Segmentation’s goal should be to put similar groups together.  You might have a lapsed donor segment, because a male lapsed donor looks more like a female lapsed donor than he looks like a male active donor, for example.

This came to mind when I was looking at NonProfit Tech for Good’s Data for Good.  This survey of online donors has some interesting data points on year-end giving (some are no surprise: did you know that online donors generally prefer giving online?).  But one of the goals form the study is to talk about “the impact of gender, generation, and ideology [and how they] impact philanthropic giving.”

I think all of us have been asked about how to get younger donors (or, worse, Millennial donors).  Or had recommendations to focus on one side of the gender continuum (in fact, I once had two agencies recommend focusing more on men and two recommend focusing on women – simultaneously.)

But how do those segmentations stack up against ideology – a non-demographic factor?  It’s certainly not going to be as good as something that is charity-specific (e.g., I give to the charity that aims to end the disease from which my loved one died). So, if it was more predictive of giving than demographics, then we probably should be ignoring demographics as a segmentation strategy. (Spoiler: it is and we should.)

So let’s go to the tape:

Gender

The largest change in preferences is about 20%. Women are more likely to support animal and women/girls charities. Men are more likely to support religion and faith and health and safety charities.

Generation

The biggest difference is millennials are 32% more likely than average to support human and civil rights organizations.  Boomers are slightly more likely to support religion and faith charities. Most other differences are in rounding error territory.

Ideology

Here we see some real differences.  Conservatives are almost three times as likely as the average to support religion and faith charities. Liberals don’t have religious charities in their top five groups.  Liberals are 50% more likely to support human and civil rights organizations and 25% more likely to support environmental causes.

So, if you wanted to figure out whether someone would support your organization and could get only one data point – gender, generation, or ideology – you’d pick ideology every day and twice on Sundays (when many conservatives would be at the religious service of their choice).

And, as mentioned earlier, ideology isn’t even a particularly strong predictor.  You’d much rather know if they are cat or dog people if you are an animal charity.  Or if a disease has harmed them if you are a health charity.  Heck, I’d rather know if someone is an active attendee at religious services than their ideology if I were at a religious charity.  After all, there are religious liberals and secular conservatives.

In short, demographics are the second-best way to segment your file.  The best way is almost any other way.  Especially those ways that focus on what the donor has told you about their identity and preferences.