Card Sorting: An (Actually Really Fun) User Research Method
Admittedly, card sorting is a dull set of words—it sounds like an activity akin to organizing your sock drawer—but I’m here to tell you that it’s one of the most illuminating (and fun!) user research methods for any product team to have in their toolkit.
Not familiar? No worries. In this article, we’ll walk you through what’s so great about card sorting, the various types of usability testing available, and exactly how to do it.
What Is Card Sorting and Why Is It So Useful?
Whether you have an app with tools or a website with information and products, and whether or not you like fancy product terms, you have something called information architecture. Information architecture, simply put, is how you organize and present information and features in your product.
Information architecture is a key part of the user experience, and card sorting is how you check its effectiveness
Let’s say that you have an e-commerce website selling sustainable clothing. In addition to your search bar, you’ve likely categorized all of your clothing items so that visitors to your site can find what they’re looking for or browse things that are relevant to them. Maybe you’ve organized clothes by gender and age, or perhaps by clothing type—maybe both, in a hierarchy.
Whatever you’ve done, your overall goal is to organize and present your items in a way that is intuitive for your users, since the easier it is for them to browse or find a specific item, the more likely they are to end their session with credit cards in hand.
Card sorting helps you check your assumptions about whether or not everything is grouped in a way that’s intuitive to your users, and it’s easy to see how that leads you closer and closer to your key KPIs.
How exactly does card sorting work?
As with most UX research methods, there are a variety of ways to execute a card sorting session, but all of them involve giving users various words and phrases that represent information within your product and asking users to group them according to their own sensibilities, and usually to name the category.
For example, imagine that your team has a photo editing app. It’s important that your users can find the precise editing tools that they need, and you want to check that your toolbar and sub-toolbars are organized in an ideal way. In each card sorting session, you’ll give users cards, each with the name of an editing tool. They’ll group them however it makes sense to them and name each category.
After doing this with multiple users and analyzing the results, you’ll be able to check user groupings against how you’ve organized the features in the app and make adjustments so that the information architecture—that is, the grouping of photo editing tools—is intuitive for users and overall a good user experience.
What Are the Different Types of Card Sorting?
The intention and basic actions of each type of card sorting are more or less the same, but it’s best to browse the different types below and make a decision with your team as to what best suits your exact needs.
Closed card sort
A closed card sort is when users are given a predetermined group of category names and asked to put various words in the appropriate category. For example, if you have a website with copious amounts of recipes and you organize them by meal type (breakfast, lunch, dinner, dessert), you’d give your users the names of meals (lemon squares, chickpea tahini salad, etc.) and ask them to put them in the place that makes most sense to them.
A closed card sort is most useful when you are totally sure about your high-level categories, perhaps because in your field, there is already a dominate convention for how things are organized.
The downside of a closed card sort is really only that if you’re not 100% sure about your high-level categories, you miss the opportunity to get user generated insights and ideas about how you might categorize things differently.
Open card sort
An open card sorting session assumes nothing about categorization: users are asked to group items or tools and then name the categories on their own. For example, if your team has a SaaS tool for accessing basic internal marketing data, you’ll give card sorting participants cards with metrics on each of them. Users will group each of the metrics as it makes sense to them, and then name the categories themselves.
What’s useful about an open sort is that you’re totally open to surprises. Not only may users group the cards in a way that you hadn’t thought of, but how they conceptualize and summarize each group may really enable you to think about your information architecture in a completely new way.
Hybrid card sort
If you’re uncertain as to whether or not open or closed card sorting makes sense, rest assured that there is somewhat of a happy medium: the hybrid card sort. In a hybrid card sort, you give users the high-level group names as in a closed card sort, but you instruct the users that they can create different category names if they don’t find a good fit.
This is primarily useful when you’re 99% confident in your high-level categorization, but you want to double and triple check and be somewhat open to new categories. What you’ll want to keep in mind, though, is that by virtue of giving over category names to the user to begin with, you’re introducing a bias toward those names. You won’t know for sure whether the user would have thought of them on their own, so again, it’s really only the best method if you’re pretty sure. If that makes you uncomfortable, it’s likely that an open card sort is the best card sorting method for you.
Online or remote card sorting
Card sorting isn’t a new research method, but like with most things, a digital version has made it to the mainstream. There are a ton of card sorting and other usability testing tools out there that you can use in order to do card sorting sessions with users via Zoom, or even an unmoderated session.
These sessions tend to be moderated or done live with you on the call as the facilitator while the user sorts each digital card so that you have an opportunity to ask the user questions as they go. For this, you can create cards using a Miro template, or use a specific virtual card sorting tool like UXMetrics. Overall, the functionality of these tools tends to be quite similar, so don’t agonize over the choice.
That said, live card sorting sessions are time-consuming, especially if you consider the logistics of scheduling with users. If you prefer efficiency over the benefits of being on the call with each user, you can go the unmoderated online card sorting route, using a tool like UserTesting. When you do unmoderated card sorting, you will receive recordings of users doing their session and can watch and analyze them on your own time and at your own pace.
In-person card sorting
Call me old-school, call me sentimental—but the truth is that, as a user researcher, there isn’t much that I enjoy more than a good, face-to-face card sorting session. First of all, who among us doesn’t love index cards? And beyond that, the interaction with the user during your session, and your ability to notice everything from furrowed eyebrows to general hesitation truly deepens what you can learn from card sorting.
The obvious downside of doing card sorting face-to-face is that it’s time-consuming, and there is the hassle of scheduling users and bringing them to a specific physical space. But if you can manage the logistics, you’ll reap the rewards.
When you’re in-person, as we mentioned before, you can notice the subtleties of body language. For example, if a user hesitates or looks like they’re having an internal debate about how to categorize a specific card, you’re more likely to notice in the moment and can ask them to externalize their thoughts. When categorizing a specific card is tricky, their internal debate can be illuminating. Going back to our recipe example, a user looking at a fancy salad recipe who is hesitating may say something like: “Well, on one hand, this feels fancy like dinner. But on the other hand, it could really elevate a lunch that’s meant to impress, like when I’m hosting guests.” That statement alone, if repeated by multiple users, may inspire you to do something like experiment with a special category with recipes designed to impress while hosting guests!
Additionally, while it's not a form of card sorting, tree testing is a similar form of usability testing that resembles card sorting. The methodology is different, but as our tree testing guide explains, it also seeks to find and resolve issues with a product or website's information architecture.
How to Prepare for a Card Sorting Session
The preparation for card sorting sessions is crucial for success. Make sure that you do all of the following before you’re in front of a user.
Decide definitively on your methodology
Talk it out with your team: will your card sorting be open, closed, or hybrid? Remote or in-person? If remote, moderated or unmoderated? This greatly affects what you need to do to prepare, so you can’t really wing it.
Create your set of cards
Whether you’re using physical cards or going the remote route, you’ll need to make sure that you're including all of the cards that represent the ‘items’ in your product so that nothing is missing. Once you’ve created your initial set of cards, ask a colleague to go over them to make sure that there isn’t anything you’ve forgotten. Often, I find that my colleagues in UX design or data analytics remember areas of our products that I overlook.
Write out an introduction and instructions for the user
Whether you’ll be remote or face-to-face, it’s important to explain to users what the session is all about and what they need to do. Even when you’re face to face, you want to make sure that all users get uniform and clear instructions.
Tip: make sure that you communicate to your participants that this exercise is not a test as to what they know, but rather a way for you to understand users’ mental models. In other words, you want them to group the cards in the way that is most intuitive to them. Emphasizing that there are no right answers is key in terms of getting the right insights from card sorting.
Schedule your sessions ahead of time
Now it’s time to find your users, whether you’re using an online platform or bringing users somewhere face-to-face. As with all user research sessions, let users know what to expect in terms of time and assure them that they don’t need to do anything to prepare in advance.
Pro Tips: How to Get the Most Out of Card Sorting
Remember: the magic of card sorting lies in the why
This is worth mentioning more than once. Though the final grouping of cards according to category labels is super important, the thought processes that led users to do what they did says just as much, if not more, about users’ mental models. This data is going to help you create actual action items from your research, so be sure to consistently ask users to explain why they make certain choices or why they’re unsure.
Blank cards are your friend
Though it’s not always done, a powerful addition to all of these research methods is to give users blank cards for anything that they feel is missing. So, with our photo editing app example, if users realize that an editing tool that they often use isn’t in your set of cards, they can add it. Not only will you learn about how it’s grouped, but you’ll also potentially get some insights about missing features.
Put some thought into user groups and sample size
Depending on your product and the market within which it operates, you may want to conduct card sorting sessions with more than one user group so that you can compare and contrast. For example, if your target audience includes both professional content creators and amateur social media enthusiasts, it could be that they conceptualize the relevant information architecture differently, and in order to find a good overall solution, you’ll need adequate representation from both.
As for your sample size, or the number of participants that you conduct card sorting sessions with, as with many qualitative methods - it’s complicated. It will depend on the extent to which you’re relying on card sorting results to inform your product iterations and more. We recommend using this guide to sample sizes as a starting point.
Consider conducting mini-user interviews in your sorting sessions
If you’re sitting with a user or doing moderated remote sessions, you have an opportunity also to ask your participants some interview questions. This helps with card sorting itself because it gives you context: who is this person, and how does it influence their mental models around our information architecture?
Beyond that, it’s also a great way to learn more about your target audience in general, in terms of everything from motivations to unmet needs. For more information about how to conduct user interviews, check out this guide.
Analysis and follow-up are key
Once you’ve collected all of your data, you’ll need to analyze it and create action items. It’s a lot of information, so it’s important not to neglect this part of the process and assume that you have enough of a sense from simply being present in the study sessions.
Now, There’s Nothing Left To Do But Get Started!
You’re armed with all of the relevant details—why not get started? To warm up, you can do a practice internal card sorting study just to get the hang of it before you start planning your real card sorting test—or you can jump right into it. Whatever you choose, you’re bound to have some exciting and KPI-bumping insights coming your way sometime soon.
Don’t forget to subscribe to The Product Manager newsletter, which will keep you updated on all things user research methods, and a lot more.