Let’s Revisit the Perceived Information Architecture Test (PIA)
It’s been a very long time since I talked about the Perceived Information Architecture Test (PIA). However, some folks are redesigning a web site and they’re looking to use the PIA. Here are some of the questions they asked:
- Have you written anything new about the PIA?
- It’s been 5 years since the article, do you use the test? Still like it?
- How do you word the instructions? What tips can you provide?
- What other insights can you offer?
Here’s my response, pretty much word for word. Minor edits only, to protect the innocent…
– — – — –
We’ve used the PIA test a few times with great success. It’s definitely a “niche” tool, meaning that you only want to use it narrow circumstances. It’s best for web sites that are already designed or prototypes that are high fidelity. The reason is that you want users to absorb a lot of information and then reconstruct the architecture for you. In our opinion, users need to absorb a lot of material in order to form their mental models.
Unfortunately, I have not written any more papers on the PIA. We’ve used it a couple more times as a low fidelity, back pocket tool. It’s great to get people to produce something as a result of their interactions. In any case, I do not have anything else to pass along. I wish that I did.
Regarding instructions, you’re right that the instructions matter. I’m going to stick to what’s in the original article. The main thing is to make people VERY comfortable with the task. We’ve given people some examples in the instructions like:
- “Imagine that you have to create an outline of the web site”
- “Maybe you want to draw some boxes or circles and connect them”
- “You could literally draw some of the key pages and how they’re linked”
We provide these examples to get the juices flowing, but we leave it up to users to create their own PIAs. We do answer questions, but never say “yes” or “no” to the best way to generate the PIA. We do provide the examples in the same order for each user but not much else. It’s very free form. It’s a very creative task.
The best way to analyze the data is to lay out the drawings, diagrams, and related artifacts. Look at what sticks out from artifact to artifact.
Important: Be certain to gather comments along the way as you conduct the sessions. What people say as they create their PIAs is pure gold. We simply look for common words, themes, information groupings, location of information (e.g., upper right button location), and so on. You’re looking for emergent concepts, if that makes sense.
I still like the PIA Test. As I said, I use it more informally now than I did 5 years ago. It’s a wonderful back pocket test.
So, I do still recommend the test, for sure. It’s an easy test to run and you can get solid additional information for next to nothing. And, it’s fast.
Caveat: Don’t depend on only this test. But, use it to supplement other UX and IA work you’re doing.
I hope this helps. If you have more questions, please let me know. I’m also available for additional consultation if you’d like more formal assistance.
– — – — –
That’s all for now. Is the PIA right for you?