Description
Testing a text-only navigation tree with task scenarios and coding each failure as wrong branch, backtrack failure, label failure or false confidence to target fixes.
Tree testing isolates information architecture from visual design: participants navigate a text-only hierarchy to find where they would complete a task scenario. Standard analysis reports success, directness and time per task. Gopinath's framework adds a navigation error taxonomy so that failures point to specific remedies. Each failed task is coded as wrong branch (the top-level taxonomy misled), backtrack failure (participants could not recover), label failure (right area, but item labels did not discriminate) or false confidence (a wrong answer accepted without doubt). Full click paths and first clicks are analysed because the first click strongly predicts eventual success. The taxonomy converts success rates into a diagnosis: restructure, relabel or add recovery cues.
Attributes
- Discipline
- Community-Proposed Frameworks
- Loss type
- Community-proposed
- Tested
- Not yet tested
- Stage
- Design, Evaluation
- Data
- Quantitative
- Effort
- Days
Steps
- Build the tree: Text-only representation of your IA — no visual design, no colours, no icons. This isolates IA findability from visual design quality.
- Task scenarios: Write 8–12 task scenarios. Each should be answerable by finding a specific item in the tree. Scenarios must be written in participant language — not using the exact label names in the tree (this defeats the test).
- Record full navigation paths: Track every node click, not just the final answer. The path is the data — correct destination via incorrect path is as informative as an incorrect destination.
- Error taxonomy coding: Code each failed task by type: (a) wrong branch (participant went to entirely wrong area), (b) backtrack failure (couldn't find the way back after going wrong), (c) label failure (correct branch but couldn't identify specific item from labels), (d) false confident (reached wrong destination and didn't try again).
- First-click analysis: What was the first node each participant clicked? First-click accuracy predicts overall task success with ~87% accuracy — and first-click errors reveal where the IA's top-level taxonomy fails to match users' mental model.
Sources
- Tree Testing Playbook (Gopinath)
- Information Architecture (O'Reilly)
- How to Make Sense of Any Mess (Covert)
Also known as
- Tree testing with error taxonomy
- Reverse card sort with error coding
- Gopinath tree test error taxonomy
Categories
- Community-Proposed Frameworks