methodResearch & behavior

Card Sorting

A categorization method in which participants group and sometimes label content items to reveal patterns in how they expect information to be organized.

Download Markdown (.md)

01What it means

Card sorting asks participants to organize labeled content cards into groups that make sense to them. In an open sort they create and name groups; in a closed sort they place cards into supplied categories; hybrid sorts combine both. The method reveals similarities and terminology patterns that can inform an information architecture, but it does not produce a finished navigation system automatically.[1][2]

Participant groupings reveal expectations about organization Prepare representative items, observe grouping, compare patterns, then test a proposed structure
  1. 01Define scopeInformation problem, participants, and sort type
  2. 02Prepare cardsRepresentative, clear, and non-leading content items
  3. 03Participants sortOpen, closed, or hybrid grouping and labels
  4. 04Analyze patternsAgreement, disagreement, rationale, and segments
  5. 05Propose structureCombine evidence with content and product constraints
  6. 06Validate navigationUse tree testing or realistic tasks

Card sorting informs information architecture but does not directly produce it; task context, labels, content relationships, and navigation testing still matter.

02Why it matters

Teams often organize content around departments, databases, or internal vocabulary. Card sorting introduces evidence about how intended users associate concepts and what labels they use. Comparing individual structures highlights strong clusters, ambiguous items, and differences between audiences before a navigation hierarchy becomes expensive to change.

03When to use it

Use card sorting when creating or revising an information architecture, especially for content-rich sites, help centers, catalogs, or settings. Open sorting supports discovery of candidate groupings; closed sorting evaluates whether an existing category set can absorb the content.

04How it works

  1. Select representative content at a consistent level of granularity and write concise, distinguishable card labels.
  2. Choose open, closed, or hybrid sorting based on whether the goal is discovery or evaluation.
  3. Recruit participants who understand the domain as intended users do, not merely internal experts.
  4. Ask participants to sort independently, explain uncertain cards, and name groups where applicable.
  5. Analyze recurring clusters, labels, disagreements, and audience differences; draft a hierarchy for later testing.

05A practical example

Example

A university redesigning student services gives current students cards such as “replace an ID,” “appeal a grade,” and “find emergency housing.” Students group urgent life needs together rather than under administrative departments. The team drafts task-oriented categories, then uses tree testing to check whether students can find specific services within the proposed hierarchy.

06Where it breaks

Sorting cards removes content from the navigation, page, and task context in which it will be used. Participants may group by superficial wording, and aggregate clusters can hide meaningful segments. Results depend heavily on card selection and granularity. Card sorting suggests structures; it does not validate findability, navigation labels in context, or final menu depth.

Common mistakes

  • Mixing broad categories, specific tasks, and duplicate concepts in the same card set.
  • Treating a similarity dendrogram as a complete information architecture.
  • Skipping follow-up tree testing because participants produced tidy groups.

07Sources

  1. Card Sorting: Uncover Users’ Mental Models for Better Information ArchitectureNielsen Norman Group · Professional guidanceCard Sorting: Uncover Users’ Mental Models for Better Information Architecture
  2. What is Card Sorting?Interaction Design Foundation · Professional referenceWhat is Card Sorting?
View saved topics

Start typing to search titles, definitions, categories, and frameworks.