Driven Mad by Over-Segmentation, Part 2: Basic Principles of Sound Segmentation

In my first post on this topic I discussed common approaches to direct mail fundraising segmentation, and posited that we ought to step back and consider these two fundamental questions:

  1. Do we really know whether we’re getting meaningful ROI on our segmentation?
  2. Are we even set up to determine the answer to #1?

My argument, having worked with dozens of clients, is that the honest answer to both questions is no. I would further argue that the way we “do segmentation” can actually hinder our ability to answer these questions in the affirmative.

In my next post I will discuss an analytics-based approach to segmentation, but since a certain amount of audience-based segmentation is likely to continue as well, I will first suggest a few basic segmentation principles to inform that. There is no single correct answer, and it is not the main point here. However you decide to slice your constituent universe up, one of the most fundamental tenets of direct marketing is to do it in a way that actually permits you to test the effectiveness of some aspect of your effort – channel, content, timing, who signs the letter, etc. Too much segmentation and too many variables make it very difficult to do so. Perhaps we could take our lead from the world of academic research, in which experiments are carefully designed to test one variable at a time.

One consequence is that segments should be stable over some period of time. If constituents are constantly being grouped into different segments, the ability to gauge the effectiveness of different messages, channels, etc., can be lost. If the content is changed from one appeal to the next, a the composition of segments changes as well, differences in results cannot be accurately ascribed to the change in content. This may sound fairly obvious, but I do think this fact gets overlooked in practice due to the (sometimes vast) complexity of segmentation.

A closely related principle, therefore, is the use of test groups. If we want to test different content within a certain population, let’s say parents of current students, the most scientific approach would be to pull a random selection out of the overall group to use as the test group. There are different ways to accomplish this, and the more sophisticated fundraising CRM systems have functionality for this.

In my next post, I’ll present a data-driven alternative approach to direct marketing segmentation.

Brandon Ferris, Senior Director of Strategic Services and Fundraising Counsel


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 )

Facebook photo

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

Connecting to %s