I have been asked to evaluate a clients proposed IA based. The IA is only in excel and we have no wireframes or mockups to test.

The most obvious method would be a card sort but due to the sheer number of pages (600+) this will be impractical.

Having said that, and because we are testing different personas, I think a focussed card sort with different sections of the site would be a good compromise.

Has anyone used this technique or do can you recommend any other evaluation methods for assessing the validity of large, complex IAs?

website is here (old IA) www.RSAMD.com

  • Surely each of the 600 pages isn't relevant for a card sort. The goal of a card sort is to figure out general priorities, not index every single available item. Try cutting things down to 50 items and then getting people to prioritise those.
    – Rahul
    Commented Jun 20, 2011 at 21:11
  • it's tricky, the client wants data back on everything so, at times, the research has to demonstrate a fair degree of granularity. But I have cut the focus down to one or two areas I'm having most trouble with and this represents only about 30 items...
    – colmcq
    Commented Jun 21, 2011 at 15:21

1 Answer 1


My favorite answer to most UX questions is: "It depends".

How will the redesign be done? Full redesign with a rewrite of navigation and content? Or keep the existing content as is, and just redesign navigation?

My next question is: What are you trying to organize? 600+ words sounds like a lot of words to organization, and I don't believe that you really need to card-sort all of these words?

Well, here comes my tips:

1) Focus on user tasks rather that information about the organization.
What are the most important tasks for the visitors? Is it to read about the "Postgraduate" or is it to "Find relevant studies" or is it to "Find staff"?
User tasks should be the primary motivation in your IA work.

2) Start with navigation.
The user needs to know how to find stuff, and he needs to know where he is.

3) Do some measurement.
IA can be measured. "Task success" is definitely the best metric to use when you evaluate the navigation. You can also measure the "lostness" by counting the number of steps the user used to find a page, and compare this to the minimum number of steps that is actually needed to find that page.

3) Use other methods.
If you really need to test several hundred words, you should turn the card sort method upside down. Instead of presenting all words at once and then let the user organize them into groups, you could give the user some groups and let them fill the groups with their own key words. (This method probably have a name, but I can't remember at the moment :)

4) Use a tool.
There are several online card sort tools that will help you. Just google "card sort tools" and take a look at some of them.

5) Training.
To ensure a good IA over time, training is important. A couple of days with the web editors (or maybe all contributors) will ensure a more uniform IA to the site.

BTW, Creating a wire-frame isn't that hard. Take a look at www.wirify.com, which I used to create this in seconds:
enter image description here

  • I would say come end of play, you've got the 50 :)
    – colmcq
    Commented Jun 21, 2011 at 10:54
  • @colmcq Don't forget to award the bounty. You can award it at any time, you don't need to wait until the end of its life.
    – ChrisF
    Commented Jun 25, 2011 at 12:30

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.