Using card sorting in UX research for better information architecture

Optimizing information architecture (IA) without user input—whether for your website or another digital interface—is bound to fail. 

You need to understand what your users call things, how they mentally group different concepts, and how they create categories within the more extensive architecture.

Enter card sorting: an exercise designed to extract that exact information from your audience. 

It's an easy way to get into the mind of your audience and build a digital infrastructure they will respond to. 

In this guide, we'll explore the concept of card sorting in-depth: from its benefits to the different types of sorting and examples of successful sorting exercises.

Card sorting is a relatively broad research term describing any activity in which participants group cards with possible terms of features into categories or based on their similarities. 

You can use it to understand your audience, design your website or products, and more. Notecards can be physical or digital (using a card sorting software tool).

Card sorting is commonly used when building the information architecture (IA) for your website (or any other digital presence). Ultimately, you could use it anytime you’re organizing information.

Card sorting is an exercise that provides the background information you need to structure all info you’re presenting into an intuitive, accessible interface for your users.

Card sorting typically gathers three types of info from your users

• Terminology your audience uses when describing the things you're looking to show and categorize

• Categories into which your audience groups the individual items within the IA, including how these categories should be named

• Relationships between the individual items and categories—like how closely related two concepts are, or which order they would naturally group them in

Combined, these three types of feedback allow you to build a more intuitive IA, from creating categories based on audience recommendations, to labeling your site navigation and the relationship between different categories.

What is card sorting in human-computer interaction (HCI)?

Card sorting plays a central role in human-computer interaction (HCI) by designing software systems that are intuitive for users.

For instance, card sorting can reveal how users like items grouped in navigation windows or menus as they move through your site or app.

Card sorting allows researchers to understand how audiences typically interact with that interface and how that interaction impacts the navigation and hierarchy of buttons to be displayed.

The basic concept for card sorting remains identical in HCI. It's just a different research application for a general concept designed to draw out your audience's thoughts on what to name various items in an informational hierarchy, how to sort them, and how they relate.

Is card sorting a form of research?

Card sorting is a type of user experience (UX) research that seeks to understand your audience's domain knowledge and expectations related to the concepts you're asking them to sort. As such, it's an open-ended research tool that depends on qualitative audience input.

The most significant advantage of card sorting is its ability to garner a comprehensive understanding of your audience's thought process when naming and categorizing different pieces of information they encounter. That, in turn, helps you better organize the information on your product or website. 

Card sorting has considerable benefits for UX researchers:

• Quick, cheap, and easy insights. As long as you can get a group of users together, participation is quick and digital tools can help to simplify and synthesize insights.

• Audience empathy. As a research tool, card sorting allows you to stop relying on hunches when categorizing content and information. Instead, you'll see the information from your audience's perspective, placing their insights and needs front and center.

• Categorization and name ideas. When working with a plethora of information (like multiple products or content types), card sorting helps you find the correct categories to put them in. Naming suggestions allow you to use category labels that reflect your audience’s natural vocabulary instead of imposing your internal product jargon on them.

• Improve your IA. Improving the intuitiveness of information also works vertically within the hierarchy. With the proper card-sorting exercise, you can understand how content and categories should be structured and related when presented online.

Why do we use card sorting?

Card sorting is a popular research tool in part because it requires no significant investment or high-tech tools. Instead, you can gain quick and valuable insights about your IA and user expectations of categories and category labels.

Card sorting is beneficial because you can apply it to such a wide range of situations, including:

• Organizing e-commerce products online

• Categorizing an extensive FAQs section

• Creating a simple sitemap for your website

• Building an intuitive IA for a smart device

• Prioritizing features based on importance, favorability, or other descriptive scales 

You can use card sorting at any stage of the design process—to develop the initial structure of a website or app or to refine an existing navigation structure. You can also use card sorting to test the effectiveness of existing navigation structures. 

While the concept is relatively simple, researchers can use three basic types of card sorting to gain user insights:

1. Open card sorting

2. Closed card sorting

3. Hybrid card sorting

None of these methods is inherently better than the others. Instead, the best option depends on your situation, use case, and audience.

1. Open card sorting

In an open card sorting approach, the researcher presents users with information cards but no categories. Users then create categories for these cards, labeling them according to their preferences and expectations.

Open card sorting is the most flexible of the three options. It places the fewest limits or restrictions on users. Users can add cards to the existing inventory or change the labels in some open card approaches.

This type of sorting is a great option when idea generation is a core need at the beginning of a new IA project. That said, it can be limited in generating specific ideas or reaching any consensus among the feedback.

2. Closed card sorting

A closed card sorting approach presents users with the names on the cards and the individual categories into which to place them.

It constricts user freedom in the categorization process, making it useful primarily for understanding how your existing IA meshes with user needs and expectations.

Closed card sorting is more fixed and can lead to faster, more easily interpreted results. But its restrictions on your audience also mean that it can miss improvement opportunities where your audience would have created different categories or named those categories differently had they been given the freedom.

Still, it's a great way to follow up on an open card sort to ensure you have the proper categories in place.

3. Hybrid card sorting

A hybrid approach combines the best of both open and closed card sorting. Participants receive pre-defined categories for their cards but also have the freedom to create new categories if needed.

Hybrid card sorting can help you test your existing architecture and uncover potential holes you've missed. It can also help users create categories for complicated information structures, allowing them to focus their brain power on the non-obvious categories that don’t yet exist.

Finally, you can use it to remove the dreaded miscellaneous category in an IA— pieces that seemingly do not fit any individual category.

These eight steps can help you run a successful card-sorting exercise:

1. Choose a research goal. Choosing a goal involves understanding precisely what you want to take away from the card sort. For example, are you looking to review your existing IA? Or create a new structure?

2. Document your expectations. Write down what you expect the results of the card sorting to be. This step helps you stay grounded as you work to confirm or contradict the status quo.

3. Determine your card sorting type. An open, closed, or hybrid approach may work best, depending on your research goal. Choose your type early to help structure your study.

4. Prepare the exercise. Create the cards you're looking to organize into categories. For a closed or hybrid approach, identify and write down the categories into which participants will sort the cards.

5. Launch the card sorting study. Work with at least 10-15 users for relatively representative results. Set a time limit between five and 30 minutes on the study, depending on the number of cards and categories, to prevent potential procrastination.

6. Debrief your participants. After the study, ask your participants whether the card amount was right, whether your categories were well-named, and what else you could improve for future repeats.

7. Analyze the data. Look for frequently paired items, category names that appear multiple times, and other common trends. Look for patterns you can use to extrapolate insights.

8. Repeat the card sort as needed. You'll rarely get conclusive insights on your first card sort. Take your learnings from the exercise and debrief to run another exercise with new participants, and repeat as needed.

Formats of card sorting

Beyond these basic steps, you can format your card sort as a moderated or unmoderated session.

• Moderated card sorting In this format, the researcher actively moderates and asks users to narrate and rationalize their choices to gain insights into the reasoning behind their decisions.

• Unmoderated card sorting In this type of session, the researcher doesn't interact with study participants. Instead, users do the sorting themselves, and the researcher only evaluates the outcome of their actions. Unmoderated card sorting is fast and quick to do electronically. But its emphasis on the final sort over the rationale behind it can also limit the insights gained about user expectations and needs.

Non-traditional card sorting methods

In addition to traditional paper-based card sorting, you can also use a digital technique to get similar insights.  

Digital card sorting uses online tools and software to create a similar drag-and-drop experience to its physical equivalent. That increases flexibility and speed by not requiring your users to be in the same place as you. At the same time, the integrated analytics capabilities of these tools can also make analyzing the outcome easier.

However, digital card sorting includes fewer opportunities for the researcher and participants to interact or ask clarifying questions. Paper-based card sorting is thus a better choice for more complex structuring with many cards and categories that may require clarification.

How to effectively conduct card sorting technique

An effective card sort depends not necessarily on one perfect type or technique but on optimizing for the individual situation. An intimate understanding of your users and the research question to answer can help you build a compelling study designed to gather tangible insights.

For example, testing the IA of your e-commerce store requires talking to a distributed audience about a relatively straightforward set of topics and categories.

In this case, an unmoderated, closed, or hybrid digital study might be sufficient for initial insights. On the other hand, organizing an extensive knowledge base for your software solution may require a moderated, open approach to finding suitable categories and structures.

In-person, paper-based card sorting is simple. All you need are enough index cards for both the card and the categories and writing utensils for every participant.

A whiteboard can help to put the results on the wall for everyone to see.

Digital card sorting requires software designed for this type of research. These tools can help:

• OptimalSort offers easy setup and collaboration features, as well as automated analysis of the results. A basic, browser-based version is free, while more complex plans start at $166/month.

• UserZoom offers open and closed card sorting and analytics ranging from completion time to category similarities across participants. Pricing starts at $250/month.

• Miro is an online : tool with features that work for basic card sorting. It's a broader playground that offers no post-study analytics, but paid plans start at only $10/month per user.

• UXMetrics allows researchers to run open, closed, hybrid moderated, and unmoderated card sorts. It shines through simplicity and has a free version for unlimited card sorts. 

• Maze is a comprehensive UX research platform that counts card sorting among its features. The browser-based tool is free for individual projects and starts at $25/month for ongoing research needs.

Even tools with paid plans typically offer at least some free version or trial.

Before committing to a paid plan, test their advantages and disadvantages and their integration into your other research tools.

A few simple tricks can help you get the most out of any card sorting exercise, regardless of type or format:

• Keep it simple. Participants should be able to understand the assignment easily, which calls for clear task descriptions, categories, and instructions.

• Provide context. Let your participants know precisely why they're sorting the cards to get them in the right mindset for the exercise.

• Avoid leading questions. As with any UX research, all instructions and participant interactions should be as unbiased as possible to get an accurate view of your users' perspectives.

• Offer multiple types of sorting. Participants who can sort the cards by type, frequency, or user task can provide richer insights into their thought processes.

• Use visual cues. For example, color-coded categories can make it easier for participants to sort cards in the right spots quickly.

• Record the session. Video or audio recordings can help you, and other stakeholders review a previous session for further insights or clarification.

• Keep the categories and cards manageable. Too many topics or cards can overwhelm users and provide fewer insights. Try to stay below 40 cards and 10 categories.

Card sorting shines as a research tool because it's fast, reliable, and cheap Card sorting can be an excellent first step to building new IA or ensuring that your existing IA aligns with user expectations.

Of course, it's not a solve-all solution to UX research.

You should use it in conjunction with other tools and techniques.

When well-integrated and implemented, sorting can become a core technique for any researcher looking to expand and improve their understanding of users.