Card sorting
Posted in Usability on: Sunday, July 24, 2005 by Paul Boag
I am currently involved in some usability consultancy for an intranet that is going through a major redesign. One of the tools we will use to decide on the sites new information architecture is card sorting.
At its core card sorting is probably one of the simplest, and yet most powerful ways of improving a site’s information architecture. It is valuable because it gives an insight into the users mental approach demonstrating how they sort and structure information within their own heads.
How to do card sorting
All you need is a stack of index cards, some post-it notes and a pen. Not exactly a hi-tech approach!
Simply label 20-30 index cards with headings from the various sections, subsections and pages of your website. Depending on the complexity of the site, you might also wish to include a brief description. It is also useful to number the cards so that you can more easily analyse the results of your test later.
Untitled cards
It is also possible to start with no predefined headings but rather allow the user to specify their own section titles. Although this approach initially sounds good because it introduces no bias from the tester, the reality is that it can often be incredibly challenging to the user and so progress is often slow. Often it is better to have existing titles but encourage the user to comment on or change titles if they perceive it as appropriate.
Obviously, 20-30 cards will probably involve some considerable editing on your part but more cards than this can overwhelm the user you are testing. If it is necessary to cover more ground than this, it is possible to have some organisation already in place so that users are responding to an existing information architecture rather than starting from scratch.
Testing normally involves approximately 15 users and requires them to sort the stack of cards into piles that make sense to them. Often you also ask users to name these piles and this label is written on a post-it note that is then attached to the pile.
Card sorting approaches
Beyond this basic approach there are numerous ways you can structure a session however basically this breaks down into two approaches; quantitative or qualitative.
Quantitative
The quantitative approach uses card sorting as a data collection tool and is largely orientated around producing measurable statistics against which to judge. For example, it might establish that 83% of people placed the "contact us" section under "about us".
Qualitative
Although the quantitative approach is perfectly valid, it is easy to prejudice the results and does not particularly help to understand the users’ reasons for the ordering. Personally, I believe more is to be learnt from the qualitative approach. Qualitative testing is a much more interactive and less observational allowing you as the tester to question the user and dig into some of the specifics of how they organise the deck. The aim is to encourage the user to articulate their thoughts and frustrations so you can understand their underlining approach.
What are your thoughts?
It is pretty obvious from this entry that I see a lot of potential in card sorting, but what is your opinion? Have you tried card sorting? Was it valuable? What works for you and what doesn’t? Do you take a quantitative or qualitative approach? I would love to hear your thought!
5 Comments
Comments are for the discussion of this post. If you have other questions / comments then post them to the forum or send me an email
I have to admit the one time I tried card sorting was far from a success. Judging by your comments above I used far too many cards and simply overwhelmed the users. As a result they struggled to order anything into logical sections.
I’ve used card sorting in the past, and it worked really well for me. I had vaguely decided upon the seven top level headings that I wanted to use, so on each card I wrote either a section heading or a piece of content that I wanted on the site. Then I asked customers to arrange the content under the headings they would expect to find them in. The results were quite interesting. I primarily used them quantitatively, but I did note any comments that the customers made whilst carrying out the exercise, and they informed the final structure of the site as well. I think I was lucky that I was creating a very flat structure. I could imagine that the number of cards would have been overwhelming and the whole exercise would have suffered if I’d tried to create hierarchies. I also don’t think the customers would have been willing to sit there for too long trying to organise the cards!
The thing I wonder about card sorting excercises is how you deal with the stack of cards once you get them back?
Do you use some kind of Excel spreadsheet to note their card placement order? What does that Excel spreadsheet look like?
How do you analyze the results if you have, say, 10 people providing card sorts?
What we don’t do is analyse the cards in a particularly statistical way. We don’t believe in simply adding up the results and going with that information architecture. We work through the groupings of cards and allow those to influence our thinking. Sometimes we put them into an excel document but not that often. Most of the time it’s just a matter of examining the groups and looking for consistencies.
I found a good guide to card sorting here:
http://www.boxesandarrows.com/view/card_sorting_a_definitive_guide
Definitely worth taking a look if you’re unfamiliar with the process.
I’m about to embark on the huge task of organising content for a student support site. Let’s hope this card sorting strategy works!