving in FY10-11. This excludes Donor 2 from our test.

 

Donor 3 enters our volatility matrix in FY04-05, leaves in FY07-08, reenters in FY10-11, and maintains volatility to current day, and, thus, enters into our test solicitation.

 

While all three of these donors are lapsed, and are all SYBUNTs, only Donor 1 and Donor 3 are, by our definition, volatile.

 

Solicitation strategy and results

 

We now had a pool of constituents who were at least two years lapsed in giving, who all had a history of inconsistent, but not unsubstantial, contributions to the university. In an email solicitation, we presented constituents with both upgrade language and an aggressive ask matrix, beginning at a minimum of +60% of their highest ever v_value, regardless of where they were in the ebb and flow of their volatility cycle. Again, the goal of this test was to (1) identify donors with high capacity (2) whose giving to the university was erratic in frequency and loyalty and (3) encourage these donors to reactivate at greater than their previously-established high-end giving.

 

In our results analysis, we broadened our examination to include any gifts received from our testing pool within the subsequent four weeks, not just gifts linked to this particular solicitation code, to verify the legitimacy of tagging these donors as volatile – that is, having a higher-than-average probability to reactivate at a high-end giving level.

 

An important part of our analysis included comparing our testing pool to a control pool, pairing each of our volatile donors with a non-volatile twin who shared as many points of fiscal and biographic information as was possible.

 

Within the four-week time frame, our test group had about a 7% activity rate, whereas our control group had an activity rate of about 5% (average for the institution during this timeframe). Within our volatility test group, 50% of donors gave an amount that would plot a valid point on our volatility matrix.

 

Conclusion and next steps

 

Through our experiment, we sought to identify volatile donors, and test if we could trigger a reactivation in giving, ideally at, or greater than, their highest level on record.

 

Since not all of the donors within our test group made their gifts to the coded solicitation with the volatile ask matrix, it is indiscernible whether being presented with language and ask amounts that reflected their elusive giving behavior prompted a gift – volatile or otherwise. However, we do feel confident that we’re onto something when it comes to identifying and predicting the behavior of a particular, valuable set of donors to our institution.

 

Our above-average response rate (both versus the control group, and institution-wide) supports our “theory of volatility”, insofar as that volatile donors are an existing pool with shared behaviors within our donor population. We plan to re-run this test again at the same time next year, continuing our search to find a pattern within the instability.

 

Were we able to gather definitive results that will define and shape future annual giving strategy? Not exactly. But as far as data goes, this was definitely cool.

 

Jessica Kostuck is the Data Analyst, Annual Giving at Queen’s University in Kingston, Ontario. She can be reached at jessica.kostuck@queensu.ca.

 

————-

1. Varadi, David. “Volatility Differentials: High/Low Volatility versus Close/Close Volatility (HVL-CCV).” CSS Analytics. 29 Mar. 2011. Web. Winter 2015.

Odd but true findings? Upgrading annual donors are “erratic” and “volatile”

In Annual Fund, Leadership giving typically starts at gifts of $1,000 (at least in Canada it does). For most schools, these donors make up a minority of all donors, but a majority of annual revenue. They are important in their own right, and for delivering prospects to Major Giving. Not surprising, then, that elevating donors from entry-level giving to the upper tiers of the Annual Fund is a common preoccupation.

It has certainly been mine. I’ve spent considerable time studying where Leadership donors come from, in terms of how past behaviours potentially signal a readiness to enter a new level of support. Some of what I’ve learned seems like common sense. Other findings strike me as a little weird, yet plausible. I’d like to share some of the weird insights with you today. Although they’re based on data from a single school, I think they’re interesting enough to merit your trying a similar study of donor behaviour.

First, some things I learned which you probably won’t find startling:

• New Leadership donors tend not to come out of nowhere. They have giving histories.
• Their previous giving is usually recent, and consists of more than one or two years of giving.
• Usually those gifts are of a certain size. Many donors giving at the $1,000 level for the first time gave at least $500 the previous year. Some gave less than that, but $500 seems to be an important threshold.

In short, it’s all about the upgrade: Find the donors who are ready to move up, and you’re good to go. But who are those donors? How do you identify them?

It would be reasonable to suggest that you should focus on your most loyal donors, and that RFM scoring might be the way to go. I certainly thought so. Everyone wants high retention rates and loyal donors. Just like high-end donors, people who give every year are probably your program’s bread and butter. They have high lifetime value, they probably give at the same time of year (often December), and they are in tune with your consistent yearly routine of mailings and phone calls. Just the sort of donor who will have a high RFM score. So what’s the problem?

The problem was described at a Blackbaud annual fund benchmarking session I attended this past spring: Take a hard look at your donor data, they said, and you’ll probably discover that the longer a donor has given at a certain level, the less likely she is to move up. She may be loyal, but if she plateaued years ago at $100 or $500 per year, she’s not going to respond to your invitation to join the President’s Circle, or whatever you call it.

Working with this idea that donor  loyalty can equate to donor inertia, I looked for evidence of an opposite trait I started calling “momentum.” I defined it as an upward trajectory in giving year over year, hopefully aimed at the Leadership level. I pulled a whole lot of data: The giving totals for each of the past seven years for every Annual Fund donor. I tried various methods for characterizing the pattern of each donor’s contributions over time. I wanted to calculate a single number that represented the slope and direction of each donor’s path: Trending sharply up, or somewhat up, staying level, trending somewhat down, or sharply down.

I worked with that concept for a while. A long while. I think people got sick of me talking about “momentum.”

After many attempts, I had to give up. The formulas I used just didn’t seem to give me anything useful to sum up the variety of patterns out there. So I tried studying some giving scenarios, based on whether or not a donor gave in a given year. As you might imagine, the number of possible likely scenarios quickly approached the level of absurdity. I actually wrote this sentence: “What % of donors with no giving Y1-Y4, but gave in Y5 and did not give in Y6 upgraded from Y5 to Y7?” It was at that point that my brain seized up. I cracked a beer and said to hell with that.

I tried something new. For each donor, I converted their yearly giving totals into a flag that indicated whether they had giving in a particular year or not: Y for yes, N for no. Imagine an Excel file with seven columns full of Ys and Ns, going on for thousands of rows, one row per donor. Then I concatenated the first six columns of Y/Ns. A donor who gave every year ended up with the string “YYYYYY”. A donor who gave every second year looked like “YNYNYN” — and so on.

I called these strings “donor signatures” — sort of a fingerprint of their giving patterns over six years. Unlike a fingerprint, though, these signatures were not unique to the individual. The 15,000 donors in my data file fit into just 64 signatures.

A-ha, now I was getting somewhere. I had set aside the final year of giving data — year seven — which I could use to determine whether a donor had upgraded, downgraded or stayed the same. All I had to do was take those 64 categories of donors and rank them by the percentage of donors who had upgraded in the final year. Then I could just eyeball the sorted signatures and see if I could detect any patterns in the signatures that most often led to the upgrading behaviours I was looking for. (This is much easier done in stats software than in Excel, by the way.)

All of the following observations are based on the giving patterns of donors who gave in the final two years, which allowed me to compare whether they upgraded or not. This cut out many possible scenarios (eg., donors who didn’t give in one of those two years), but it was a good starting point.

I confirmed that the more years a donor has given, the more likely they are to be retained. BUT:

• The more previous years a donor has given consecutively, the LESS likely they are to upgrade if they give again.
• A donor is markedly more likely to upgrade from the prior year if they have lapsed at least one year prior to giving again.
• Specifically, they are most likely to upgrade if they have one, two or three years with giving in the previous five. More than that, and they are becoming more loyal, and therefore less likely to upgrade.
• Donors who give every other year, or who have skipped up to two years at a time, are most likely to upgrade from last year to the current year.

I told you it was counter-intuitive. If it was just all obvious common sense, we wouldn’t need data analysis. Here’s more odd stuff:

• In general, the same qualities that make a donor more likely to upgrade also make a donor upgrade by a higher amount.
• By far, the highest-value upgrader is a last-year donor who lapsed the previous year but had three years of giving in the previous five.
• The next-highest donor signatures all show combinations of repeated giving and lapsing.
• As a general rule, the highest-value upgraders have about an equal number of years as a donor and as a non-donor.

The conclusion? Upgrade potential can be a strangely elusive quality. From this analysis it appears that being a frequent donor (three or four years out of the past six) is a positive, but only if those years are broken up by the odd non-giving year. In other words, the upgrading donor is also something of an erratic donor.

I thought that was a pretty nifty phenomenon to bring to light. I decided to augment it by trying another, similar approach. Instead of flagging the simple fact of having given or not given in a particular year, this time I flagged whether a donor had upgraded from one year to the next.

Again I worked with seven fiscal years of giving data. I was interested in the final year – year seven – setting that as the “result” of the previous six years of giving behaviour. I was interested only in people who gave that year, AND who had some previous giving in years 1 to 6. The result set consisted of “Gave same or less” or “Upgrade”, and if upgrade, the average dollar upgrade.

The flags were a little more complicated than Y/N. I used ‘U’ to denote an upgrade from the year previous, ‘S’ to denote giving at the same level as the year previous, ‘D’ for a downgrade, and ‘O’ (for “Other”) if no comparison was possible (i.e., one or both years had no giving). Each signature had five characters instead of six, since it’s not possible to assign a code to the first year (no previous year of giving in the data to compare with).

This time there were 521 signatures, which made interpretation much more difficult. Many signatures had fewer than five donors, and only a dozen or so contained more than 100 donors. But when I counted the number of upgrades, downgrades and “sames” that a donor had in the previous five years, and then looked at how they behaved in the final year, some clear patterns did emerge:

• Donors who upgraded two or more times in the past were most likely to upgrade again in the current year, and the size of their upgrade was larger, than donors who upgraded fewer times, or never upgraded. Upgrade likelihood was highest if the donor had upgraded at least four times in the previous five years.

• Donors who gave the same amount every year were the least likely to upgrade — this is the phenomenon people were talking about at the benchmarking meeting I mentioned earlier. Donors who never gave the same amount from one year to the next, or did so only once, had higher median upgrade amounts.

• And finally, the number of downgrades … this paints a strongly counter-intuitive picture. The more previous downgrades a donor had, the more likely they were to upgrade in the current year!

In other words, along with being erratic, donors who upgrade also have the characteristic that I started to call volatility.

I wondered what the optimum mix of upgrades and downgrades might be, so I created a variable called “Upgrades minus Downgrades”, which calculated the difference only for donors who had at least one upgrade or downgrade. The variable ranged from -4 (lots of downgrades) to plus 5 (lots of upgrades). What I discovered is that it’s not a balance that is important, but that a donor be at one extreme or the other. The more extreme the imbalance, the more likely an upgrade will occur, and the larger it will be, on average.

ERRATIC and VOLATILE … two qualities you’ve probably never ascribed to your most generous donors. But there it is: Your best prospects for an ambitious ask (perhaps a face-to-face one) might be the ones who are inconsistent about the amounts they give, and who don’t care to give every year.

By all means continue to use RFM to identify the core of your top supporters, but be aware that this approach will not isolate the kind of rogue donors I’m talking about. You can use donor signatures, as I have, to explore the extent to which this phenomenon prevails in your own donor database. From there, you might wish to capture these behaviours as input variables for a proper predictive model.

At worst, you’ll be soliciting donors who will never become loyal, and who may not have lifetime values that are as attractive as our less flashy, but more dependable, loyal donors. On the other hand, if you put a bigger ask in front of them and they go for it, they may eventually enter the realm of major giving. And then it will all be worth it.