CoolData blog

14 July 2010

Does “Do Not Solicit” mean they won’t give?

Filed under: Alumni, Annual Giving, Peter Wylie — Tags: , , , — kevinmacdonell @ 8:19 am

Guest post by Peter Wylie

(Click here: Do Not Solicit – 071310 to download Microsoft Word version of this paper.)

I think studying higher education alumni databases is a lot like studying a foreign Language. Take French. I’ve been a student of it off and on for over 50 years. I’ve gotten pretty good at it. So much so that native speakers often compliment me on my fluency. When they do, I thank them graciously and quickly add “Plus je sais, plus je ne sais pas.” The more I know, the more I don’t know.

For over a decade I thought I “knew” that alumni who tell their alma maters not to ask them for money meant two things: (1) Don’t ask me, and (2) I’m not going to give you anything. Then a good friend and colleague sent me some data for a project that may end up being pretty cool. Of the 50,000 or so records she sent, over 7,000 had a “do not solicit” tag. Okay. That’s clear. Don’t call ’em. Don’t send ’em letters or e-mails. And definitely don’t go knocking on their doors just because you happen to be in the neighborhood. That’s all pretty clear.

But I’m nosy; I checked to see if any of those 7,000 or so had ever given the school any money. It turns out that a little over a third of them had given some money. I couldn’t tell how much because my friend had not given me dollar amounts. Just whether or not they’d ever given a hard credit gift to the school. (We’d agreed that actual amounts would distract us from the goals of the project.)

I couldn’t stop there. I started digging around to see what some of the differences were between the ones who’d given and the ones who hadn’t. I think I found some interesting stuff. Of course, that doesn’t mean what I found applies to other schools. Maybe it does. Maybe it doesn’t.

For me, that’s not so important. What’s important is that young folks (like Kevin MacDonell, the creator of this great blog) who carry the vanguard of data-driven decision making in our field will take this topic farther than I have. And that can’t be a bad thing.

In the rest of this piece I’ll cover three topics:

  • Some of the interesting differences between the givers and non-givers (who say “Don’t solicit.”)
  • The model I used to “predict” the givers
  • Some concluding thoughts

Some Interesting Differences

In a moment we can take a look at Figures 1-6 that show some variables on which “Do not solicit” alums differ markedly when it comes to giving. Before we do that, however, I think it’s important to point out something we don’t know. We don’t know when or under what circumstances any of these alums told the school they did not want to be solicited. For example,

  • Did they do it recently or a long time ago?
  • Did they do it on a permanent basis, or just temporarily? It’s possible that some of them are parents of kids attending the school and their attitude for the next four years is, “Hey, I’m paying an arm and a leg for tuition. Until I’m done with that, please don’t go asking me for money while I’m in hock up to my underwear.”
  • Are they bent out of shape over the beloved football coach who was finally let go after ten consecutive losing seasons – something they may eventually get beyond? Or is it simply a case of, “I’ll give you something, maybe a lot, when I’m ready. In the meantime, don’t bug me?”

Again, we don’t know. But before any advancement person uses the results of a predictive model like the one I lay out here, they should consider these sorts of possibilities. More about that later.

Let’s take a walk through the figures. After each one I’ve made a short comment or two to make it clear what the figure is conveying.

There are huge difference in the giving rates among three types of alums. Undergraduate alums are almost twice as likely to give as graduate alums and more than six times as likely to give as non-degreed alums.

Alums who have attended at least one reunion are two and a half times as likely to give as alums who have never attended a reunion.

Alums who have never attended an event are less than half as likely to give as alums who’ve attended one event and about a third as likely to give as alums who have attended two or more events.

Alums who are members of the online community are twice as likely to give as alums who are not members.

Alums who were members of a Greek organization as undergrads are almost twice as likely to give as alums who were not Greek members.

Alums who are children of alums are two and a half times as likely to give as alums whose parents are not alums.

Clearly, these figures (and several I haven’t included) show there are a number of variables in the alumni database at this school that can be used to predict which “do not solicit” alums may be more likely to give in the future. One way to find out if we’re right is to (a) build a model that yields a “likelihood of giving score” for each of these alums and (b) begin testing the model.

Building a predictive model

I have to admit that, here, I was a bit torn about how much technical detail to go into. If I take you minute step by minute step through the model building process, I risk both confusing you and boring you. That wouldn’t be good. On the other hand, if I go too light on the details, you may say, “Come on, Pete, you haven’t given me enough info to see if I can test your results at my own school.”

So … how about this as a compromise?

Using multiple regression (if you know what that is, great; if you don’t, not to worry), I created a score for each alum where EVER GAVE (0/1) was the outcome variable and where the variables you see above as well as the following were the predictor variables:

  • Count of current volunteer activities
  • Count of past volunteer activities
  • Year the alum graduated (or should have if he or she had completed a degree “on time”)

The model generated well over 2,500 different score levels into which the 7,393 alums  could fall – way too many for anybody to get their conceptual arms around. The adjusted R squared for the model was about 36%. (Again, don’t worry if you don’t know what R squared means.)

I divided the 2,695 score levels into 10 groups of “deciles” containing about 740 alums each. As you look at Table 1, you’ll see these groups varied some in size. In Decile 1 (the lowest scoring 10% of alums) there are 726 people. In Decile 10 (the highest scoring 10% of alums) there are 739 people.

TABLE 1: Frequency Breakdown of “Do Not Solicit” Alums by Score Decile

If the model is to be useful in identifying “Do not solicit” alums who are likely to give, the number and percentage of givers should increase as the deciles increase. A look at Table 2 and Fig. 7 show that these numbers and percentages do just that. For example, in the first decile, of the 726 alums, only 12 (1.7%) have ever given anything to the school. In the tenth decile, of the 739 alums, 649 (87.8%) have given to the school.

TABLE 2: Number and Percentage of “Do Not Solicit” Alums Who Gave by Score Decile

Some Concluding Thoughts

You may have read other stuff I’ve written on data mining and predictive modeling. If you have, could be you’re tired of hearing me say that higher education advancement offices ignore most of the data they have on their alums as they go about the business of raising money from those alums.

Well, no rest for the weary here. This little study is a good example of what I’m talking about. Who would have thought there would be such striking differences between givers and non-givers who ask not to be asked? Not me. I just stumbled onto it because I was playing around with data that had been put together for a totally different reason. And I think that’s my point. When we’re talking alumni databases, there are oceans of data that could help us save a lot of money and generate a lot more revenue for some very worthy missions. But with the drops of analysis we’re currently doing on all that data, we’re not saving all that much money nor generating that much more revenue. We’re not. And that needs to change.

If you’re reading this, I suspect I’m preaching to the choir. It’s probably not you that needs the convincing on this matter. It’s probably the big bosses where you work that need the exhortation and cajoling. So the next time one of ‘em makes noises about spending big money on some product or service that’s designed to “prepare for the campaign” or whatever, you might say something like: “That’s cool. That’s great. But let’s not forget about all that data we’ve got just sitting there waiting to help us identify individuals who can play a major role in this project.” If they look intrigued, start pitching them. If they don’t, don’t give up. Take another run at them in a few months.

Back to the specific topic of this paper. Try to build a predictive giving model for your own “Do not solicit” alums. If you’re not proficient with using statistical software, find somebody in your school who is and get them to help you. If you have an Office of Institutional Research (or some similarly titled entity), that’s a good place to look. Just make sure the person you choose grasps the basic idea of what you’re trying to do and has the capacity to explain technical stuff in plain English.

Then do some in depth research on the high scoring alums that emerge from your model, especially those who’ve been generous givers over the years. Share the names with some of your colleagues, whether they’re involved in the annual fund, prospect research, or part of your cadre of gifts officers. My bet is that at least one alum is going to pop out of the mix who is teed up for a real nice appeal if your group comes up with the right strategy.

Good luck and let us know how it turns out.



  1. I’m surprised at the differences from the low to the highest deciles for the model you built, but there’s one issue that you didn’t touch on. We have these entities coded “Do not solicit” by their own choice – how do we get around that fact when trying to solicit them again? Putting them back into the calling pools could create a lot of negative comments and annoy a lot of people who already told us to stop calling them at least once. Don’t we need to abide by their wishes? If we all we have is “Do not solicit” we can’t just remove that code arbitrarily -it’s not a “Do not solicit for 3 years, then it’s ok” code.

    Comment by Peter — 14 July 2010 @ 8:44 am

    • Peter – yes, it is surprising, and I think the important message in Peter Wylie’s paper is that while we tend to write off the ‘Do Not Solicit’ people as donors, in some cases we may be doing so a little too hastily. HOW we address that issue? That’s not a simple answer, but definitely it’s a case of “Proceed with extreme caution.”

      I think we should think of DNS in the same way we think of every other type of leakage and attrition in our prospect pools such as lost alumni, disconnected phone numbers and so on: We need to follow up to potentially rectify the situation, and prevent it from happening in the first place. I have a few thoughts on this, which I will share in a follow-up post on a future day.

      Comment by kevinmacdonell — 14 July 2010 @ 9:29 am

  2. I’ve always wondered about the “Do not solicit” codes. Something tells me that creative fundraisers can find a way to involve those individuals – especially if there is a method to finding the most likely to give!

    Comment by afpsuncoast — 15 July 2010 @ 3:11 pm

  3. ummm… that comment was not really the AFP Suncoast chapter it was actually me. Sorry, I didn’t realize I was logged in!

    Comment by aspireresearch — 15 July 2010 @ 4:50 pm

  4. […] started talking about the topic of Do Not Solicit codes in university databases, the subject of his guest post from last week, I knew it would be a hot topic. Restrictions on solicitation and contact are coming under […]

    Pingback by Come Here, Go Away: Rethinking “Do Not Solicit” « CoolData blog — 20 July 2010 @ 3:16 pm

RSS feed for comments on this post. TrackBack URI

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

Blog at

%d bloggers like this: