Card Sorting: Learn More About Your Users and Create Better Products

To design products with excellent UX, you need to understand how your users think and organize information. Yet, when it comes to organizing content in our own products, all too often, we structure information based on what we think is optimal for our users, and not what our users think makes sense to them. Card sorting is a technique that helps product teams discover how their users understand and categorize information. In this article, you’ll learn about the different types of card sorting and 11 best practices for running card sorting sessions.

Information organization is an essential ingredient in creating a good product strategy—to design products with excellent user experience, we need to understand how our users think and organize their research. Yet, when it comes to organizing content in our products, all too often, we structure information based on what we think is optimal for our users and not what our users think makes sense to them. Fortunately, there is a technique that helps product teams discover how their target users understand and categorize information. This technique is called card sorting.

In this article, I will review the concept of card sorting and share some practical advice for running effective card sorting sessions.

What is card sorting?

Card sorting is a research technique used by product teams to design or evaluate the organization of information in a product. When researchers conduct card sorting, they prepare a set of topics, write them on digital or physical cards (each topic on a separate index card), invite people who represent their target audience, and allow them to organize topics into categories that make sense to them. This method uncovers the user's mental model, and the insights are used to organize information in a way that matches the user's expectations. 

Types of card sorting

There are multiple different types of card sorting, depending on your different needs.

Please note that while I will discuss online and offline (in-person) card sorting, you should only engage in offline card sorting sessions once your governing body has said that it is safe to do so with respect to physical distancing rules for COVID-19.

Open, closed, and hybrid card sorting

Depending on the needs of your project, you can use open or closed card sorting. When you conduct open card sorting, you ask participants to organize topics into groups that make sense to them and then name on each group. Open card sorting is equivalent to open-ended questions in a user interview—participants don’t have any restrictions for naming. This type of sorting is especially useful when you want to design a new product. When you analyze card sorting results, you will see what parts of your product structure (content groups, navigation) match up with how people would organize the same information.

Closed card sorting, on the other hand, requires participants to organize content into predefined categories. Closed card sorting requires less effort from test participants because all they have to do is match content to existing categories. Since closed card sorting is more like an evaluative exercise, it works best for prioritizing and ranking features. Researchers often use this type of sorting when they want to improve the usability of an existing product. 

For example, a team can apply it to features available in a product. Study participants can distribute all available features of a product between groups called "Essential," "Frequently used," or "Optional."

In some cases, it’s possible to combine two approaches. The mixed approach is called hybrid card sorting. For example, a team can have a list of predefined categories and a few empty slots that participants can name. Hybrid card sorting works best when you know that you have some missing categories in your current category structure and want to fill this gap.

You might also like: How to Learn More About Your Users with a Contextual Inquiry.

One-on-one or group card sorting

Card sorting may be performed individually (as a one-on-one in-person session with an observer) or in groups. The major benefit of individual sessions is flexibility. Scheduling individuals can be easier than scheduling groups. Also, you can ask study participants to think out loud, debrief them, or ask clarifying questions without worrying about creating uncomfortable conditions for other study participants. 

Alternatively, one benefit of group sorts is that they typically provide less biased results. Participants in a group can work together and come up with better solutions on how to sort cards into groups. 

Moderated and unmoderated card sorting

The key difference between these two types of sessions is an impartial observer who is present during the session. Generally, moderated card sorting can provide more valuable insights because the moderator can ask participants some clarifying questions and learn more about the rationale behind individual decisions. Questions like “Why did you decide to group these cards?” can help to understand the underlying logic behind individual choices. However, moderated card sorting typically requires more resources and preparation and can be expensive. 

"Generally, moderated card sorting can provide more valuable insights because the moderator can ask participants some clarifying questions and learn more about the rationale behind individual decisions."

The alternative to moderated card sorting is unmoderated card sorting. This type of card sorting is conducted online, using tools like Userzoom or Optimal Workshop. Unmoderated card sorting is very helpful as a supplement to moderated card sorting sessions. A product team can run a moderated card sort with a limited group of participants in an attempt to define the top-level categories. Following this, the team can run another, unmoderated card sort with the larger group using the categories defined in the first session to see how well this structure works for the larger audience.

Online and offline card sorting

Card sorting can also be done either face-to-face and offline (again, only if your governing body has said that it is safe to do so with respect to physical distancing rules for COVID-19), or remotely (online). In the first case, researchers invite representatives into the same space (such as a conference room) and allow them to interact with actual, physical cards that contain topics. The topics are written on pieces of paper or cards, which are spread out on a table. The most significant advantage of offline card sorting is a low learning curve for the participants. Basically, all researchers have to do is to provide clear instructions to study participants. Also, the fact that the moderator is in the same room with participants means the moderator can guide them through the process.

When you conduct online card sorting, the research team invites participants to join an online session in card sorting tools like Optimal Workshop or Userzoom.

Closed card sorting example in UserZoom. Image by UserZoom.

Online tools are beneficial for statistical analysis—they can create dendrograms (diagrams that show the grouping of information within hierarchies) that will visualize how topics are related.

UserZoom can generate visual diagram data in the form of a dendrogram. Image by UserZoom.

The major disadvantage of onlinecard sorting tools is the fact that you lose the personal touch—online sessions cannot capture the body language of participants, and it becomes harder to understand the rationale behind individual decisions. 

Best practices for card sorting

To have a successful card sorting session, there are some guidelines to follow.

1. Recruit enough participants

You’ll need enough participants to detect patterns in users’ mental models. The number of participants you need depends on the format of the card sorting. NNGroup recommends hiring 15 participants to get statistically useful results. If you’re running a group card sort, you can have five groups of three participants per group (total, 15 participants).

Ensure that every person you hire matches your user profile. If you don’t have much information about your target user, it’s recommended to invest in user research before starting card sorting. 

You might also like: Conducting User Interviews: How to do it Right.

2. Don’t go overboard with the number of topics

The first step in conducting a card sort is to determine the list of topics. Topics can be individual pages, navigation labels, or functionality of your product. When conducting card sorting, it’s important not to overwhelm participants with too much information. Always consider the cognitive load on the participant—remember that the more content you have, the longer it will take for participants to group it, and the less motivated they will be. 

"When conducting card sorting, it’s important not to overwhelm participants with too much information."

With the goal of “quality over quantity,” aim for something in between 30 and 60 topics per set.

Use the following strategies to come up with actual topics:

• For existing products: conduct content inventory, identify the most important or most frequently used content, and create the list based on this information. You can use Google Analytics to find out the most visited pages and use these as a starting point.
• For new products: brainstorm information you want to include on your website and prioritize it according to the needs of your target audience and business. 

3. Ensure that cards are “sortable”

Ensure that the cards you provide to study participants can be logically grouped. Selected topics should have enough similarity to allow groupings. At the time when you have a list of topics, conduct a trial run of card sorting with someone from your team just to ensure that the list is good enough. 

4. For offline card sorting, use properly-sized cards on thick paper with the same styling

Cards are the most used element during card sorting sessions. Donna Spencer, the author of the book Card Sorting: Designing Usable Categories, recommends using 3″ x 5″ (10cm x 15cm) cards because the text on such cards is easy to see from a distance.

Cards created from thin paper such as stick sheets can easily be torn. Cards that use thick paper are more durable. 

Last but not least, keep your cards uniform. Use the same visual style for all cards—same color for paper and same font size for text. Any variation in visual style can bias sorting results because cards that look different will immediately capture the attention of your study participants.

You might also like: 22 Basic UX Laws That Every Designer Should Know.

5. Be careful with text on cards

The text you put on the cards can have an impact on card sorting session results. Ideally, the labels should be short enough so the participants can read them quickly, but long enough to convey the meaning. 

The words/phrases you write on cards should be self-explanatory.

Jakob Nielsen advises against using topics that contain the same words because participants tend to group those cards. He illustrates this problem using the following sets of cards: 

Avoid using sets with synonyms (like this one). Example by NNGroup.

Given the set A, most participants will sort all “strawberry” cards together and all “wheat” cards together. Likewise, given the set B, most users will sort cards by activity “planting,” “growing.”

6. Schedule enough time for the session

Before running a card sorting session, you need to estimate how long the card sort will take. Generally, you should plan about one hour for each session. Provide the estimate to participants at the beginning of the session to help them better gauge the required time and effort. 

Take the following into account:

• The time required to describe the instructions to study participants
• The time required to master a tool (if you run online card sorting)
• The time required for debriefing study participants

7. Clearly explain the goal of the session

Explain not only what you ask your participants to do but also why. People work better when they know how the results of their work will be used. For example, if you plan to redesign a website information architecture, you should tell study participants that the insights you gain during the card sorting will help you to create a better navigation experience.

"Explain not only what you ask your participants to do but also why."

8. Ask participants to think out loud while working

It will help you to understand the participant's thoughts and rationale. It’s possible to use this approach both during the moderated and unmoderated sessions. In the latter case, you can simply ask users for permission to record the sound. 

You might also like: Lean UX: A Guide for Remote Teams.

9. Don’t force participants to select a category

If the participant isn’t sure about a particular topic, it’s fine to leave it off to the side. You can even create a special category for such cards called “Uncategorized.” By doing this, you will minimize the risk of users making random choices. 

10. Allow participants to change their mind

The more participants work with the topics, the better ideas they will have on how to structure them. That’s why during card sorting sessions, it's normal for participants to change their minds and move cards from one pile to another. You shouldn't prevent them from doing so. In fact, it’s even recommended to provide clear instruction at the beginning of the session that there is no right or wrong answer.

11. Analyze card sorting results

After you gather research data, you need to invest time in finding patterns. Look for common groups and for items that were frequently paired together. For example, if the goal of your open card sorting was to improve the website navigation system and you noticed that participants created too many different groups, you should try to combine the groups that have similar properties. After that, you can run another card sorting (this time, close card sorting) to validate whether the groups you created work for your users. 

"After you gather research data, you need to invest time in finding patterns."

Finding topics that participants struggle to classify is another thing that you need to consider during the analysis. Such topics might cause findability problems, so it’s important to understand why participants faced this problem of categorization (whether it’s because the card labels are unclear or because the content itself didn’t fit in with other categories).

Understanding how users think

Card sorting is an excellent way to understand how your users think about the structure of your content. This technique will help you understand how to group and label information in a way that makes sense to users. At the same time, card sorting is not a silver bullet; it cannot be used as the only technique during user research. Card sorting works the best when it is used in tandem with other user-centered design techniques such as usability testing and contextual inquiry.