A research method where participants organize topics or content items written on cards into groups that make sense to them. Results reveal users' mental models for how information should be categorized, informing information architecture decisions.
Common contexts
- Running an open card sort with 20 users before redesigning a help center's category structure
- Using a closed card sort to validate that a proposed navigation taxonomy matches user expectations
- Identifying which product features users group together to inform a settings menu reorganization
Use when
Run a card sort before making decisions about navigation labels or category structures — especially when moving existing content to a new information architecture. Open sorts are best for discovery; closed sorts are best for validation after you have a proposed structure.
Avoid when
Card sorting is a poor tool for testing whether navigation works in practice — it reveals mental models for categorization, but not whether users can successfully find things. Always follow a card sort with tree testing or usability testing to validate that the resulting structure actually helps people navigate.
The labels users give their groups in an open card sort are often more informative than the groups themselves — the language they use is your navigation copy waiting to be written.
Real-world examples
- Spotify used card sorting with users to restructure its settings menu, discovering that users expected 'Downloads' under 'Playback' rather than a top-level menu item.
- GOV.UK used open card sorting extensively when redesigning its navigation, allowing citizens to group government services in ways that reflected their mental models rather than departmental silos.
- Atlassian ran remote card sorting sessions to reorganise Confluence's template library after users reported difficulty finding the right starting point for new pages.