Tree Test Template for User Research

[Card sorting](/templates/card-sorting-template) asks users to organize items into groups and label them. It answers: "How do users think about these items?" Tree testing asks users to find items within a predefined structure. It answers: "Can users find things in our proposed IA?" Card sorting is generative (creates a structure). Tree testing is evaluative (tests a structure). Run card sorting first to inform the IA design, then tree test to validate it.

Optimal Workshop's Treejack is the industry standard. It handles tree construction, task randomization, results visualization (pietrees showing path distributions), and benchmarking. UXtweak and PlaybookUX also support tree testing. For budget-conscious teams, you can run a manual tree test using a clickable text outline in a shared document, though you lose the path analytics and automated scoring.

8-12 tasks is the recommended range. Fewer than 8 does not cover enough of the IA to be useful. More than 15 causes participant fatigue and lower data quality on later tasks. Include a mix of easy (top-level, unambiguous) and hard (deeply nested, potentially ambiguous) tasks. Start with 2-3 easy tasks to build participant confidence.

Yes. Some items legitimately appear in multiple places (e.g., "Integrations" might logically live under both "Settings" and "Tools"). Mark all valid endpoints as correct answers. In your analysis, note which path was most popular. If users are split 50/50 between two locations, consider cross-linking or duplicating the item in the actual UI.

If your full IA has 100+ nodes, test subtrees rather than the entire structure. Focus each test session on one major section (e.g., "Settings and Admin" or "Content and Reports"). Test the top-level structure separately with tasks that require navigating across sections. This modular approach keeps sessions under 15 minutes and maintains data quality. ---