Card Sorting
A research method where participants group your content into categories, revealing the mental model they already hold.
Process & MethodsCard sorting hands people the content — labels on cards — and asks them to group it. Open sorts let participants name the groups themselves, which surfaces the vocabulary they use and is what you want before an information architecture exists. Closed sorts give them your categories and measure how cleanly the content falls into them, which is what you want when you have a structure and suspect it is wrong.
The output is a matrix of agreement: which items were grouped together by how many people. The interesting part is rarely the consensus, which usually confirms what the team assumed. It is the items that split the room — those are the ones with unclear labels or genuinely two homes, and they are exactly where users will get lost in navigation.
Sorting is not the same as finding
Card sorting tells you how people would organise things. Tree testing tells you whether they can find something in the structure you built. They answer different questions and the usual sequence is sort, structure, then test.
In practice
An insurance site ran an open sort with 30 customers on 48 content items. “Claims” and “Report an incident” were grouped together by 27 of them; the site had them in separate top-level sections, three clicks apart. Merging them removed the most common path to the support phone number.
Where teams get it wrong
- Sorting too many cards. Past about 50, participants get tired and the data degrades.
- Using internal labels on the cards, which tests your vocabulary rather than theirs.
- Running a closed sort before there is any structure worth testing.
- Reading only the consensus and ignoring the items that split participants — those are the findings.
- Treating the result as the IA. It is input to a structure, not the structure.
Learn more
You may ask
Frequently Asked Questions
What is the difference between open and closed card sorting?
Open sorts let participants create and name their own groups, revealing their vocabulary and mental model. Closed sorts ask them to file content into categories you supply, testing whether your structure holds.
How many participants does a card sort need?
Around 15–30 for a quantitative agreement matrix. Fewer than that gives you directional themes but not reliable grouping percentages.
What is the difference between card sorting and tree testing?
Card sorting asks how people would organise content; tree testing asks whether they can find something in the structure you already have. Sort first, build, then test.
Related terms
All terms- Tree TestingTesting whether people can find things in a proposed structure, using a bare text hierarchy with no design, navigation or search.
- Information ArchitectureThe structure of a product's content and navigation — how things are grouped, labelled and related so people can find what they need.
- User ResearchThe systematic study of users' goals, behaviour and context — through interviews, observation and analysis — to inform what gets designed and built.
- SitemapA structured list of a site's pages and their hierarchy — used both as a design artefact and as an XML file that tells search engines what exists.
Defined by Mara Last reviewed .
Let's talk about your product.
Happy to look at what you're building and say where design would move the needle.
Contact Us