Link-Label Scent Audit¶
Checklist — instantiates Predictive-Cue Wayfinding Design
A recurring review pass that walks every label, heading, button, and link and checks it against what an agent actually finds after clicking, flagging weak, ambiguous, or mismatched cues.
The Link-Label Scent Audit is a disciplined walk through a space's textual cues — labels, headings, button captions, link text — comparing what each says against what an agent finds on the other side. It is a recurring inspection driven by a checklist: for each cue, does its wording predict the destination for the intended audience, or is it vague, insider jargon, overloaded, or quietly wrong? The defining idea is that it is an activity performed against reality — a periodic pass that produces a findings list — rather than a stored artifact or a live meter. It grades cues one at a time and hands back a prioritized set of mismatches to fix.
Example¶
A company's employee intranet has accreted labels over a decade. The people team runs a scent audit before the annual benefits enrollment. Working from a checklist, a reviewer walks each navigation item and asks the same questions: What would a first-year employee expect behind this? What is actually there? Does the wording match? "My Requests" turns out to hold IT tickets and time-off approvals — an overloaded cue two audiences read two ways. "Total Rewards" is HR jargon that novices read as a loyalty program, not compensation. "Forms" leads to a page that has not held the enrollment form since it moved to a new system two years ago — a stale label pointing at a promise it no longer keeps.
Each finding is logged with a severity and the evidence behind it, including the help-desk tickets and search logs showing employees routing around the labels to reach the same pages. The output is not a redesign but a ranked worklist: split "My Requests," rename "Total Rewards," repoint "Forms." The audit's job is to find the mismatches, credibly and repeatably; fixing and owning them is downstream.
How it works¶
What distinguishes the audit is that it is a structured, item-by-item comparison anchored in outcomes, not opinion:
- Checklist-driven coverage. A fixed set of questions is applied to every cue, so weak spots are found systematically rather than where a reviewer happens to look.
- Promise-versus-reality for each item. Each cue is graded on the gap between what it implies and what its destination actually contains.
- Outcome evidence, not just judgment. Where available, the reviewer pulls the trace of what agents did after the click — the mismatch signal — to ground each flag in behavior, not taste.
- A prioritized findings list. The deliverable is a severity-ranked set of specific, actionable mismatches.
Tuning parameters¶
- Scope — every cue versus the high-traffic or high-stakes subset. Full coverage is thorough but slow; a sampled pass trades completeness for cadence.
- Cadence — one-off, quarterly, or triggered by a content change. More frequent audits catch drift sooner but cost reviewer time.
- Reviewer perspective — an insider (fast, but blind to jargon) versus a novice or outsider (slower, but reads cues the way lost users do).
- Evidence source — the reviewer's own judgment versus real post-click behavior. Behavior-grounded audits are far harder to dismiss.
- Flag threshold — how bad a mismatch must be to make the list. A low bar surfaces everything but buries the severe cases.
When it helps, and when it misleads¶
The audit is the cheapest way to catch cue drift — labels that were once accurate and slowly stopped being so — and to expose insider jargon that reads clearly to the team and opaquely to everyone else. Its behavioral tell is pogo-sticking: agents bouncing into a destination and straight back out, a signature the audit is built to hunt.[1]
Its central failure is the curse of knowledge: a designer auditing their own labels rates them clear because they already know what is behind them, so a self-review of insider cues reliably passes cues that novices cannot read. It is also a snapshot — accurate the day it is run and decaying afterward — and its severity grades are subjective unless anchored in evidence. The guarding discipline is to audit with outsiders or, better, with real post-traversal data, and to re-run on a cadence rather than trusting one clean pass.
How it implements the components¶
local_cue_surface— the audit systematically walks the full inventory of textual cues, which is the surface it inspects.cue_destination_correlation_model— each checklist item grades how well a specific cue predicts its actual destination, cue by cue.wayfinding_feedback_signal— it pulls post-click behavior (bounces, re-searches, help tickets) as the evidence that grounds each flag.
It reviews cues for honest error but does not adversarially hunt for cues engineered to mislead (misleading_scent_guardrail, false_scent_exception_log) — that offense-oriented search is Misleading-Cue Red Team; nor does it maintain the owned cue-to-value register (cue_freshness_update_rule), which is Cue-Destination Alignment Matrix.
Related¶
- Instantiates: Predictive-Cue Wayfinding Design — supplies the recurring review that keeps textual cues honest against their destinations.
- Consumes: Scent Clickthrough Trace Dashboard — the audit leans on its post-traversal telemetry to ground each flag in behavior.
- Sibling mechanisms: Breadcrumb and Landmark Trail · Cue-Destination Alignment Matrix · Destination Preview Card · Misleading-Cue Red Team · Progressive Disclosure Preview · Route Recovery Pattern · Scent Clickthrough Trace Dashboard · Task-Based Wayfinding Test
Editorial Notes¶
Form Classification¶
Form family: Assessment, Review & Assurance
Rationale: Link-Label Scent Audit operates as a bounded evaluation of existing evidence or work that produces a finding or disposition because it a recurring review pass that walks every label, heading, button, and link and checks it against what an agent actually finds after clicking, flagging weak, ambiguous, or mismatched cues.
Independent corroboration: The frozen evidence defines Link-Label Scent Audit as 'A recurring review pass that walks every label, heading, button, and link and checks it against what an agent actually finds after clicking, flagging weak, ambiguous, or mismatched cues', so its operative form is Assessment, Review & Assurance.
Review outcome: Independent reviewer agreement; high confidence.
Origin Attribution¶
Primary origin: Human-Computer Interaction
Origin pattern: Cross-disciplinary synthesis
Present-day reach: Specialized
Rationale: Information-scent and link-label testing are established HCI and usability practices for navigation design.
Related originating lineages:
- Cognitive Science — Information-foraging theory and predictive-cue cognition materially shaped the concept of information scent.
- Library & Information Science — Information architecture and navigation labeling contribute the systematic audit of paths and destinations.
Review resolution: Both independent reviews assign primary provenance to human_computer_interaction. The queued secondary differences (alternate_origin_disagreement, origin_mode_disagreement) are reconciled by retaining cognitive_science, library_information_science only as formative or independently established lineage(s), not merely as application domains. origin_mode=cross_disciplinary_synthesis records the provenance relationship, while domain_reach=specialized separately records applicability breadth. confidence=high preserves the more cautious assessment, and encyclopedia_synthesis=false records whether either reviewer identified a corpus-specific synthesis.
Review outcome: Reconciled after independent review; high confidence.
References¶
[1] Nielsen, J. E-commerce User Experience. Nielsen Norman Group (2001). Defines pogo-sticking as users repeatedly moving from a results hub into destination pages and back again. registry ↩