UX Glossary · C
Card sorting
Also: Open card sort · Closed card sort
Asking users to group and label content so the structure reflects their model rather than the org chart.
Card sorting asks participants to organise items into groups that make sense to them. Each item goes on a card, physical or digital, and the sorting reveals the categories people actually use.
Open sorts let participants name their own groups, which tells you both the structure and the vocabulary. Closed sorts give them your categories and test whether items land where you expect. Hybrid sorts offer categories and allow new ones.
What it is good and bad at
Card sorting is good at surfacing the mental model: which things people consider similar, and what they call them. It is bad at telling you whether a finished navigation works, because sorting cards on a table is not the same task as finding something on a website.
That second question belongs to tree testing, where participants are asked to locate a specific item in a proposed structure. Sorting generates the structure, tree testing validates it. Running only the first is the common mistake.
In practice
Somewhere around 15 to 30 participants is typical for reasonably stable groupings, and useful signal appears earlier than that. The most valuable output is often not the clusters at all but the labels: the words participants reach for are the words your navigation should use.