Key Takeaways: 2011 NTC Segmenting Session
This session was called “A Scientist in Your Communications Department,” and Jeff Shuck, the presenter, had a ton of energy and insights to share. (It looks like the whole video is posted so you can see for yourself!)
Sally Heaven over at Convio posted her takeaways from this session, too. All the points she listed were ones I had written down, too, but I had somewhat different top takeaways:
Takeaway #1: Predictive analysis is different from the more common descriptive analysis. With predictive analysis, instead of just looking back at who did what, you use what you know to anticipate what people might do. I thought the example was really cool. An organization figured out to send information about the high-dollar giving program to people who engaged with them well after an event was over, since they were more likely to be committed to the cause.
How I might use it: I thought immediately of Artomatic, which is based around a huge arts event that happens only once every year or two. By this reasoning, the people who contact us when there’s no event on the horizon could be more likely to become volunteers. (That’s how I got involved, in fact…)
Today, someone posted on the Artomatic wall on Facebook asking when the next event is. We get those inquiries a lot, and I usually don’t give people an opening to volunteer. This time, I did, and sure enough, the poster was rarin’ to go. Within an hour she was connected with current Artomatic volunteers who can absolutely use her help.
Jeff mentioned that this kind of analysis involves fancy statistics like regression analysis, which I definitely did not do, but it was great to see the basic idea work out so nicely so quickly.
Takeaway #2: Don’t use averages alone! In very uneven distributions, the mean average paints a skewed picture. Jeff gave the example of a fundraising event that had an average gift of $58 — pretty nice. But that hid the fact that the median gift was zero.
How I might use it: We’re in the midst of putting together a new metrics dashboard for EDF’s web site. When we do one-off analyses our analyst will often toss out outliers that seem to be clouding the results. But incorporating medians into our new dashboard seems a more rigorous way of solving the same problem.
Takeaway #3: Ask why people support you. It’s the piece of data that can most affect how you communicate with them. For example, if you are running a walk to support cancer research, knowing that a walk participant is a cancer survivor or lost a relative to cancer is extremely important.
How I might use it: This is going to be really helpful at Project Create, a tiny nonprofit with the very big mission of giving free arts classes to at-risk and homeless kids. We have a new fundraising database that supports better interest tagging, and we realized just last week that we don’t actually know enough about why individual donors support us to use those tags well!
We’re having our annual art auction on April 7 (please join us!), and will talk with people there about their connections to us. We’ll follow the event with a survey seeking the same data in a less open-ended format. I’m looking forward to seeing the results of that.