5,000 classrooms — three years of continuous data
Heating systems ran when rooms were empty more often than not. The pattern held across both fleets independently.
Two local authorities. Different boilers, different controls vendors, different facilities teams. Three years of continuous monitoring across roughly 5,000 classrooms between them.
The finding that surprised nobody on the data side and everyone on the operations side: heating systems were running into empty rooms more often than they were running into occupied ones.
Weekend mornings. Half-term weeks. The first hour before the school day actually started. Wings of the building that had been mothballed years ago but were still on the boiler's timetable.
The two fleets had no operational overlap. The pattern was independent in each. The pattern was the same.
What the data made obvious.
When the timetable disagrees with the calendar, the calendar wins. Boilers don't know about INSET days. They run to a schedule someone set years ago and nobody has revisited since.
Continuous monitoring made the disagreement legible — and then made it actionable. The fix isn't sophisticated. The visibility is.
Names withheld by request. Stories are published anonymously until customers approve attribution. The pattern, the data, and the finding are real.