Hybrid working broke the assumptions every office was designed around. A space planned for a hundred people at desks now holds forty on Tuesday and twelve on Friday, and almost nobody knows those numbers with any confidence — which makes every decision about space, rent, heating and refurbishment a guess.
Measuring occupancy properly is straightforward, cheap and legally sensitive. Getting the last part right is what determines whether it works.
Decide what question you are answering
Occupancy data is collected for very different reasons and the method should follow the reason.
“How much space do we need?” Needs peak and average headcount by day, over months. Coarse data, no individual identification, and by far the most valuable question.
“Which rooms are wasted?” Needs per-room utilisation. Room-level sensors, still no identification.
“Is this meeting room free right now?” Needs live per-room presence, and it is the question staff most appreciate having answered.
“Should we heat and light this floor today?” Needs live zone occupancy, and it is where the operational saving is.
“Who is in the building?” Needed for fire roll call and nothing else. This is the one that requires care, because it identifies individuals.
“How much is each person in the office?” This is monitoring individuals, it is legally fraught, and it damages trust badly. We would advise against collecting it, and if an employer insists, it is a matter for formal consultation rather than a technology decision.
The methods, and what each really tells you
Passive infrared motion sensors. Cheap, entirely anonymous, and they tell you whether a space is in use rather than by how many. Their weakness is well known: someone sitting still reads as absent, so a meeting room with four people looking at a screen may report empty.
Presence radar. Detects stationary people, and better models report a count. Anonymous, considerably more accurate for meeting rooms, and more expensive.
People counters at doorways, using optical or thermal sensing. Give an accurate running headcount for a floor or a building, and they need careful placement and calibration — two people walking abreast is the classic error.
Desk sensors, under the desk or in the chair. Accurate per desk, and a lot of devices to maintain.
Wi-Fi association counts. Free, already available from your access points, and it counts devices rather than people — one person with a laptop, a phone and a tablet counts as three. Useful for trends, poor for absolute numbers, and depending on how it is done it may be personal data.
Badge and access control data. Accurate for entries and it identifies individuals, which puts it squarely in employment and data protection territory.
Carbon dioxide as a proxy. Underrated. People exhale it, so the rate of rise in a room correlates with how many people are in it. Entirely anonymous, cheap, and it also tells you whether your ventilation copes — two useful answers from one sensor.
Booking system data. Free and misleading on its own, because the gap between rooms booked and rooms used is frequently large. Comparing bookings against sensor data is where the insight is.
The legal part, which is not optional
In the EU and UK, monitoring in a workplace engages data protection law as soon as data relates to identifiable individuals, and employment law adds obligations on top.
Principles that keep you on the right side:
Aggregate and anonymous by default. A sensor that reports “three people in room 4” is not personal data. One that reports who they are is.
Purpose limitation. Collect for a stated purpose and do not repurpose. Occupancy data gathered for space planning should not later be used to assess individual attendance, and doing so is exactly what destroys staff trust.
Data minimisation. Do not collect identity if a count will do, and do not retain raw data longer than the analysis needs.
Transparency. Tell people what is being measured, where, why and for how long. Signage at the entrance and a short written notice. Covert workplace monitoring is a serious matter in most jurisdictions.
Consult. Where there is a works council, union or staff representative body, involve them early. In several European countries this is a legal requirement rather than a courtesy, and Germany and the Netherlands in particular have strong co-determination rules on monitoring.
No cameras for counting where a sensor will do. Cameras in workspaces are a substantially higher bar to justify, and thermal or optical counters that produce no identifiable image are a better answer.
What the data usually reveals
Some findings recur often enough to be worth expecting.
Peak day utilisation is lower than everyone believes, and the peak is almost always Tuesday or Wednesday.
Meeting rooms are booked far more than used. Recurring meetings that nobody cancelled, and rooms held “just in case”. This is why automatic release — freeing a room that shows no occupancy after ten minutes — is the highest-value intervention in most offices.
Large rooms are used by small groups. A twelve-person room hosting three people is the commonest pattern, and it argues for more small rooms rather than more space.
Whole zones are empty on Fridays, while being heated, lit and ventilated as though full.
Quiet space is the real shortage. Open-plan offices with plenty of desks and nowhere to take a call is the complaint that consistently comes out of occupancy work.
Turning it into operational savings
The measurement pays for itself here rather than in the property decision.
Consolidate onto fewer floors on quiet days, and heat and light only those.
Zone ventilation to actual occupancy, using carbon dioxide, which both saves energy and improves air quality where people actually are.
Automatic meeting room release, which increases effective capacity without any building work.
Lighting and heating on real occupancy, with generous timeouts so nobody sits in the dark.
Cleaning on use rather than schedule, which is a genuine saving in a building where half the floor was empty.
Right-size the lease at renewal, with months of evidence rather than an argument.
How we would start
- Write down the questions you want answered, and the ones you will deliberately not ask.
- Consult staff and any representative body before installing anything.
- Start with carbon dioxide sensors in meeting rooms and open areas. Anonymous, cheap, and they answer occupancy and air quality together.
- Add presence radar in meeting rooms where you need accuracy.
- Compare bookings against actual use for a month.
- Implement automatic release, which will pay for the sensors.
- Zone heating, lighting and ventilation to what you now know.
- Report aggregates only, and delete raw data on a defined schedule.
Steps one and two are what make the rest acceptable. Occupancy data collected without a stated purpose and without telling anyone is how a sensible energy project becomes a trust problem.