Workplace Utilization: Measuring How Offices Are Really Used

On a Tuesday in March, a facilities team pulled badge-swipe data showing 71% of staff had entered the building. The same afternoon, a walk-through of floors three and four found eleven people spread across 140 desks. Both numbers were correct. Neither answered the question the CFO was asking, which was whether the lease on floor four should be renewed at all.

That gap between "who came in" and "what space was actually used" is the whole problem of workplace utilization. Most organisations have the first number in abundance. Very few have the second in a form they would stake a multi-year property decision on.

Why the data you already have keeps misleading you

Three sources get used as a proxy for workplace utilization, and each fails in a predictable way.

• Access control. Counts entries, not exits, and not where people go. Tailgating through a propped door is invisible. A badge swipe at 08:40 says nothing about whether that person left at 11:00 for client meetings.
• Room booking systems. Measure intent. Studies of booking-versus-actual use in corporate estates routinely find a third or more of booked meeting rooms sit empty for the reserved slot. A calendar full of bookings can describe a building that is half-vacant.
• Wi-Fi association. Counts devices, not people. One person with a laptop, phone and tablet becomes three. Access point coverage rarely aligns with the rooms or neighbourhoods you actually want to measure.

None of these are useless. But if the decision in front of you is consolidating two floors, converting a bank of 60 desks into collaboration space, or committing a university to a new teaching block, the margin of error in proxy data is larger than the margin of the decision.

Measure at the level you intend to change

The most common mistake in a first space utilization project is measuring everything at the same resolution. Building-level counts are cheap and tell you about commuting patterns and peak days. They tell you nothing about whether the quiet zone on floor two is quiet because people love it or because nobody can find it.

A practical hierarchy looks like this:

• Building and floor: people counting at entrances and stair/lift cores gives peak workplace occupancy by day and hour, the number that drives cleaning contracts, catering volumes and security staffing.
• Room: presence counting inside meeting rooms, lecture theatres and project spaces, so you can compare booked capacity against actual heads. A 12-seat room averaging 3.4 occupants is a design finding, not a scheduling one.
• Desk: a desk occupancy sensor at individual positions, only where the decision genuinely hinges on desk-level behaviour, such as deciding a sharing ratio for a department moving to hybrid.

VemSpace was built around this layering. The same platform measures room, desk, floor and building utilisation, with real-time occupancy and alerts when a space hits a predefined limit. That last feature matters more than it sounds. A facility manager in a university does not need to know afterwards that a 200-seat hall was over capacity during exams; they need a notification while it is happening.

The practitioner's warning about the first six weeks

Anyone who has run a workplace occupancy deployment will tell you the same thing: do not present the first month of data to leadership. Not because the sensors are wrong, but because the building is not behaving normally yet.

Staff notice new devices on ceilings and under desks. Some avoid the monitored areas; others crowd into them out of curiosity. Facilities teams, cleaners and IT contractors walk the floors more than usual during commissioning, and every one of them is counted unless the system is configured to exclude them. Modern AI sensors with staff exclusion handle the second problem technically, but the behavioural novelty effect only fades with time. Six to eight weeks is a reasonable period before you treat the data as a baseline, and you should annotate the dataset with anything unusual: a fire drill, a department offsite, a public holiday that moved.

On accuracy, be careful about what vendors promise and what you repeat upstairs. Vemco's contractual minimum is 96%, and installations typically run at 98 to 99% when lighting, layout and movement patterns allow. That is good enough for any property decision, but it is not 100%, and quoting a flat 99% in a board paper is the kind of claim that gets picked apart later.

From dashboard to decision: three uses that actually pay

Utilization dashboards are easy to produce and easy to ignore. The projects that justify their cost tie the data to a specific recurring expense or a one-off capital choice.

Right-sizing the footprint. If a floor peaks at 38% occupancy on its busiest day across a full quarter, that is evidence, not opinion. It supports a lease negotiation, a sublet, or a reconfiguration into fewer, better desks and more rooms, since the booking data usually shows meeting space is the thing people fight over.

Cutting the cost of always-on systems. Most office HVAC, lighting and ventilation schedules were set once, to the design occupancy of the building, and never revisited. A floor heated and ventilated for 180 people while 40 are present is paying to condition empty air. VemFusion passes live occupancy data into the HVAC, BMS and security systems so set-points and air volumes follow actual presence rather than the timetable. In public buildings, where opening hours are long and attendance is uneven, this is frequently the fastest route to a measurable energy saving.

Planning the next building. Universities in particular commission teaching space on enrolment projections. Room-level occupancy from the existing estate gives the planning team an evidence base for how many seats a 40-student module actually fills, and how often the "flexible" rooms are used flexibly at all.

What to put in the specification

Before tendering, write down the five decisions you expect the data to inform in the first two years. Then specify measurement only where those decisions need it. Insist on anonymous counting with no personal identification, which simplifies the conversation with works councils and privacy officers considerably. Require that the system can both report historically and push live occupancy to other building systems, because the energy use case will arrive whether or not it was in the original brief. And ask for accuracy as a contractual figure, not a brochure figure.

Frequently asked questions

How is workplace utilization calculated? At its simplest, utilization is occupied capacity divided by available capacity over a defined period, for example 42 people in a 120-desk zone during core hours gives 35%. The useful versions separate peak utilization (the busiest hour or day) from average utilization, and measure at the level being decided on, whether desk, room, floor or building. Frequency matters too: a room used for ten minutes every hour and a room used solidly for one hour a day both show low utilization but call for different responses.

If you are weighing a lease decision, a hybrid-working redesign or an energy project and want to know what a sensor layout for your specific floors would look like, talk to the Vemco team about a scoped VemSpace or VemFusion assessment for your building.