What is Card Sorting in Information Architecture (IA)

Card sorting is a research method where participants group content elements (cards) by meaning and give names to the groups. The goal is to discover natural mental models for navigation and inform the structure and labels of menus, categories and help sections.

Types of Card Sorting

• Open card sorting: participants come up with names for the groups. Useful in the early phase to discover the language of the users.
• Closed card sorting: pre-defined categories; participants only distribute the cards. Useful for validating an existing structure.
• Hybrid card sorting: proposed categories, but new ones can be added. Combines discovery and validation.

When to use

• When creating or redesigning navigation and information architecture.
• When there is low discoverability of content or confusing labels.
• After content audit, to group content elements into logical sections.

How it works

• Prepare a list of cards: key pages, products, themes.
• Choose a session type: in-person with physical cards or remote with online tool.
• Instruction: goal, context and rules (no wrong answers).
• Grouping: participants arrange the cards into meaningful piles.
• Labeling: give names to the groups (open or hybrid format).
• Discussion: short notes why they grouped them that way.

Sample and size

• 8 to 15 participants give stable models for qualitative analysis.
• For greater confidence, use 20 to 50 participants in remote sessions.
• Select people from the target audience and key segments.

Analysis and outputs

• Similarity matrix: how often two cards are grouped together.
• Dendrogram and clusters: suggest candidate structures.
• Navigation proposal: categories, subcategories and labels.
• Test hypotheses: what to validate with tree-testing.

Tools

• Optimal Workshop
• UXtweak
• UserZoom
• Useberry
• Maze

Best practices

• Use user language on the cards, avoid internal jargon.
• Limit the number of cards to 30 to 60 to avoid fatigue.
• Balance of detail: cards should be specific enough, but short.
• Divide by scenarios, if you have a lot of content (several sessions).
• Collect and qualitative notes: why the cards are together, alternative names.

Common mistakes

• Смесване на съдържателни единици с действия (страници и задачи в едно).
• Too specific or internal terms, not known to the users.
• One format: only open or only closed without subsequent validation.
• Interpretation without statistical check of similarity and stability of groups.

Link to Tree-Testing

Card sorting generates hypothetical structure and labels, and Tree-testing validates whether people quickly find what is in this structure. The best process is to use them together in iterations.

Mini scenario

• Goal: new IA for online store.
• Cards: product types, services, return policy, delivery, help.
• Format: hybrid card sorting with 20 participants.
• Next step: create a candidate menu and validate with tree-testing.