Tree-test paths diagnose label and hierarchy problems
Aliases: path diagnosis · label confusion · hierarchy mismatch
What it is
Tree-test paths localize label and hierarchy mismatch through the first divergence, backtracking, and competing sibling nodes. They identify where and suggest why, but failure rate alone cannot decide whether to relabel, move content, or offer multiple entrances.
Why it happens
Convergence on one wrong node indicates stronger scent there; dispersion at the top suggests broad category ambiguity; failure inside the correct parent points lower. Similar paths can still arise from task concepts, domain knowledge, or structure and need interpretation.
Studying it
Analyze first-click matrices, path flows, and backtracking by task and segment. Probe interpretations of competing labels after tasks without replacing path evidence with explanation. Change label or hierarchy separately and retest.
Where it stops holding
Aggregate success hides distinct failures and small path shares are unstable. A single correct target creates artificial failure when content legitimately belongs in multiple places. Search and cross-links may repair paths excluded by the tree.
Applying it
- Locate first divergence and common competing nodes.
- Form separate label, hierarchy, and multiple-entry hypotheses.
- Change one mechanism at a time and retest.
- Verify search and alternative entrances in the full prototype.