- Related metric
refrigeration_temperature_compliance- Trigger
asset_temperature > configured_upper_band, sustained ≥ persistence_window (with sensor health confirmed)- Value drivers
-
Product loss prevention
Customer experience
Avoided truck rolls
- Preconditions
- Temperature sensor operational and validated; upper band and persistence window configured per asset and product category; sensor health confirmed (invalid, stale, or suppressed intervals excluded).
- Human role
- Primary when a product check, dispatch, or store communication is required; secondary review after AI on validation.
- Action plan
-
(1) Validate the sensor and data quality — confirm the reading is real, not a failed or drifting probe, a stale feed, or a suppressed maintenance window; exclude invalid intervals.
(2) Assess duration and magnitude against the asset's configured upper band and persistence window — a brief in-window blip is not the same as a sustained excursion.
(3) Assess product and category risk — how much product, what category, and how close to the safety limit for that category.
(4) Correlate door events where door telemetry exists — a prolonged door opening or a burst of activity often explains the excursion.
(5) Compare with peer assets and rack-level patterns — a single asset points to that unit; a whole rack points upstream.
(6) Remote triage — where a controllable cause is found (schedule, defrost timing exposed by the controller, a resettable state), attempt it remotely first.
(7) Recommend a product check to the store when the excursion crosses the category's safety threshold for its persistence window; record the outcome.
(8) Scope a dispatch only when remote triage cannot resolve it, with the correlated evidence attached; capture closure evidence with root cause on close.
- Escalation
-
(1) Excursion crosses the product-safety limit for a high-value or high-risk category → out-of-cycle alert to store and customer ops with a product-check recommendation; do not hold for the MBR.
(2) Remote triage fails or no controllable cause is found → dispatch within SLA with the correlated evidence and a scoped diagnostic starting point.
(3) Many assets on one rack or site breach at once → suspect a rack, power, or connectivity event; escalate the site, not the individual cases.
(4) The same asset breaches repeatedly after a repair → flag for asset or controls review at the MBR; candidate for replacement.
- Prevention
-
(1) Configure the upper band and persistence window per asset and product category, and confirm the asset-to-category mapping, so an excursion is scored against the right target.
(2) Keep a per-asset excursion log and review assets in the upper half of their band range at cadence — catch a drifting asset before it becomes a sustained excursion.
(3) Refresh the sensor-health and suppression rules so data-quality gaps are excluded and do not either mask or manufacture excursions.