The short answer
The test is simple: if the reading arrived, what would change? If the answer is nothing, the sensor produces a dashboard rather than a benefit.
That test rules out most proposed installations and makes the remaining ones easy to justify.
Which ones earn their place
| Sensor | Decision it changes |
|---|---|
| Leak detection | Attend immediately, before damage spreads |
| Temperature and humidity | Intervene before stock or fabric is affected |
| Door and access | Investigate out of hours activity |
| Consumable level | Restock before it runs out, not on schedule |
| Occupancy | Clean by actual use rather than fixed rota |
The top two justify themselves through avoided damage. The bottom two only pay off if the schedule genuinely changes in response, which requires operational will as well as data.
The costs people forget
- Batteries, which need replacing across many devices
- Connectivity in basements and plant rooms
- Mounting and access for maintenance
- Someone to act on alerts, including out of hours
- Client permission to install in their building
The fourth is the real one. An alert with nobody to act on it is worse than no alert, because it creates a record that you knew.
Start with one site
- Pick the site with the clearest problem.
- Install the smallest number of sensors that addresses it.
- Define in advance what response each alert triggers.
- Run for a season, because conditions vary through the year.
- Decide on evidence whether to extend.
Running through a full season matters for environmental sensing. A reading pattern in summer tells you little about winter.
Avoid the dashboard trap
A wall of readings looks impressive and changes nothing. Prefer alerts on thresholds that require action over continuous displays that require interpretation.
If you want a display, make it show exceptions rather than everything, so that a glance is enough.