Open card sorting reveals participants' categorization schemes
Aliases: open card sort · category scheme · participant labels
What it is
Open card sorting lets participants group and label content, revealing perceived similarity and category language. It produces schemes under a supplied card set and task, not a complete, stable, or unique user mental model.
Why it happens
Grouping externalizes similarity and labels expose the dimension used. Card selection, granularity, and instructions constrain possible structures. One person may reasonably organize by topic, task, or object, while aggregation can erase consequential minority strategies.
Studying it
Collect groups, labels, explanations, and hard-to-place cards. Use similarity matrices and clustering as aids rather than treating a dendrogram as navigation. Compare segments and plausible stable solutions and return to original sorts for interpretation.
Where it stops holding
Open sorts explore language and organization but do not test findability in an existing tree. Sorting cards is not navigating an interface. Too many, specialized, or mixed-granularity cards confound categorization with workload.
Applying it
- Use independently understandable cards at comparable granularity.
- Permit alternative groupings, renaming, and unclassified items with reasons.
- Generate candidate structures from common and critical minority patterns.
- Validate candidates through tree testing rather than shipping a cluster diagram.