The sensor worked. Deciding what the alert should trigger was harder.

Connecting a sensor and seeing a number on a screen feels like finishing the project.

Often it is only the beginning.

Data becomes useful when it changes a decision: stop something, notify someone, register an incident or keep evidence for later review.

Measuring is not the same as monitoring

A sensor can report perfectly every minute and still be useless. You need to decide which range matters, how long a value must stay outside that range, who gets notified and what happens if nobody responds.

Then come the less attractive questions: what if connectivity drops, how do we know the device stopped reporting, what happens to impossible values and how much history do we need?

The physical world fails differently

Software requests can be retried. In the field there is power, connectivity, moisture, distance, cables and people unplugging things because they needed the outlet.

The objective is not to have sensors. It is to know something early enough to act on it.