Measured attendance is 34 percent. Reported attendance is not.

What the research found

The AWA Hybrid Working Index measures occupancy directly rather than asking about it. The fourth edition covers 55 offices, roughly 68,000 employees, across 19 countries and 35 cities.

Average office attendance came out at 34 percent of the working week. Average desk utilisation, meaning how much of the desk stock was in use at peak, came out at 55 percent.

Set that against survey based studies from the same period, where employees report spending well over half their week in the office, and the gap is not marginal. It is roughly double.

What we see happening

Both numbers are true. They are answers to different questions, and the trouble starts when an organisation treats them as interchangeable.

A survey captures a self-image. People remember the busy weeks more vividly than the quiet ones, they count travel days and client visits as office time, and they answer in the direction they believe is expected. A sensor captures a fact. Neither is dishonest. But only one of them belongs in a decision about how many desks to keep.

What we increasingly see is a third number in the mix, which is the booking. Booking data feels objective because it comes out of a system, and that is exactly why it misleads. A reservation records what someone planned on Thursday for the following Tuesday. It does not record the cancelled meeting, the sick day, or the four hours the desk stood empty while its user was elsewhere in the building.

Our expectation for the next two years is that workplace teams will be asked, routinely, to state the provenance of their occupancy number. Not just the figure, but where it came from and what it counts. That is already normal in energy reporting and it is arriving in space reporting. Teams that cannot answer will find their recommendations discounted, not because the recommendations are wrong, but because they cannot be verified.

There is also a management practice worth borrowing from finance here. Nobody presents a revenue figure without stating whether it is booked, invoiced or collected. Those are three legitimate numbers describing the same business, and confusing them is a firing offence. Workplace reporting has no equivalent discipline yet, which is why the same slide can carry a survey figure, a booking figure and a sensor figure without anyone noticing they are not comparable.

The organisations that fix this do something simple. They label every occupancy number with its source before it leaves the workplace team, and they never let a survey figure appear in a capacity decision. It costs nothing and it changes how seriously the whole function is taken.

What this means for how you run a workplace

The fix is not to stop surveying. Surveys tell you why people behave the way they do, which sensors never will. The fix is to stop using survey answers where measurement is required.

iotspot Insights sits underneath your existing iotspot booking data and adds the measured layer: actual use per desk, zone, floor and hour. That gives you three numbers instead of one, and the ability to say which is which. What people intended, what people did, and what people say. The distance between them is usually the most interesting finding in the whole dataset.

For HR and workplace teams, the practical payoff is a hybrid policy that can be evaluated. Which days fill, which teams co-locate, whether anchor days are the days people actually choose. All anonymised and aggregated, so it answers questions about the workplace without ever being about a person.

The honest caveat

AWA's index is a strong dataset because it measures and states its sample, but it skews toward organisations that already invested in occupancy measurement, which are not a random slice of the market. Sector matters too: a client facing consultancy and a claims processing operation will sit far apart on the same chart. Use 34 percent as a reference point, not as the number your own building should produce.