Card sorting is a UX research method where participants group labeled cards into categories that make sense to them. It reveals how users mentally organize content, which you use to shape information architecture and navigation.
Use cases
When your navigation mirrors how your org is structured instead of how users think, people can't find things. They give up, fall back to search, or leave.
The internal-jargon menu: A bank groups pages by department names like "Retail Solutions." Users hunt for "checking account" and find nothing. A card sort surfaces the words people actually use.
The overstuffed redesign: A content team has 60 pages and no agreement on grouping. An open sort shows the natural clusters instead of another whiteboard argument.
The validation check: A PM proposes a five-category structure. A closed sort tests whether users can place items where the team expects, before engineering builds the menu.
How it's used in practice
Pick your type by stage: Open sort to discover groupings, closed to validate a proposed structure, hybrid when you have a partial one.
Keep labels neutral: Avoid repeating keywords across cards. Matching words push people to sort by wording instead of meaning.
Right-size the set: Roughly 30 to 60 cards. Too many exhausts people; too few hides patterns.
Recruit 15 to 30 real users: That range gives stable results without burying you in analysis.
Read the dendrogram and similarity matrix: Find the clusters most people agree on, then the messy cards that split the room.
🪄
Pro-tip: The disagreements are the gold. Cards that land in five different groups are telling you the label is ambiguous or the concept doesn't belong. Consensus tells you what's safe; the outliers show you where the IA will break.
Challenges & limitations
Groupings aren't navigation: A card sort tells you how people cluster content, not your full menu, hierarchy, or final labels. You still design that part.
Context-free sorting: People sort cards in isolation, without the task or page around them. Results can drift from real behavior.
Tools
UXtweak — best value. Open/closed/hybrid, image cards, solid reporting, and a genuinely usable free plan.
Lyssna (formerly UsabilityHub) — clean, fast, good for quick unmoderated sorts alongside first-click and five-second tests.
Useberry — unmoderated research suite with card sorting baked in; decent if you want everything in one place.
