Rachel Lovinger (@rlovinger) just published a great piece on categorizing, called “Splitting Tigers, Lumping Rabbits,” on Scatter/Gather. I love her simple, elegant advice: “You just need to find the right balance between lumping and splitting.”

Since I read it, I’ve been wondering: How do you find that balance? Is it just some feeling that comes upon you when you have all the pieces in the proper order? Is it like sorting male and female chicks?—something that is learned unconsciously through experience? Is there some way to work it out systematically?

I believe that finding the balance lies in discovering which distinctions make the most difference for the users of your content. If you can articulate what makes this thing different from that one, and why that difference matters to your users, then you will have identified the dimensions of difference. You will also have created a test for your categories, your labels, your navigation, and perhaps even the whole content strategy for your website.

Discovering the dimensions of difference

So let’s imagine a standard categorizing scene. Say we trot out my favorite brainstorming tool: The Affinity Diagram. You know the one. First, I ask a group of hapless users to write all sorts of content ideas onto sticky-notes. Then I have them post them randomly on a wall, and then organize them—all together—into clumps. Finally, I ask them to name the clumps. Inevitably and with supreme confidence, they dub one of the groupings “resources.”

There’s always a “resources” category, isn’t there? As professional categorizers, you and I are pretty confident that “resources” wields little descriptive power. Here’s where Rachel’s finding balance comes in for me.

You have to dig into the clumps for some more scrutiny. Take the stickies apart to identify all the ways in which the items are alike and different. Name the distinctions, at this point, rather than the categories. It’s another round of brainstorming, really. Get crazy! Split the hairs more and more finely, until you need a microscope to see them.

Then clump the differences together as a sort of meta-categorization, and rank them (subjectively, of course) according to their clumping strength. Which of these dimensions account for most of the difference among the items?

Patterns should begin to emerge, which should point pretty clearly toward the labels that will have the greatest power to communicate the items they organize.

Patterns within patterns

As a sort of bonus track, I just remembered another piece I’ve read recently—Indi Young’s Mental Models, in which she uses “Task-Based Audience Segmentation” to uncover the underlying dimensions of difference that shape users’ mental models. I believe that she is talking about the same, basic process as finding the balance between lumping and splitting: In order to identify the distinctions that make the most difference to your users, you must first understand the underlying patterns that shape those distinctions.

Just as a community is more than a collection of individuals, but rather an organic whole of interrelationships, so a categorization scheme is more than a collection of labels: It is a complex weave of distinctions that make all the most important differences to the ones who rely upon it. The labels should point toward and name the underlying patterns of difference, even when the users aren’t consciously aware of them. When you can describe those patterns of difference, you’ll have a way to find the right categories and name them intuitively for your users.