Perceived-vs-Measured Performance Dashboard¶
Metric dashboard — instantiates Appearance vs. Reality Distinction Audit
Shows subjective user/operator perceptions alongside behavioral, telemetry, or task-performance measures without collapsing one into the other.
Perceived-vs-Measured Performance Dashboard puts two live streams next to each other — what people perceive about a system and what instruments measure about it — and refuses to collapse one into the other. Its defining move is temporal juxtaposition with a divergence alarm: it tracks both channels over time and fires when they pull apart, treating the gap itself as the signal worth investigating. Where a static table classifies one case, the dashboard watches two streams continuously and asks, in real time, "when perception and measurement disagree, which one is telling us something the other is missing?" It never assumes the metric is the reality and the perception the error — a divergence is data about both feeds.
Example¶
A streaming-video team runs a dashboard with two panels. The perceived panel aggregates in-app quality complaints and thumbs-down ratings; the measured panel shows rebuffering rate and startup latency from telemetry. Across most regions the two panels track together. Then, in one region, perceived quality craters while the measured metrics stay green. The dashboard raises a divergence alert.
Because the alert re-opens the question rather than declaring the users wrong, the team investigates the gap. They find the telemetry samples only sessions that reached playback — viewers who abandoned during a long spin-up never entered the measured feed at all, so the metric looked healthy precisely because the worst experiences were invisible to it. The dashboard then routes the finding: the fix is not to reassure users that things are fine, but to close the measurement blind spot and repair the abandonment path. Perception, in this case, was the more accurate channel — and the side-by-side view is what let the team see it.
How it works¶
- Run two independent feeds. A perception feed (ratings, complaints, survey pulses) and a measurement feed (telemetry, task times, error rates), each on its own axis, never averaged together.
- Align them in time. Plot both over the same window so agreement and divergence are visually obvious.
- Alarm on divergence. Define a gap threshold; when the streams separate beyond it, raise a trigger that re-opens the question instead of resolving it.
- Route by divergence type. Perceived-bad-while-measured-good and measured-bad-while-perceived-good point to different fixes — a measurement gap versus a communication or expectation gap — so the dashboard maps each pattern to an intervention owner.
Tuning parameters¶
- Divergence threshold — how far the streams must separate before the alarm fires; tight thresholds catch subtle gaps but cry wolf.
- Aggregation window — the time and population grain of each panel; too coarse hides regional or cohort divergences, too fine drowns in noise.
- Perception instrument — passive signals (complaints, ratings) versus active pulses (surveys); passive is cheap but self-selected, active is representative but intrusive.
- Routing map — how richly each divergence pattern is tied to an owning fix; a detailed map speeds response but hard-codes assumptions about causes.
- Baseline handling — whether panels show absolute levels or deviations from a learned baseline; deviations surface change faster but can normalize a chronically bad level.
When it helps, and when it misleads¶
Its strength is that it makes disagreement between appearance and measurement visible and actionable in real time, and it resists the reflex to trust the number over the human. The divergence alarm turns "users are complaining but the metrics look fine" from an argument into an investigation.
Its failure mode is Goodhart's law:[n1] once the measured panel becomes the target everyone optimizes, it drifts from the reality it was meant to track, and a green metric can coexist with a genuinely bad experience — exactly the blind spot the streaming example exposed. The classic misuse is using the measured panel to overrule the perceived one ("the numbers are fine, so the users are wrong"), which is measurement realism dressed up as rigor; the mirror misuse is chasing perception with cosmetic reassurance while the underlying measure quietly rots. The guarding discipline is to treat a divergence as a question about both channels — the metric may be mis-sampling reality just as easily as the perception may be miscalibrated.
How it implements the components¶
Perceived-vs-Measured Performance Dashboard realizes the audit's live divergence-monitoring machinery — the components that keep an experience stream and a measurement stream honest against each other:
phenomenological_report_channel— the perception panel is a standing channel for subjective experience, carried on its own axis rather than folded into the metric.measurement_corroboration_plan— the measurement panel is the instrumented feed that would corroborate or contradict the perception.decision_implication_map— the routing step maps each divergence pattern to the intervention it implies.revision_trigger— the divergence alarm re-opens the question whenever the two feeds separate past threshold.
It does not sort a claim by type with an appearance_reality_classifier (that's Appearance/Reality Audit Checklist), surface a claim's ontology_commitment_map (that's Symptom–Biomarker Crosswalk), or rewrite a claim into its experience_condition_set (that's Sense-Condition Rewrite Template).
Related¶
- Instantiates: Appearance vs. Reality Distinction Audit — the dashboard is the audit run continuously over two live streams.
- Sibling mechanisms: Claim Tagging Matrix · Appearance/Reality Audit Checklist · Observation Warrant Ladder · Symptom–Biomarker Crosswalk · Bridge Assumption Annotation Protocol · Sense-Condition Rewrite Template
Editorial Notes¶
Form Classification¶
Form family: Monitoring, Sensing & Alerting
Rationale: Perceived-vs-Measured Performance Dashboard operates as ongoing observation, sensing, or alerting that detects and surfaces state without itself executing the response because it shows subjective user/operator perceptions alongside behavioral, telemetry, or task-performance measures without collapsing one into the other.
Independent corroboration: The frozen evidence defines Perceived-vs-Measured Performance Dashboard as 'Shows subjective user/operator perceptions alongside behavioral, telemetry, or task-performance measures without collapsing one into the other', so its operative form is Monitoring, Sensing & Alerting.
Review outcome: Independent reviewer agreement; high confidence.
Origin Attribution¶
Primary origin: Human-Computer Interaction
Origin pattern: Cross-disciplinary synthesis
Present-day reach: Multi-domain
Rationale: Juxtaposing subjective experience with observed task or telemetry measures is canonical usability and user-experience evaluation practice.
Related originating lineages:
- Data Science & Analytics — Perceived-vs-Measured Performance Dashboard is rooted in data science and analytics: Analytics practice juxtaposes telemetry with subjective measures to expose proxy and Goodhart failures. Telemetry instrumentation and dashboard construction supply the measured-performance layer.
- Organizational & Management Science — Organizational and management science materially shaped Perceived-vs-Measured Performance Dashboard through coordination, organizational learning, performance, and change practice.
- Psychology — Psychology and behavioral science materially shaped Perceived-vs-Measured Performance Dashboard through perception, judgment, learning, motivation, and behavioral bias. Psychometrics and perception research supplied disciplined treatment of subjective reports as distinct evidence.
Review resolution: Light authoritative-source research resolves the primary-origin disagreement in favor of human-computer interaction and user-experience research. NIST: Common Industry Specification for Usability Requirements directly documents the defining practice or theory described in the selected origin rationale. Other listed domains are retained only where the blind reviews identify material co-development or translation; broader adoption remains separate as domain_reach=multi_domain.
Attribution caveat: The boundary with data science and analytical instrumentation is real because that field materially developed or translated the practice, but the cited provenance places the defining form in human-computer interaction and user-experience research.
Encyclopedia synthesis: The exact catalogued form synthesizes established practice rather than reproducing a single standard historical label.
Review outcome: Researched adjudication after independent review; high confidence.
Sources consulted:
Notes¶
The dashboard shares two components with Symptom–Biomarker Crosswalk — both hold an experience channel next to a measurement channel — but they are not interchangeable: the crosswalk is a static table for a single case across fixed clinical lanes, while the dashboard is a live temporal surface that watches streams and fires on divergence. Reach for the crosswalk to adjudicate one patient; reach for the dashboard to monitor a population over time.
[n1] Goodhart's law — "when a measure becomes a target, it ceases to be a good measure." Once a team optimizes the measured panel directly, it can decouple from the experience it was meant to proxy, which is why a dashboard must keep the perception channel independent rather than treating the metric as ground truth. ↩