INF2170H Chap.5 Navigation, Wayfinding and Tree Testing
Navigation, Wayfinding and Tree Testing
Test paths through a hierarchy
Navigation offers routes, while wayfinding concerns orientation and the cues used to choose a direction. Tree testing isolates hierarchy and labels from the full visual interface.
This chapter writes realistic information needs, records first choices and backtracks, and distinguishes final success from direct success.
Repair the first divergence
The earliest unexpected branch often reveals where information scent breaks. Participant explanations help distinguish ambiguous task wording, overlapping categories and weak labels.
A structural test can succeed even when the live interface fails, so conclusions remain tied to the layer the method actually examined.
Write tasks that preserve the information need
A test scenario should describe a realistic goal without repeating the target category. Context belongs only when it changes the choice.
Before a session, confirm that every task has one authoritative destination or an explicit rule for acceptable alternate routes. Pilot the wording by asking someone to paraphrase the need. If they misunderstand the scenario, revise it before evaluating the hierarchy.
Sample meaningful variation. Include high value destinations, overlapping siblings and content that has several plausible homes.
Do not let one branch dominate the set merely because its tasks are easy to write. Freeze the tree version and task wording used in a round so later results can be compared against a known structure rather than a moving target.
Separate destination success from path quality
Record first choice, every subsequent selection, backtracking, destination and participant explanation.
A successful endpoint can conceal a weak route when the participant wanders or guesses. First choice locates the earliest divergence, while direct success shows whether the information scent remains coherent along the full intended path. Inspect both before changing the architecture.
Diagnose the layer of failure.
If participants choose the intended top category but fail lower down, preserve the top label and inspect subordinate boundaries. If first choices split across plausible siblings, compare scope and language. If the tree performs well but the live interface fails, move to presentation, interaction or page content instead of rewriting a structure the test supported.
Every conclusion should name what the stripped down method did not examine.
What this chapter covers
- 01
Navigation structure
- 02
Wayfinding cues
- 03
Tree testing
- 04
Task scenarios
- 05
First choice
- 06
Backtracking
Worked example · free
Design a tree testing task
- 2State the user need in ordinary language.
- 3Record path and backtracking.
- 2Limit the conclusion to hierarchy and labels.
Key terms
- Information Scent
- Information Scent is the expectation that a label or cue leads toward the needed content.
- Tree Testing
- Tree Testing evaluates hierarchy and labels through task paths without the complete interface.
- Direct Success
- Direct Success records reaching the intended destination without backtracking or detour.
Navigation, Wayfinding and Tree Testing FAQ
What does tree testing isolate?
Tree testing isolates hierarchy and label interpretation by removing most visual interface detail. It records the path through a text based tree for realistic information needs. It does not evaluate layout, styling, interaction design or the destination page itself.
Why should first choice be recorded?
The first choice identifies the earliest point where the participant’s expectation meets the hierarchy. A participant can eventually succeed after a detour, so final success alone may hide a weak top level label or overlapping category boundary.
How should a task scenario be written?
Describe a realistic information need in the participant’s language without copying the target label or teaching the route. Include only context that changes the decision, pilot the wording and keep one authoritative destination so the path can be scored consistently.
Assessment move
Sketch a compact tree and write tasks that avoid target labels. For each trial, record first choice, path, backtracking and destination. Diagnose the first divergence before changing the whole hierarchy, then state which interface layers remain untested. Build tasks from realistic situations, not commands that repeat a branch name.
Write the expected destination separately and ask whether more than one endpoint could satisfy the information need. During a trial, capture the first branch, complete path, backtracking and final selection without coaching. Analyse the earliest point where unsuccessful routes diverge, because later errors may be consequences rather than causes.
Compare the task wording with the participant's interpretation before changing the tree. Propose the smallest structural or label revision that addresses the evidence, then rerun the same task and one neighbouring task to check for side effects. Keep tree-test claims distinct from interface claims: a text hierarchy does not evaluate layout, visual prominence, search or page content.
For wayfinding practice, explain how global, local and contextual cues help a person answer where they are, what is nearby and how to recover. Finish each diagnosis with a bounded conclusion and the next appropriate test.
