Card sorting for navigation and IA
What card sorting is
In a card sort, participants arrange cards, each representing a piece of content or a feature, into groups that make sense to them. The result shows the mental models of your audience, which helps when designing menus, categories and information architecture (IA).
Choose the type
Open
Participants create their own groups and name them. Best when exploring a new structure.
Closed
Participants place cards into categories you define. Best for testing an existing set of categories.
Hybrid
Participants use your categories but may add their own.
Sessions can be moderated (you observe and ask why people grouped items) or unmoderated (online tools collect many sorts with less effort, but without the explanation behind each choice).
Prepare good cards
- Use real content items from the product, such as page titles or feature names.
- Write clear, plain labels. Avoid internal jargon and avoid words that repeat across many cards, because they hint at groups.
- Keep the set manageable. Too many cards tire participants, and too few give thin results.
- Run a quick pilot to catch confusing labels.
Read the results
- Look for agreement: which cards do many participants place together?
- Note similar groupings that appear under different names.
- Collect the labels people propose; they often suggest better menu wording.
- Treat cards with no agreement as a signal that the item is unclear or belongs in more than one place.
Card sorting helps build a structure, but it does not test whether people can find things in it. Verify that with a tree test.
From results to navigation
Card sort results are input to a decision, not the decision itself. Use them to draft a structure and then refine it with judgement about the business and the content.
- Draft top-level categories from the groups most participants agree on.
- Choose labels from participants’ own words where they are clear.
- Decide where items belonging to several groups should live, and consider cross-links.
- Check the draft with a tree test before building the interface.
Remember that participants group what they see on the cards. If the cards miss important content, the structure will too, so review the card list against the real content inventory.
This guide was drafted with AI assistance from public standards and references. For the exact wording of a standard, use the sources listed above.