A tight threshold can notify during brief door openings or defrost. Widening it arbitrarily can also delay detection of a real excursion. The aim is not to minimise alarm count, but to send an actionable event to the right person.
Separate the threshold from the delay
First record which product or area is monitored, where the sensor sits and what condition counts as an event. Then set a temperature limit consistent with the applicable product requirements. If a notification delay is used, it filters brief fluctuations; it does not define an acceptable excursion time for the product.
Compare notifications with incident records. For each alarm, record its time, reading, sensor, operating condition and response. Classify notifications as a real temperature excursion, sensor or connection issue, or expected event such as a door opening. Review missed events too; counting excess notifications alone does not show that the threshold works correctly.
Change one setting and review again. Change either the threshold or delay, record the change date, and review alarm records alongside product measurements over the next comparison period. Reassess the configuration when product requirements, sensor location or operations change.
