Card Sorting: Open vs Closed vs Hybrid (How to Run One + Tools)

Card sorting is a great way of UX research that gives designers an idea of how users group and structure information. It is an important part of building intuitive website navigation, logical content structures, and user-friendly information architecture. 

But the right way is important. Open, closed, and hybrid card sorts are used for different research purposes and provide different insights into user expectations. 

This blog will walk you through the differences between the 3 methods, how to conduct card sorting step-by-step, and tools to help you effectively organize, analyze, and interpret your card sorting results.

What Is Card Sorting in UX?

Card sorting in UX is a research method used to understand how users group and organize information. Participants sort cards representing website pages, app features, or content categories into groups that make sense to them. UX designers use the results to create clear navigation systems and organize information in a way that feels natural to users.

A user, for example, may group “Billing History,” “Payment Methods,” and “Invoices” into a “Billing” category on a SaaS dashboard. Their selections demonstrate their expectations about how information is structured.

Card sorting has three flavors – open, closed and hybrid. Open sorting allows users to establish their own categories, closed sorting offers them existing categories, and hybrid sorting is a mix of the two methods. Your approach selection relies on whether you are building a new information structure or appraising an old one.

Open vs Closed vs Hybrid Card Sorting: What's the Difference?

The main difference between open, closed, and hybrid card sorting is how participants organize information into categories. In open card sorting, participants create their own categories. In closed card sorting, they sort cards into predefined categories. In hybrid card sorting, they can use existing categories or create new ones when needed.

Here is a comparison of the three ways:

1. Open Card Sorting

Open card sorting permits individuals to make their own groups and name them. You don't decide the categories, you just deal the cards.

Think about building a platform for project management. Your cards have tasks, team members, project reports, notifications, integrations and billing info.

Participants might create categories like “My Workspace,” “Team Management,” and “Account Settings.” Another person may sort the same objects differently.

These differences help you understand the mental models of users and discover surprising ways to arrange information.

Conduct open card sorting to:

• You are creating a new website or application.
• Your navigation as it is is confusing for users.
• You are not sure which categories make the most sense.
• You want to find language that meets what users expect.

The Nielsen Norman Group’s introduction to card sorting describes how the process discloses users’ mental models and guides information architecture.

One key limitation: the titles of the categories of the participants are study findings and not necessarily the final navigation labels. You should refine them with further research and usability testing. 

2. Closed Card Sorting

In closed card sorting, participants are given a list of pre-defined categories. They place each card in what they feel is the greatest category.

Let’s imagine your SaaS dashboard already includes four navigation sections: Projects, Team, Reports and Settings. You ask consumers to classify individual qualities into these categories.

If many participants have trouble placing “Usage Analytics” in a category, you might want to rethink your category structure or clarify labels.

When to use closed card sorting:

• You’ve already got a planned information architecture.
• You want to see how well material fits into current buckets.
• You’re adding new features to an existing product.
• You want proof to support a navigation structure before committing to it.

Closed sorting can identify mismatches, but it doesn’t necessarily tell us whether users can locate information when executing real tasks. Tree testing is generally better suited for this. 

3. Hybrid Card Sorting

Hybrid card sorting is a combination of open & closed. Participants can choose from pre-defined categories or create new ones if none of the existing categories are suitable.

A finance application could have Income, Expenses, Accounts, and Settings. If participants feel a category is deserving of its own section during the exercise, they can add a new category titled "Subscriptions".

This allows you to assess an initial structure without entirely limiting the participants’ choices.

When to use Hybrid card sorting:

• You have a draft of your information architecture and you're not sure about all the categories.
• You're adding new features to a mature product.
• Some sections of navigation are set, while others are to be discovered.
• You wish to find missing categories without beginning anew.

But the categories you define can shape how others categorize information.  If you want to understand users' natural groupings without that influence, begin with an open sort.

How to Do Card Sorting in UX: A Step-by-Step Process

An excellent card sorting session does not begin with a tool but with a clear research subject. Think about why you want to recruit people or generate cards, and how you will use the data ahead of time. Once you have a clear goal in mind, you can move on to the practical steps below:

Step 1: Define Your Research Goal

Start with one specific question.

You might want to see if customers understand how to navigate your SaaS dashboard or how potential purchasers anticipate to locate pricing information on your website.

The card sorting method you use will depend on your study’s goal.

• Card sorting, open. Pick.
• Validation: Closed card sort
• Explore and investigate: Try a mixed card sort .

Keep it reasonable in scope . Testing a whole business platform at one time can over-tax participants and make results difficult to interpret.

Step 2: Prepare Your Cards

Each card should be a single content item, feature or topic. Keep labels short, straightforward, and unbiased.

For a SaaS application, your cards could include:

• Projekt erstellen
• Invite team members
• See project reports
• Your subscriptions
• Update payment information
• Change notification preferences
• Link third-party applications
• Download invoices

Don’t use labels that tell the user which category you anticipate them to select. A card entitled “Billing Settings: Update Payment Methods,” for example, could prompt participants to club it with other billing items.

Research from the Nielsen Norman classify on card sorting language illustrates how the phrase might influence how participants classify items.

Start with a reasonable number of cards, e.g. 20-40 cards for a typical examination of a website or product. This is a useful place to start, not a universal cap. A complex product may need to be studied in several smaller studies. 

Step 3: Recruit the Right Participants

Your participants should be the people who will be using the product.

If you are redoing a project management solution, bring in project managers, team members or other appropriate users, not only internal designers.

How many participants would you want? In a first investigation, 15-30 suitable individuals can generate useful patterns for numerous card-sorting applications. However, the right size of sample relies on your audience, the complexity of your study and your research goals.

Smaller qualitative research can also uncover valuable problems. For instance, the UK Department for Education conducted an open card sort with nine participants to inform a draft information architecture. Read its card-sorting case study.

Don't take any participant number as a statistical confidence. Get enough people to probe meaningful variations across your major user segments. 

Step 4: Run the Card Sorting Session

The task is to put the cards in order.

In an open card sort, you ask participants to group related cards together and name each group. For a closed card sort, offer them the existing categories and ask them to slot each card into the right category.

You can do a card sorting session in person with physical cards or remotely with an online card sorting tool.

In a moderated session, ask participants to explain their selections. Ask unbiased follow-up questions, such as:

• What did these mental things connect? 
• Was any card difficult to place?
• Did any item seem to belong in multiple groups?
• Was there a category you expected to see but couldn't find?

According to Digital.gov's card sorting guidance, setup generally takes one to two hours, followed by approximately 15–30 minutes per participant. Actual timing varies with the number of cards and the complexity of the task.

Step 5: Analyze the Results

When they are done you look for repeating patterns on the participants. Not the structure of any one member.

If several clients bundle invoices, payment methods, and subscription plans, you might want to consider creating a separate Billing Section.

Search results for:

• Common Clumps What cards are common clumps?
• Labels: What labels do participants use most frequently?
• Disagreements: Which cards are always difficult to classify?
• Outliers: Who’s sorting differently? ‘Why?’

Features or information that people expect from your product but that it does not supply.

Most online card sorting tools generate similarity matrices, dendrograms and category summaries. They help to see the connections between cards, but they are not a substitute for your personal interpretation of the participant input. 

Step 6: Turn Findings Into an Information Architecture

Sketch out your navigation using the strongest patterns and then talk about potential categories with your product team.

Don’t think that the most popular grouping is the best answer. You also need to take into account user tasks, product priorities, accessibility and technical constraints.

A practical next step is to build a sitemap or navigation prototype and test it with users. Let your card sorting results impact your design, but don’t let them dictate every decision you make.

If you want a more extensive approach on how to tie research to product decisions, check Orbix Studio’s tutorial on UX strategy. 

Card Sorting Tools: Free and Online Options

You do not need a costly research platform to do your first card sorting experiment. This can be a shared digital board, paper cards or a specific research tool.

The best choice relies upon whether you need remote involvement, automatic analysis, reusable templates or basic collaboration. Check out this table and you will get the primary idea of which tool to use based on the needs.

Features, free plans and product availability are subject to change. Check the current pricing and plan limits of each provider before you make your choice.

What Is the Best Free Card Sorting Tool?

The best free card sorting tool depends on your research needs and the type of study you want to conduct. 

• Miro or a shared Figma board works well for quick, collaborative sessions. 
• UXtweak is worth considering for dedicated UX research, but check its current free-plan limits and analytics features. 
• For in-person studies, paper cards and a spreadsheet offer a simple, low-cost option. 
• If you already use a survey or research platform, check whether it supports card sorting before choosing another tool.

A dedicated card sorting tool can reduce the manual effort involved, by surfacing trends and collecting responses. A general purpose whiteboard is flexible but you may have to work harder to organize and make sense of the data.

If you’re using a card sort template, check it’s appropriate for the type of study you’re doing. An open card sorting template should allow participants to create the categories and a closed card sorting template should provide categories you have selected.

Card Sorting Template and Example

A basic card sorting template should have the research purpose, instructions for participants, labels on the cards, rules for categories and space for observations.

Here's a practical example of a SaaS project management dashboard.

Research Goal: Understanding how consumers anticipate to discover project management features.

• Labels Cards
• Make a project.
• Share the labor
• Add team mate Add team mate
• See reports on projects
• My subscriptions
• Payment method change
• Integrations Connect
• Configure notifications
• Edit account information
• Download invoices

Open card sorting session: Sort the cards in a way that makes sense to you. Name each category Explain any tough choices.

Closed card sort task: Sort each card into Projects, Team, Reports, Billing, Integrations or Settings.

Hybrid card-sort task: Use the same six categories to begin but if a card doesn’t go well with a category, create a new one.

After the task is finished, look at the categories you came up with and go over the cards that were hard. Develop a navigation structure for the strongest patterns.

You can use this template for a website redesign, a mobile app, an enterprise dashboard or a product onboarding flow. If you're mapping the early stages of a new product, Orbix Studio's guide to wireframes, mockups, and prototypes can help you decide how to translate the findings into a design.

Card Sorting vs Tree Testing: Which Should You Use?

Choose card sorting when you want to understand how users expect information to be grouped and organized. Choose tree testing when you want to check whether users can find specific information within an existing website or app navigation structure. 

If you're building a new information architecture, start with card sorting. If you already have a proposed structure, use tree testing to see how well it works. Here is a clear comparison for easy understanding

For example, through card sorting you may find out that consumers expect to see invoices and payment options under Billing. You may then test with tree testing if visitors can find an invoice in your proposed navigation successfully.

Both strategies reinforce each other. Create a structure using card sorting first. Then test it using a tree test and check whether people are able to use it successfully. 

For reference, the US Department of Energy’s usability guidance says that card sorting shows how customers think about content, but doesn’t necessarily give you a useful website layout.

Can You Use AI or ChatGPT for UX Card Sorting?

Yes, you can use AI tools like ChatGPT to support UX card sorting by generating card ideas, suggesting related categories, summarizing participant results, and identifying follow-up research questions.

However, AI-generated categories are only suggestions, not proof of how users naturally group information or what they expect. Use AI to support your research, but rely on real user feedback to validate your information architecture.

This Is How to Use AI Effectively for Card Sorting:

1. Generate Clear Card Labels

Provide AI with a product description, and it can help turn your existing pages, features, or content into short, neutral card labels. Ask it to identify confusing, overlapping, or repetitive items that may need clarification.

2. Explore Possible Categories

Ask AI to suggest different ways to group the cards. Use these suggestions to explore possible categorization patterns and generate hypotheses, not to make the final information architecture decisions.

3. Create a Card Sorting Template

Use AI to prepare a card sorting template that includes a clear list of cards, participant instructions, and neutral follow-up questions. You can then transfer the content into a spreadsheet or your preferred research tool.

4. Review Anonymized Study Results

Once real participants have completed the study, share the anonymized results with AI. Ask it to identify common grouping patterns, disagreements, unusual responses, and cards that participants found difficult to categorize.

5. Compare AI Suggestions with Real User Data

Compare AI-generated suggestions with the actual findings from your study. Use real participant behavior to guide your decisions, refine your navigation structure, and validate the final information architecture through user testing before launch.

But artificial intelligence is unable to tell you what your real consumers desire until you have real proof from real users. It can speed up preparation and analysis, but should not substitute for full research on the participants. 

For product teams, this distinction matters: use AI to make the research process more efficient, not to manufacture research findings.

Frequently Asked Questions

What Is Card Sorting in UX?

Card sorting is a user experience (UX) research method that helps you understand how people group and categorize information. It aids designers in developing website navigation, app menus, and information architectures.

What Is the Difference Between Open and Closed Card Sorting?

In open card sorting, users generate and label their own categories. In closed card sorting, the categories are supplied so that researchers can determine whether existing information structures are in line with users' understanding.

What Is Hybrid Card Sorting?

Hybrid card sorting uses both open and closed. Participants sort cards into current categories, but are free to create new categories if the provided alternatives don’t capture their perspective properly.

What Is the Best Free Card Sorting Tool?

Which one is the greatest free card sorting tool depends on what you require. Collaborative free card sorting tools include Miro and Figma, while dedicated UX research platforms could be able to offer analysis automation.

How Many Participants Do You Need for a Card Sort?

For many card sorting tests, useful early trends can be found from about 15-30 individuals. The ideal sample will rely on the study aims, the heterogeneity of the audience, the complexity of the project and the level of confidence required.

What Is the Difference Between Card Sorting and Tree Testing?

Card sorting reveals the way clients want information structured. Tree testing is to examine if users can find certain information with a given navigation arrangement. Either of these solutions can lead to a superior info-architecture.

Can You Use AI or ChatGPT for Card Sorting?

Yes, ChatGPT can help generate card labels, suggest categories, organize research data, and identify patterns. However, always validate AI-generated recommendations against real participant feedback before making final design decisions.

Final Thoughts

If you’re establishing a SaaS product, or redesigning an existing digital experience, combining user research with good information architecture can help you make decisions about your navigation based on data, not internal assumptions.

We are Orbix Studio. We enable startups and developing organizations to better digital products with UX/UI design, SaaS design and product development. If your team needs to translate user research into a clear, scalable product experience, check out our UX design principles and best practices to plan your next move.