Hybrid offices are commonly sized by taking average daily attendance and providing desks to match. The average is arithmetically correct and operationally misleading, because attendance is not evenly distributed.
Attendance clusters rather than spreads
Given a choice of days, employees converge. Midweek fills because people come in to see colleagues, and coming in on an empty day defeats the purpose of coming in.
This is self-reinforcing. Once a day is known to be busy, it attracts more attendance, and the quiet days get quieter.
An office sized for the mean therefore overflows on peak days and stands largely empty on the others, at the same total occupancy.
Overflow is worse than it looks on a spreadsheet
Running out of desks does not merely inconvenience the last arrivals. It teaches everyone that attendance is a gamble, which changes the calculation for the next week.
Employees who cannot rely on a workspace start arriving earlier, booking speculatively, or deciding the trip is not worth it.
Speculative booking then corrupts the utilisation data, because reservations no longer indicate intent to attend, and the next capacity decision is made from worse information.
Meeting rooms constrain earlier than desks do
People come to the office largely to meet, so peak days generate demand for rooms disproportionate to headcount present.
Room shortage pushes meetings back onto video from inside the building, which is the outcome the office was meant to avoid.
Many hybrid offices are consequently mis-specified in composition as well as size, holding too many individual desks and too few enclosed spaces.
Assigned days shift the problem rather than solve it
Mandating particular days per team smooths the peak on paper. It also breaks the cross-team encounters that justified the office, since teams assigned different days never overlap.
The teams most likely to need each other are often the ones most likely to be scheduled apart, because scheduling is usually done to balance numbers rather than to match collaboration patterns.
Any assignment scheme is also a commitment that becomes hard to change once employees have arranged childcare and commutes around it.
Measuring peaks costs little and changes decisions
Badge or network data already records who was present on which day, and the peak-day figure is as easy to compute as the mean.
Sizing to something near the peak, rather than the average, produces a larger and less efficient-looking office that people can actually use.
The alternative is an office that reads as well-utilised in a report and functions as unreliable in practice, which is how many hybrid programmes lose support from the people they were built for.