I suggest that you start by doing card sorts one on one with your participant. In other words, you sit and watch as one participant completes the sort. This allows you to really focus on the groupings they create and listen to their justifications. Once you've run and analyzed a few sessions, this way you might decide to do sorts with a group of two or three participants working together at the same time. A group will probably talk more between themselves, which can be useful information, but can also be too hard to follow when you're just starting out.
Before you even start with participant sessions, do a run through with a friend or colleague to check for errors, typos, or misconceptions that they have. I can't emphasize this enough. There will be mistakes or confusion, and it's better to catch this with a friendly helper than to have to throw out real user data. During a run through of the whole process, we'll also make you more confident when you're working with real participants. As the session moderator, your job is to make sure that participants feel comfortable and understand what they're supposed to be doing, and then just sit and watch quietly.
I like it to read my introduction and instructions from a moderator's script, so that I say the same thing to each participant. Then I hand them the stack of cards and let them get on with it. Getting participants to think out loud will you give a wealth of additional information. When participants have doubts about where to place an item their comments and the vocabularly they use can help you to work out what alternative arrangements might be feasible. Similarly when they're very sure about an item, it can give you an increased confidence about its placement. Have a notepad handy to write down what each participant says, because the terminology they use and the reasons they give can be very valuable to you as you do your analysis and create your information architecture.
My usability testing course gives you the full details on moderating a session using the Think Out Loud Protocol. And we've also included a sample card sort moderator script in the exercise files for this course. I start by getting the participants to spread all the cards out on the table. That lets them get familiar with the types of information the cards contain, which in turn, means they can create better groups. Then, they start grouping the cards. Once the participant has placed all the cards into groups, check for groups that contain too many cards, perhaps ten or more.
And ask the participant to break that stack of cards into smaller groups. We do this because otherwise, the groups they create won't have sufficient differentiation. You might end up with a group called products, when you really need to know how people distinguish between your different product categories. It's also good to ask participants to see if they can combine groups of just two cards with other groupings. Sometimes that just won't be possible, but such small groups really don't help us with analysis.
If participants feel really strongly that a certain card doesn't group well with any other cards that's okay. However find out in detail why they feel that way then you'll know whether it was because of the set of cards you used. For instance not having any other similar cards or whether it was because of their mental concepts of the site. Now, it's time for participants to tell you what they think each group should be called. The best way to do this is normally to give the participants some blank index cards and a pen. Then, they can write each group name down on a card, and place the card on top of each group.
Getting participants to physically write their names down and place the cards, means that they consider all the groupings in combination. That way, the names they've given are more likely to be unique and descriptive. I've even had participants who choose to re sort some cards after they've tried unsuccessfully to give names to the groups they've created. That's just fine. At the end of the session you may have some questions about the groupings that a participant chose, I'd suggest the following 3 questions as a good starting point to get the information you need.
The first is, what is the overall pattern here? Asking this question, let's us understand the stretcher they used as a basis for the groups. Next, I'd ask, were there any groups difficult to create? This suggests items that may not fit well together. If there are any, then ask them why those were difficult to create. Then I'd ask were any cards difficult to put into a group, and if so why.
This suggests items that might not be structurally related to each other. Be careful asking any other types of questions. Remember asking leading questions or ones that suggest that the participant choose the wrong grouping will just make your participant feel bad and won't help you at all. Instead ask mutual questions like: can you tell me more about this grouping or can you explain what you mean by this category name. Finally thank the participant. Give them their gratuity, and answer any questions they may have. After they're gone, record the numbers of all the cards in each group on the card the participant used to label the group.
That way, you have a written record of the card sort. Now, you can gather up all the card sort cards, shuffle them so they're in a random order, and leave them ready for the next participant.
- What is information architecture?
- Why do research?
- Creating and running a paper card sort
- Recruiting test participants
- Analyzing paper card sort results
- Running a computer-based card sort
- Creating abstract information architecture
- Validating your plan with a reverse card sort
- Translating information architecture to navigation and layout
- Watching the server after you go live