Card sorting and tree testing

Two methods help you design and validate a structure with real users.

Card sorting

In a card sort, participants group content items written on cards. In an open sort, they create and name their own groups, which reveals how they think about the content. In a closed sort, they place cards into categories you provide, which checks whether your categories make sense. Card sorts can be run on paper or online; 15 to 30 participants give useful patterns for online sorts.

Tree testing

A tree test shows participants only the text structure (the "tree") with no visual design, and asks them to find items: "Where would you go to change your delivery address?" You measure whether they succeed, whether they went straight there, and where they went wrong. Because there is no design, it tests the structure and labels alone.

Using them together

A common sequence: open card sort to generate ideas, draft a structure, tree test it, revise, tree test again. Card sorting is generative; tree testing is evaluative.

Designers and stakeholders tend to organise content the way they understand it, which is rarely how users do. These methods replace opinion with evidence before anything is built, when changes are cheap. Tree tests in particular give clear success rates that are easy to share and compare between versions.

An outdoor clothing shop is restructuring its menu. An open card sort with 25 customers shows that people group by activity ("hiking", "running") more often than by product type ("jackets", "trousers").

The team builds a structure with activity as a main route and product type as a secondary one. A tree test with eight tasks shows strong success for most, but only 35% of people find "waterproof trousers" because they look under "Hiking" rather than "Rain wear". The team adds waterproof trousers in both places and renames "Rain wear" to "Waterproofs". A second test raises success to 78%.