The technique came from library science and was popularised in UX by Donna Spencer's Card Sorting (2009) and the Nielsen Norman Group's workshops. Three variants: open card sorts let participants create and name their own groups; closed card sorts give participants pre-named categories; hybrid card sorts combine pre-named categories with the option to create new ones.
Modern card sorting runs on tools like Optimal Workshop, Maze, UserTesting or dedicated whiteboard tools. In-person sessions use physical index cards. Each has trade-offs: digital scales to larger samples, physical surfaces the thinking-out-loud more clearly.
Card sorts reveal mismatches between the team's mental model of the domain and the users'. A product team that calls a feature "Fulfilment Operations" when users call it "Shipping" has a card sort problem. Running the sort before the design starts saves a reorganisation six months later.
Before any substantial navigation design, taxonomy work, help centre reorganisation, or search project. Also useful during content migrations and when consolidating product lines after an acquisition. Not useful for small feature additions — the card sort overhead outweighs the signal.
A typical card sort uses 30–60 content items with 15–30 participants. Open sorts produce more insight but messier output; closed sorts validate specific hypotheses. The results feed into cluster analysis (participants who grouped items similarly) and dendrograms (visualising category similarity across the sample).
Pair with tree testing — card sorts generate the structure, tree tests validate whether users can find things once it's built.
Card sorts feed into information architecture and complement tree tests. The outputs inform navigation, search and content strategy. Often paired with user interviews to surface the vocabulary that explains the groupings.
Reviewed 1 October 2026 · Editorial standards