Happy, happy Wednesday, my SEO heroes and comrades! Search Engine Academy brings you yet another little article on information architecture (IA) and search engine optimization (SEO).

We’re talking about how to implement IA, and one of the ways you want to figure out what’s the best IA for your particular site is to get feedback. Last time, we talked about some card sorting strategies; this week we are going to go over a few different ways to do that card sorting.

Oh by the way, I learned what IA is, how it ties into SEO, and how to make it happen in this hella-great book “Information Architecture for the World Wide Web” (Peter Morville/Louis Rosenfeld). It’s been several posts since I’ve given it a shout out.

Ready? Here goes…

Open/Closed Card Sorting

Alrighty then, you can offer open card sorting where the users create their own cards and category labels. Get several users to do this, and if there’s a majority trend, go with that for your IA. On the other hand, closed card sorting is where you’ve created the cards and categories. Use the open method to discover what users would like to have on your site; use the closed method to validate your open results.

Phrasing Card Sorting

You create cards with a keyword, a keyword phrase or images. Feel free to make the labels a question, an answer or a task.


It’s up to you if you want to make your cards really high level or very down-in-the-weeds detailed. Maybe you’re only working on a sub-domain of your site, or just a specific content subject matter, like “IT Security” or “Self-Help Books.”


If you have your primary site navigation set, and you’re trying to create alternation navigation streams, use cross-listing in multiple categories to help you set up those additional navigation schemes on your site.

Card sorting is so popular, there are software packages that allow you to set up and administer random card sorting. Here’s one called Mind Canvas. It’s even specifically for IA, whoo hoo!

I have no idea how it costs; the site is one of those where you fill out the contact form. Prices aren’t listed. I guess it’s lot like buying a large yacht – if you have to ask how much it is, you can’t afford that sucker. But anyways.

Back to more card sorting – track the amount of time two cards are rubbed together. The more they are together, the higher the likelihood is they are very, very related.

Also, if a specific card is put in the same category – what percentage of time is that? The higher the time, the more it is a candidate for that label.

If you are very sophisticated, have a lot of cash, and are a stats geek, take this data and shove it into statistical analysis software and see what pretty pictures come out the other end. They can be very complicated, and you probably want to have Nate Silver translate for you, but it’s great for seeing patterns.

Otherwise, if you are a do-it-yourself kind of creature, do your best to make a diagram into an image called an “affinity model.”  It’s a pictorial map or diagram of the hierarchy your card sorting experiments yielded.

Please keep in mind that the more users you have to test and do your card sorting, the better your model will be. If you can only drag five people into a room and four of them are your relatives, that’s not a comprehensive, neutral test. They might be a little biased for you!

This is enough for today, SEO podnas. Next up, we’ll talk about user testing and IA. Until next week, keep it between the ditches!

All the very best to you,