Card Sorting: A Complete Guide to Understanding User Mental Models

What Is Card Sorting?

Card sorting is a generative UX research method that helps you understand how people naturally group and label information. Users sort cards into categories--either ones they define themselves (open card sort), ones you provide (closed card sort), or a combination of both (hybrid card sort)--and their choices reveal how they think about information organization.

The method originated in psychology and has been adapted for UX research to help designers, product managers, and researchers understand user expectations for information architecture. By asking people to sort cards with different topics or items into groups, we can see how they think about information and then organize it in ways that match their mental models. IxDF explains how card sorting reveals user mental models.

Every user brings their own mental model to your website or application--a framework for understanding how information should be organized based on their past experiences, cultural background, and domain knowledge. When your information architecture doesn't align with these mental models, users become frustrated, bounce rates increase, and task completion suffers. Card sorting helps you discover these mental models before you invest time and resources in building the wrong structure. This is a critical first step in our web development process, where understanding user expectations shapes every design decision.

One of card sorting's greatest strengths is its simplicity. Unlike complex research methods requiring specialized equipment or extensive training, card sorting can be conducted with basic materials--index cards, sticky notes, or any digital tool. This accessibility means teams can validate their assumptions quickly and inexpensively, making it an ideal first step in any information architecture project. UXtweak highlights the simplicity advantage of card sorting for teams.

Types of Card Sorting

Open Card Sorting

In an open card sort, participants group cards into categories they define themselves. This approach reveals natural user preferences without bias from predefined categories, making it ideal for early exploration when you need to generate new ideas for categorization and naming conventions. Great Question explains open card sorting methodology.

Open card sorts are best for early design exploration when you need to generate new ideas for categorization and naming conventions. Use cases include initial website structure planning, organizing products in an e-commerce store, categorizing blog articles, and organizing features on a SaaS website. Great Question provides open card sort use cases.

Closed Card Sorting

In a closed card sort, participants are given cards and predefined categories for sorting. Unlike open card sorting, this method provides constraints that help assess the effectiveness of a proposed structure from the users' perspective. IxDF defines closed card sorting.

While closed card sorting doesn't fully reveal how users would naturally categorize information, it does allow you to evaluate whether proposed categories make sense to users. Closed card sorts are best for later validation: validating a proposed information architecture, confirming whether users understand your content categorization, clarifying unclear category names, and comparing multiple sets of predefined categories. Great Question outlines closed card sort use cases.

Hybrid Card Sorting

A hybrid card sort combines elements of both open and closed study types. You give participants predefined categories but also allow them to create new categories or suggest changes where they see fit. Great Question describes hybrid card sorting.

Hybrid card sorts are ideal when you've conducted initial research and have a preliminary structure but want to ensure it aligns with user expectations and are open to adjustments. They're also valuable when stakeholders have certain non-negotiable categories but you still want to gather user-driven insights. Great Question discusses hybrid card sort scenarios.

Tree Testing (Reverse Card Sorting)

While technically a different method, tree testing evaluates the effectiveness of an existing information architecture by asking participants to find items within a hierarchical menu structure. Great Question compares card sorting and tree testing.

Card sorting generates IA ideas and options from user mental models, while tree testing evaluates the effectiveness of an actual IA by assessing findability. Instead of sorting cards into categories, tree test participants navigate a text-based hierarchy and try to locate specific items or complete tasks. This helps identify which parts of the IA work well and which cause confusion or frustration. IxDF explains tree testing methodology.

The real power comes from following up with closed card sorts to test proposed hierarchies and then tree testing to assess goal-based success with a new information architecture. As UX expert Thomas Stokes notes, combining these methods provides comprehensive validation of your information architecture decisions. Great Question shares expert insight on combining methods.

By combining card sorting with usability testing and tree testing, you create a complete research cycle that validates your information architecture from multiple angles.

How to Conduct a Card Sort: Step-by-Step

Step 1: Define Your Goal

Before creating any cards or recruiting participants, clarify what you want to achieve. Are you looking to generate new category ideas for a new information architecture? Test the effectiveness of predefined categories? Compare multiple structural approaches? Your goal should be specific and actionable. Great Question provides guidance on defining goals.

Step 2: Choose Your Card Sorting Type

Based on your goal, decide whether you'll conduct an open, closed, or hybrid card sort. Open card sorts work best for early exploration when you have no existing structure. Closed card sorts suit validation phases when you have specific proposals to test. Hybrid approaches balance exploration and validation when you have some direction but want to remain open to improvements. Great Question offers type selection guidance.

Step 3: Prepare Your Cards

List the items you want to test on cards--these might be names of web pages, product features, use cases, service offerings, or any other type of information you need to categorize. Make sure each card contains only one item or topic. Clarity in card labels directly impacts the quality of your results. IxDF covers card preparation best practices.

When creating cards, break information down to the smallest possible unit. If you have a card labeled "Insurance," you'll never know if people would have put "Car Insurance" and "Renter's Insurance" in different categories. The number of cards matters for participant fatigue--in experience, it takes about 30 minutes to sort 50 cards. More than 30-40 cards can lead to fatigue and rushed decisions. Great Question shares expert advice on card creation.

Step 4: Set Up Your Test

For remote card sorting using online tools, ensure your study includes clear instructions for participants so the session runs smoothly. Randomize display order to avoid bias from sequence effects. UXtweak covers remote setup requirements.

For in-person card sorting, reserve a quiet room with a large table where participants can spread out cards comfortably. Provide physical materials and consider using a recording device if you want to capture think-aloud protocols.

Step 5: Recruit Participants

Aim for a sample that represents your target users. Typically, 15-30 participants is enough for card sorting studies. Consider adding an incentive upon study completion to thank participants for their time and insight. Government employees typically can't receive monetary incentives--consider offering a donation to charity in their name instead. UXtweak recommends participant numbers. Great Question provides recruitment guidance.

Step 6: Conduct the Study

For remote studies, once you create your study and invite participants, results flow in automatically. For in-person sessions, start with clear instructions explaining the task. Encourage participants to think aloud as they sort, providing valuable qualitative insight into their reasoning. Observe quietly without intervening unless participants are clearly confused about the task itself.

Analyzing Card Sort Results

Starting Your Analysis

After completing your card sort, you have a mountain of data ready for analysis. This is where the magic happens--don't speed past it. Many designers and junior researchers spend a lot of time working on the card sort itself but not enough thinking about how they'll analyze the data or what decisions the findings should support. Great Question shares expert analysis insights.

Start by getting a broad overview of results. Scan for recurring patterns or trends in how participants sorted cards or the category labels they proposed during open sorts. Look for consensus areas where most participants agree, as well as areas of divergence that may indicate multiple valid perspectives or confusing items.

Qualitative Analysis

For open card sorts, review and standardize category labels. Participants might label categories in slightly different ways through varied spellings, capitalizations, or phrasing. By standardizing labels, you get a clearer picture of the general consensus. Identify common themes or groupings that emerge consistently, and take note of unique or outlier categorizations as they can provide insights into different user perspectives. Great Question outlines qualitative analysis approaches.

Quantitative Analysis

A similarity matrix shows how often each pair of items is grouped together, revealing which items participants consider most similar. Dendrograms visualize hierarchical clustering, showing how items relate at different levels of abstraction. Agreement scores measure the degree of consensus among participants, helping you identify which categorizations are widely accepted versus which remain contested. Great Question explains quantitative metrics. UXtweak covers quantitative analysis tools.

AI-Assisted Analysis

Modern UX research tools increasingly offer AI features that make it easier to spot hidden patterns and misalignments in user mental models. AI isn't a replacement for human analysis or decision-making but can be a helpful way to cut through noise and jumpstart analysis. Use AI-generated summaries as a starting point, then apply human judgment to interpret findings in the context of your specific project. Great Question explores AI in UX research analysis. Our AI automation services can help integrate these insights into your broader digital strategy.

Common Mistakes to Avoid

Not Having a Clear Goal

A successful card sort requires setting and communicating a clear goal. While that goal should be specific and actionable, don't lose sight of the bigger picture. Ensure you know what's important to customers and the business--you don't want to present results only to have leadership ask where their main revenue-generating feature is, but you didn't include it because no customer explicitly mentioned it. Great Question shares expert insight on goal setting.

Unclear Card Labels

Whether using internal jargon or labels too vague to mean any one thing, confusion kills the validity of your results. When cards are unclear, participants interpret them very differently, making results fuzzy. Be specific. Product teams are often so used to seeing certain terms that they assume users share the same language--they often don't. To avoid this mistake, do a small pilot study with a few participants before launching a full study. Great Question warns against unclear card labels.

Leading or Repetitive Cards

In creating cards, break information down to the smallest possible unit. If you have cards labeled "Tuxedo Pants," "Denim Pants," and "Palazzo Pants," you'll end up with a category called "Pants" when what might be more appropriate is separate categories like "Formal Wear," "Casual Wear," and "Resort Wear." Avoid leading or repetitive terms in card labels. Always ask what you'll learn from each card--if another card will tell you the same thing, pick one and remove the other. Great Question provides card creation guidance.

Too Many Cards

An effective card sort balances depth and duration. About 30 minutes to sort 50 cards is a reasonable benchmark. Be mindful of overloading participants with too many cards, which leads to fatigue and rushed decisions. Be as realistic as possible about the amount of time required, and when in doubt, err on the longer side--better to have users finish early than feel rushed. Great Question offers timing recommendations.

Skipping Follow-Up Methods

Too many teams conduct an open card sort and call it complete. Instead, consider following up with a closed card sort to test out a proposed hierarchy and even a tree test after that to assess goal-based success with a new information architecture. Card sorting is powerful but works best as part of a broader research strategy. Great Question emphasizes comprehensive research approaches.

Sources

1. Dscout: Guide: How to Conduct an Effective Card Sort for UX Designers - Comprehensive guide covering card sorting methods, recruiting participants, conducting sessions, and analyzing results.

2. Interaction Design Foundation: Card Sorting: The Ultimate Guide (2025) - Extensive coverage of six card sorting types, pros and cons, tools, step-by-step methodology, and best practices.

3. UXtweak: Card Sorting - The Complete Guide - Practical guide with focus on online vs. offline approaches, participant recruitment strategies, and analysis techniques.

4. Great Question: Card Sorting: 2025 Guide to Your Users' Mental Models - Modern guide featuring expert insights, tool comparisons, analysis techniques, and real-world case studies.