What is Card Sorting?

What if the best way to design a website's structure was to hand the job to the people who'll actually use it? Card sorting is a simple UX research method where participants group and label a set of topics in whatever way makes sense to them, showing how they expect information to be structured. It can inform navigation menus, site maps, and content hierarchies, replacing guesswork with real patterns in how people think

When to use card sorting

Card sorting is most valuable when you're making decisions about how information should be organized and you don't yet have evidence of how your users think about it.

Here are some valuable places you can hold a card sort:

Before You Build
Before any structure has been set in stone, running a card sort at the very beginning of a new website, app, or product, helps define information architecture. You get evidence of how your future users naturally group and label content.

During a Redesign
If analytics show high bounce rates on key pages, search logs are full of queries for content that already exists, or support teams keep fielding "where do I find...?" questions, those are signs your current structure doesn't match how users think. An card sort can reveal where your categories and labels diverge from user expectations, giving you a clear starting point for the redesign.

When Content Expands
As products mature, content piles up: new features, product lines, help articles, and services get added until the original structure strains under the weight. A card sort helps participants place new items into your existing categories, and suggest new ones when nothing fits.

When Stakeholders Disagree
Any time your team can't agree on the "right" way to organize things, card sorting can serve as a neutral tiebreaker. Participant data replaces opinion, seniority, and internal jargon with real patterns in how people think.

Before Validation: Pairing with Other Methods
Card sorting tells you how people group and label information, not how they actually navigate a finished design. That makes it a strong first step in a research sequence. Use card sorting to refine a structure, then follow up with tree testing to check whether people can find things in it, and usability testing to see how it performs in context.

How to conduct card sorting

1. Define your goals and choose a type

Start by deciding what you need to learn. Depending on your goals, you can select from different types of card sort tests:

Open sort: If you're generating a structure from scratch, run an open sort, where participants create and name their own categories.

Closed sort: If you're testing or extending an existing structure, run a closed sort, where participants place items into predefined categories.

Hybrid sort: Allows participants use your categories but add their own when nothing fits.

2. Select and prepare your cards

Pull together the content or features you want to organize, aiming for roughly 30 to 60 items. Write each as a short, clear label, and avoid internal jargon or labels that share obvious keywords (such as multiple items starting with "Account"). This will cause participants to group them by word rather than meaning.

3. Recruit participants

Recruit participants who reflect your real or intended users. For qualitative insight, 8 to 10 participants per user group is often enough to see patterns, while quantitative analysis, such as statistical clustering, typically calls for 15 or more, with many practitioners recommending around 30 for confidence.

4. Run your test

This test can be done with physical cards, digitally, moderated, or unmoderated. Either way, provide clear instructions, encourage participants to think outloud and provide area for participants to provide feedback.

5. Analyze results

Look for patterns across participants. A similarity matrix shows how often pairs of cards were grouped together, while a dendrogram visualizes clusters at different levels of agreement. Pay attention to category names too, since the language participants use can inform your navigation labels.

Cards that get scattered across many groups are signals of ambiguity or items that may belong in more than one place.

Card sorting with User Sage

1. Define your goals and choose a type

After defining your goal, select the type of card sort you’d like to run. List clear instructions for your users to follow.

2. Select and prepare your cards

You can upload up to 100 cards to User Sage. You can add cards individually, or paste a list.

Remember: write each as a short, clear label, and avoid internal jargon or labels that can unintentionally bias users’ choices.

Depending on the type of card sort you’ve chosen, you will have the option to add predefined categories for users to view. You can add up to 30 categories with User Sage.

Follow up with your users by asking them follow up trailing questions!

3. Recruit participants

With User Sage, you can utilize an AI-simulated panel to get a fast, baseline answer to your hypothesis.

Make sure to test this with real users! You can set up your own real panel, or utilize our GenPop Panel to get insight.

4. Run your test

It can take a few seconds for your AI panel results to come in. Watch as results are updated live.

5. Analyze results

View your results at-a-glance and analyze any common grouping patterns. Notice where your participants didn’t agree.

The Dendrogram and Similarity Matrix are wonderful tools to understand the degree of category agreement(s).

Things to look out for:

• Higher branching on dendrogram: indicates higher agreement.
• Darker similarity matrix boxes: indicates stronger agreement.

Ready to get started?

Start testing with User Sage’s card sorting feature. Sign up for free and validate your designs in minutes.