Learn how to run a card sorting session in order to better understand your users’ mental models in this tutorial with Heather O’Neill, CEO of Pixels for Humans.
In this tutorial you’ll learn what a card sort is and how to conduct a successful card sorting session.
Card sorting involves creating a list of terms or phrases and then asking people to organize them into groups. Card sorts can be very useful in building something from the point of view of your user, rather than making assumptions.
It’s common to perform different types of card sorts at different stages of a project. Let’s look at two different types:
- An in-person, moderated card sort.
- An online, unmoderated card sort.
Imagine you’re part of a team helping a hospital redesign their website. For the initial session you’re the moderator in the same room as a single participant. You explain the project and then give them a number of content items written on cards such as:
- Visiting hours
- Get directions to the hospital
- Phone numbers
- Pay a bill
- And about 20 other items
Next, you ask your participant to sort the cards into categories in a way that makes sense to them.
This leads us to Big Idea, Number 1: To get the most in-depth feedback, use an in-person, moderated card sort.
Whenever you’re a moderator, you should encourage your participants to think out loud during the process. When they’re done sorting, you can follow up on things you observed.
Here’s an example of how that can lead to potentially interesting data:
[Beginning of example]
[Visuals: A number of cards sorted into piles. Will be a still image and we’ll highlight the two cards being discussed.)
Moderator: I noticed when you found those two cards you put them together immediately and they never moved. Do you mind reading those two aloud?
Participant: Yeah, those are the “Patient Billing” and “Insurance Coverage” cards.
Moderator: Do you think there’s another pile they could be grouped into? Or could any of the other cards be added to those two?
Participant: Hmm, I don’t know, I feel pretty strongly these two would always belong together in their own group.
Moderator: Ok, thanks.
[End of example]
This strong opinion is interesting, but it might be unique to this participant. Take note and move on.
For this next step we’re using an open card sort which asks the participant to label their groups of cards when they’re finished.
Closed card sort asks participants to put their groups into labels you provide them.
Each method has its pros and cons which we don’t have time to explore here, but if you want to learn more, there’s a link to a useful article in the Resources section for this video.
Big Idea, Number 2 is true no matter which method you use: Capturing the process is just as important as the card sorting results.
Here, the moderator uncovers something that could only have come up in conversation.
Moderator: Were any group names particularly difficult to create?
Participant: Yeah, this one here, which I ended up calling “Services”.
Moderator: Why was it difficult?
Participant: It was difficult because the original pile was really easy to make. All of these items like: Orthopedics, Cardiology, Radiology, and so on are pretty similar and obvious. So I thought the name of the group would also be obvious, but I struggled between choosing “Treatments” and “Services” and I just decided “Services” was fine…
You know… weird, the phrase “Medical Services” just popped into my head. Is there still time to change my mind?
The moderator would never have gotten the insight if they were just looking at the final results and not engaging with the participant.
This leads us to our Best Practice: Repeat your card sorts multiple times in order to obtain relevant data.
Somewhere between 20 and 30 participants are needed in order to get valid data and to observe legitimate patterns.
Not everyone has the time and budget to schedule this many in-person card sorts. Luckily, there are other options, which leads us to Big Idea, Number 3: Online unmoderated testing allows you to validate your designs at scale.
Thanks to online tools, you can open up your tests to broad audiences who can do card sorting remotely. It’s typically easier to recruit participants this way, since they can do the card sort at their own convenience.
This is an unmoderated card sort using an online tool called Optimal Sort. In unmoderated sorts, participants do the card sort on their own and you are sent the results. The benefit here is you will receive more responses, since you don’t have to be available for each test.
In this example, you note that 42 people agreed that the items “Patient Billing” and “Insurance Coverage” should be put under the “Billing and Records” group and only 8 people agreed that they should be put under the “Patient and Visitor Information” group.