Skip to content
UX Atlas
Plate RES·005 · Research methodGrid G2 · Rev. 2026-08-14

Card Sorting

Asking users to group content, which reveals their mental model of a domain rather than your org chart.

Research methodEstablished researchIntermediate

Definition

Participants group items into categories that make sense to them. Open sorts let them name the groups; closed sorts use predefined categories to test an existing structure.

Established research. Originates in published research and has been replicated. The effect is real. Its size in your product still needs measuring.

On this plate
  1. 1Definition
  2. 2Which variant to run
  3. 3Running one
  4. 4Preparing the cards
  5. 5In practice
  6. 6Commonly confused with
  7. 7Local chart
  8. 8Sources
  9. 9Continue from here

Reach for this when

  • Designing or restructuring navigation.
  • Users cannot find things.
  • Naming sections and categories.
  • The team disagrees about where something belongs and nobody has asked anyone outside.

Common mistakes

  • Using internal terminology on the cards.
  • Too few or too many items.
  • Treating the result as a finished navigation structure.
  • Averaging results across segments that turn out to hold different models.

Which variant to run

VariantQuestion it answers
OpenHow do users naturally group this content, and what do they call the groups?
ClosedDoes our proposed structure match how users think?
HybridDoes our structure work, and what is missing from it?

Running one

  • Use 30 to 60 items. Fewer produces thin data; more exhausts participants.
  • Label items in the users' words, not internal terminology.
  • Fifteen to thirty participants gives reasonably stable clusters for a quantitative analysis.
  • Analyse with a similarity matrix or dendrogram rather than by eye.
  • Follow up with tree testing, because a sort tells you how people group and a tree test tells you whether they can find things.

Preparing the cards

Do

  • Use the reader's words on the card, not the internal product name.
  • Keep cards to between forty and sixty. Beyond that, participants tire and sort carelessly.
  • Include the awkward items you already argue about internally.
  • Let participants leave cards unsorted, and record what they abandoned.

Do not

  • Put two concepts on one card because they live on the same page today.
  • Seed the deck with your existing category names.
  • Read the result as a structure. It is evidence about expectations, not a sitemap.

In practice

  1. i.

    Sorts reveal vocabulary

    Information architecture

    The names participants give their groups in an open sort are frequently more valuable than the groupings themselves, because they are the labels the navigation should use.

  2. ii.

    Where the agreement stops

    Reading the matrix

    A similarity matrix normally shows a few tight clusters and a long tail of items nobody agrees on. The tight clusters are settled. The tail is where the naming argument actually is, and it is worth more attention than the part that already works.

Commonly confused with

Close neighbours that get substituted for each other. The difference is stated first, then the condition that picks one over the other.

vs.

Tree Testing

Research method

Participants locate items in a text-only hierarchy, with no visual design to help.

Key difference

Card sorting asks people how they would group things, and produces a proposed structure. Tree testing checks whether people can find things in a structure you already have. Sorting generates, testing validates.

Use Card Sorting when

  • The structure does not exist yet, or is being rebuilt.
  • You need the reader's vocabulary, not the organisation's.
  • The question is what belongs together.

Use Tree Testing when

  • A structure exists and is being judged.
  • You need a findability score you can compare before and after.
  • The question is whether labels lead people to the right place.

Learn about Tree Testing

Local chart

RES·005 · Region IV · Grid G2

This entry and the entries it is stated to connect with. Each line is labelled with what the connection means.

Match Between System and the Real WorldHEUR·002 · PRINCIPLES BEHIND THISMental ModelsMOD·001 · PRINCIPLES BEHIND THISTree TestingRES·006 · OFTEN USED WITHResearch PlanTPL·001 · OFTEN USED WITHFigJamTOOL·006 · OFTEN USED WITHOptimal WorkshopTOOL·021 · OFTEN USED WITHInformation ArchitectureTERM·011 · SHORT DEFINITIONTaxonomyTERM·012 · SHORT DEFINITIONNavigationPAT·001 · RELATED CONCEPTAffinity MappingRES·020 · RELATED CONCEPTInformation ArchitectureRES·015 · RELATED CONCEPT

Sources

Where a source establishes something narrower than the popular reading of it, the note says so.

  1. [1]

    Card Sorting: Designing Usable Categories

    D. Spencer, Rosenfeld Media, 2009

    Covers open, closed and hybrid variants, and how to analyse a sort without over-reading it.

Continue from here

Each link says what the connection is, so you can tell a principle from an alternative from a thing people mix this up with.

Applies when you are designing

NavigationSettings

Principles behind this

The reasoning this solution is an application of.

Often used with

These usually appear in the same screen or the same decision.

Short definition

The one-paragraph version, for when that is all you need.

Related concept

Connected closely enough to change how you apply this.