Every Australian university library opens at 9am by reflex, seven days a week, on a schedule designed for the busiest day it will ever have. The case for matching opening hours to actual demand starts with measuring the demand honestly. This post is about the methodology a privacy-first occupancy pilot uses to make that case, not a result we've delivered, but the way we'd run a deployment when the first one starts.
01 · THE QUESTION WORTH ASKINGThe question worth asking
Before instrumentation, the conversation between a facilities lead and a library operations manager usually stalls on two things: nobody has the data to support an opening-hours change, and nobody wants to make the change without it. Anonymous occupancy counts cut the knot. The data is decision-grade because it answers the operational question directly: how full is the floor, by zone, by hour, day after day.
02 · WHAT A PILOT WOULD MEASUREWhat a pilot would measure
A first-floor library pilot is small on purpose: five zones (reading room, quiet study, café, atrium, entry), each sensed by a single edge device emitting anonymous counts. Capacity is calibrated against the fire-egress plan; counts are normalised against capacity for density. No faces, no video, no PII.
Over a twelve-week trailing window, the report we'd produce covers three things:
- Hour-by-hour utilisation per zone, with the four-week average overlay so deviations read at a glance.
- Day-of-week occupancy curves: Sundays, Saturday mornings and the first two hours after opening are the windows most often misaligned with policy.
- Annualised cost of misaligned opening hours, modelled against the building's HVAC and lighting load. Right-sizing a single floor's Sunday opening typically returns a four-figure energy saving per year.
03 · THE HVAC MATH, WITHOUT THE PILOTThe HVAC math, without the pilot
The cost-recovery model is independent of the pilot data; it's the same arithmetic in any building. Plant load during pre-cool and the first hour of conditioning, multiplied by a wholesale-to-retail blend, multiplied by the number of days per year the schedule is misaligned. The pilot supplies one input: how often the schedule is misaligned. The math does the rest.
04 · WHAT IT DOESN'T MEASUREWhat it doesn't measure
A note on what the sensors don't see, because it always comes up:
- The model runs on-device. Raw frames never leave the sensor.
- The platform sees anonymous counts per zone, every second. No faces, no biometric data, no PII.
- Conversation history and reporting is scoped per tenant.
That's the entire data flow. The number facilities needs is on the dashboard. The number a regulator would object to is never created.
05 · WE'RE LOOKING FOR THE FIRST UNIVERSITY PILOTWe're looking for the first university pilot
We don't have a university library pilot running yet. We're scoping the first one. If you're a property lead at an Australian campus and the AHU is running before the readers arrive, we'd like to count it with you. Request a demo or see how privacy works.