Separate one bad unit from the network died #

A Harbor store goes dark. A Polar Freight trailer stops checking in. Support’s first fork is simple and expensive if wrong: one bad unit, or the network (often Starlink or another ISP) died under a whole cohort?

This story is the connected-product framing of the same decision as Blame Starlink (or not) before you blame your product.

Without Dataplicity #

Every ticket looks local. Firmware bisects and truck rolls start while the corridor is dark. Customers hear “we are investigating your device” during a provider outage.

With Dataplicity #

Open the reported unit with customer and site context - confirm identity before acting.
Widen to the cohort - same class, site, or location:isp:* tag - before treating the case as unique.
Compare monitors and connection quality on the population that moved together.
Write the altitude on the incident (one unit vs network vs product) and only then choose remote fix, dispatch, or provider escalation.

Outcome #

Support spends minutes classifying the fault class - not hours repairing the wrong layer.

Separate one bad unit from the network died #

Without Dataplicity #

With Dataplicity #

Outcome #

Next #