UX Research & User Testing

Testing Your Information Architecture

Paul Boag

Paul Boag

UX Research & User Testing

Check out a free preview of the full UX Research & User Testing course

The "Testing Your Information Architecture" Lesson is part of the full, UX Research & User Testing course featured in this preview video. Here's what you'd learn in this lesson:

Paul explains the importance of testing the information architecture and introduces tree testing as a method to validate the structure. There are three types of content to test: low agreement score, not sure category, and top tasks buried in subsections. He also addresses potential problems and explains how to interpret the results, focusing on success score, directness, and time taken to complete tasks.


Transcript from the "Testing Your Information Architecture" Lesson

>> So, you've done all of this, however you've chosen to do your information architecture, whether you've just gone saw that let's get chatGPT to do the whole thing, right? Or whether you've gone or do some top task analysis and then what this and that and the other. The one thing I do recommend, however you end up with your information architect is to test it at the end, right?

You could skip everything I've said so far make up your own information architecture and just do this, right? This is your chance to validate what you've done. However you've done it, whether you've used ChatGPT, made it up, done it properly, whatever, this is your chance to make sure that it's not completely broken.

And we do that through tree testing, right? So tree testing is super, super simple. It allows you to test your assumptions and also if you've done any of the stuff before, like card sorting or any of the rest of it. Those things that were in the not sure group or were in the low agreement score group and you're going, is it gonna work, etc, now's your chance to test it, right?

So what we're gonna do? All right, let's first of all be clear about the kind of things we want to test in tree testing. So three things. Content with a low agreement score if we've done card sorting, right? So if the closed card sorting ended up with some stuff that you got, it was a bit ambiguous of where people should put it, that is a great thing to test for.

Second one is stuff aside to the not sure category, right? Great thing to test for but finally also top tasks that have been added into a subsection. So if you've buried one of those top tasks deep in the site there's a higher probability people won't want to find it and so we want to test for that so what I will do is.

I will identify which cards I'm worried about. Because it's got a low agreement score because it's in a subsection when it's a top task or it's in not sure. I try not to come up with more than about ten of them because otherwise it becomes laborious to do.

But I come up with about ten and then for each one, I'll write a question that avoids using the word on the card, right? So let's say dental care, right? I would write a question like, you have to go to a dentist and you want to know whether you are covered.

Where would you look, right? So haven't used the word dental care okay? But I've tried to word it very neutrally. Again, Chatgpt can help you with this if you're struggling. And then that is the question. So let me show you how you set it up and it will become clearer.

So first thing you do is you go in and you create a task, right? So I put that quick question in right and matter in this part case. It's got an example of imagine that you want to update your credit card details where would you find that right?

Great example of a question then what I do so I create a new tree test to do that I input the whole of the sites hierarchy right? So all of those sections, subsections the information architecture I've created. I put all of that in so it knows that. Then I create each one of my tasks like the, for example, the one about dental care or the credit card details.

And then for each task I set the proper destination of where they should be looking, so it knows whether it's the right or wrong answer, right? And then we send that out and we get people to look at it again. That is the most valuable thing. If you can only do one thing with your information architecture, that is the thing you do, right?

Things you're worried about that people might not be able to find create series of questions, put that out and do that. And then when you look at the results, what you're looking for is it will give you a screen page like you can see here. Where for each task is giving you the success score the directness and how long it's taking them to do it.

So the success score is fairly obvious that's could they complete find the piece of information the directness is did they go directly to it down the hierarchy. Or did they wander around slightly lost before finding it? And the time is how long it took them to do it.

So that each time you want to pay attention to the success score, right? And if they failed where they ended up, right? And that's a critical one. So you can actually I don't know whether you probably can't see it, but basically it tells you how many people failed.

And you can click on that and it will tell you where they ended up when they failed. Right now, that is really valuable because what you can often do is one will stand out way above the others. I see, everybody's ended up here, right? And then you can make a decision either I'm gonna swap it over and put this content in that section.

Or again, you use cross linking to redirect people back to where they're supposed to be, so that's really useful. The second thing you're looking for is the directness and the route that people took. You wanna make sure people aren't wandering around too lost, although a little bit of lostness is fine, as long as they don't spend too much of their time doing that.

And it's always good to look at the time it took people put a complete the task on average. So very quickly, what about the problem, so that you've done a series of 10 tests, and there are some problem ones there that you struggle with for whatever reason. When in most cases, it's just a matter of cross-linking to where they should be going, right?

You don't need to throw out your whole information architecture every time and go, it didn't work. I'm doomed. I have to start again. Oftentimes it just comes down to, okay, so if people have gone to that section by accident, I need to have a prominent link on that page that sends them back where they're supposed to be.

Most of the time that will do an answer. So that's really all I wanna say on information architecture. And you can see that there are lots of opportunities for doing testing and actually having that being driven by data rather than politics and opinions and that kind of stuff.

Start with your information architecture before you do anything else. Any prototype any design, anything else is such a good way to begin a project because it helps you focus on and identify those top tasks. So planning your information architecture before you do anything else is really, really beneficial.

And those top tasks can be used to inform your top level navigation through your open card sorted. And then your navigation can be validated and expanded through the closed card sorting and the tree testing.

Learn Straight from the Experts Who Shape the Modern Web

  • In-depth Courses
  • Industry Leading Experts
  • Learning Paths
  • Live Interactive Workshops
Get Unlimited Access Now