Card sorts are a useful tool in determining how information should be sorted on a website. Basically, a card sort allows users to group website content in ways that make sense to them. By conducting and analyzing a few rounds of card sorts, you can uncover a clear and useful information architecture.
Before you are ready to conduct a card sort, you will need to determine a few things:
Open or Closed Sort?
First off, will your test be an open card sort or a closed card sort? If you are starting a site from scratch, an open card sort is your best bet. It allows users to determine both the content groupings that make sense as well as the number of groups you will need to accommodate proper groupings. If you are trying to add content to an existing site or you already know what your groups will be, do a closed card sort.
Create a Content Inventory
If you are redesigning an existing site, you will need to create a content inventory. Once you have an extensive inventory, be sure to add things that you may want to add to the site in the future. If you are designing a new site, it is still a good idea to create a content inventory. Now you are ready to make cards.
Making the Cards
Users can’t comfortably process more than 100 cards during a card sort. However, many websites have more than 100 pieces of content. If that is true for your site, you’ll have to go through your inventory and pick out the most important pieces of content and the ones that have no clear grouping. If there are clear groupings, make them and include one card that represents that grouping.
Once you’ve determined the content that will be on the cards, you can make them. Each card should have a main title, a number (this can be small as it’s purpose is to help you enter the data after the test), and an optional short description if the title of the card isn’t clear.
You will need one full set of cards for each card sort you conduct. You can create a card template on your computer and get printable index cards. This way, you can print all the sets of cards on any laser printer.
Before your final printing, I recommend conducting the card sort yourself and asking some colleagues or peers do the same. This will help you ensure that the card titles and descriptions are clear.
It’s also good to know that there is card sort software that allows you to conduct in-house and mobile card sorts. So far, none of these products are strong enough to replace physical cards.
How Many Participants?
A card sort can be conducted by one to three participants at a time. With two or three participants, you can listen to how the users discuss and debate their choices. This gives you some insight when you’re analyzing the cards later. Three participants are often better than two because it is easier for three to make a majority decision. If you want more aggregate data, you can have individual participants conduct the test at the same time (each with their own set of cards).
I recommend conducting at least 4 tests with input from a minimum of 12 participants.
Also, be sure to follow tips on recruiting test participants.
Set up the Card Sort Area
Your testing area should include a large table or other open flat surface on which the participants can spread out their cards. Only one other person, the moderator/observer, should be in the space to answer questions. You should also have paper clips, rubber bands, blank index cards, and pens available.
If possible, it is a good idea to record the card sort. You can set up a video camera or webcam on a tripod to capture the participants and the table. If other people will be observing, you can hook the video camera up to a television or use software like Morae.
Conduct the Card Sort
Once the participants are comfortable, read them instructions before giving them the cards. It is helpful to write up a set of instructions beforehand to ensure you hit all the points you need to make. Be clear that participants should group things together in ways that make sense to them. If a card title doesn’t make sense to them, they should cross it out and write a name that does make sense. If they feel that a card is missing, they should create a card and insert it into the sort.
During the sort, you should take notes and observe. You shouldn’t be talking much except to answer questions about clarifying what a card means or to ensure that all test participants are having equal say. Card sorts can be frustrating for participants if they have to work as a team or if they have a lot of cards to sort. It is nice to give participants a break in the middle to refresh.
Analyze the Results
Once all the card sorts are done, you can conduct cluster analysis to determine emerging patterns and groupings. Cluster analysis is somewhat complicated - fortunately, there is software that can make it easier. The simplest software I have found thus far is IBM’s EZSort. You can also analyze card sort results with a spreadsheet template. While you are entering the results of each sort into a computer, be sure to keep track of cards that participants added or names that they changed. The software cannot account for these adjustments and they may help you in your final groupings.