Tree testing isolates structural findability from visual design
Aliases: tree testing · structural findability · information scent
What it is
Tree testing presents a text hierarchy and asks participants to locate task targets, testing category structure and label information scent. Removing layout, icons, search, and visual salience isolates structural findability but does not represent overall interface usability.
Why it happens
At each level, people use labels to predict the target location. Wrong first choices, reversals, and abandonment expose mismatches. Removing visual cues reduces confounding while also removing aids and interference that exist in the product, so the test evaluates the navigation skeleton.
Studying it
Define correct targets and acceptable alternative paths and record first click, final success, direct success, backtracking, and route. Randomize task order and compare target segments. Keep tasks and labels stable across candidate trees and disclose excluded search and cross-links.
Where it stops holding
Tree testing cannot evaluate layout, visual hierarchy, controls, search quality, or content comprehension. A text tree is not equivalent to real screen-reader navigation. One successful task does not validate the tree, and failure may originate in task wording.
Applying it
- Preserve actual hierarchy and labels in the text tree.
- Cover frequent, consequential, and confusable finding tasks.
- Inspect first choices and backtracking, not final success alone.
- Retest visual and interaction effects in a full prototype.