How Often Do People Actually Need Workspace? 19,759 Employee-Months of Data

Almost every way of buying workspace assumes a frequency. A lease assumes people come most days. A per-seat coworking membership assumes the seat gets used enough to beat a day rate. A monthly stipend assumes a monthly rhythm. These assumptions are rarely tested against what employees do, because most companies cannot see what employees do. Workspace bought through expense reports produces receipts, not usage data.

We can see it. Across Croissant's company accounts, 3,197 employees have generated 19,759 active months of workspace usage. The distribution is not close to what per-seat pricing assumes.

In a month when someone books at all, the median is two days

Three quarters of active employee-months, 76.7%, are four days or fewer. Fewer than one in eleven reach nine days, which is roughly two days a week. The median is two days; the mean is 3.6, pulled up by the small heavy-use tail.

And that is measured only across months when someone booked. Zoom out to the year and the picture is starker still. The median employee books five days across an entire year, spread over two active months. 65.3% of employee-years involve activity in three months or fewer. Only 7.4% involve activity in nine or more months.

Why the "everyone needs a desk" model breaks

The usual objection is that low usage reflects a weak benefit rather than real demand, that people would come more if the offering were better. The rest of the data argues otherwise, because the usage that does happen is highly structured rather than casual.

It is a midweek behavior, almost exclusively

Tuesday through Thursday accounts for 64.4% of all bookings, and the weekend is a rounding error. Employees are not wandering into workspace when it suits them; they are concentrating on the days their colleagues are likely to be there. Low frequency and high coordination are happening at the same time, which is exactly the pattern a fixed seat serves worst, because the seat sits empty on the days nobody chose and is oversubscribed on the days everybody did.

Most of the demand is about other people

Only 15.1% of logged hours are solo focus work, the use case a desk is designed for. The majority is collaboration, and another quarter is travel, which is inherently a different city each time. Two thirds of workspace demand is therefore tied to either being near colleagues or being away from home, and neither is served by assigning someone a permanent desk in one building.

A booked day is not a full day

Session length compounds the gap. The mean logged session is 4.3 hours and the median 3.9, roughly half a working day.

Half of all sessions are under four hours. A quarter are under two: someone dropping in for a meeting or a couple of hours between commitments. Only 17.8% look like a conventional full day at a desk.

What to buy instead

Match the purchase to the observed distribution rather than to the org chart:

The last row matters: for the heavy tail, fixed space really can win. The mistake is buying the model that suits 8.9% of employee-months for the other 91.1%. A shared pool sized to actual consumption serves both, because the light users stop subsidizing capacity they never touch and the heavy users are not rationed.

Three things to do with this

Measure days per person per month, not headcount. If your workspace program's core metric is how many employees are enrolled, you cannot tell an underused program from a well-fitted one. Days booked per active employee per month is the number that moves.

Budget on the median, plan capacity on the cluster.Spend follows the median employee (two days a month). Capacity has to survive Tuesday-to-Thursday, when 64.4% of demand lands. These are different problems and using one number for both is why programs feel simultaneously expensive and undersupplied.

Buy meeting rooms deliberately. Collaboration is 55.7% of logged hours, but meeting rooms are a small fraction of bookings, which means collaboration is largely happening in open coworking space, by default rather than by design. Rooms are the cheapest available upgrade to the majority use case.

Methodology and limits

Croissant has booked more than 4 million workspace hours since 2015: 611,226 booked sessions, each counted as a standard seven-hour workday. Logged attendance is measured separately and is lower, a mean of 3.8 hours across all visits, because booked hours describe reserved capacity while logged hours describe time present.

The frequency figures cover company-attributed usage: 3,197 distinct employees, 19,759 employee-months, 5,893 employee-years, and 71,314 sessions with a measured duration. An "active month" is a calendar month in which an employee booked at least once; months with no bookings are excluded from the frequency distribution, so the real average across all enrolled employees is lower than the figures above. Bookings include coworking day visits, guest visits, meeting room bookings, and private office bookings; cancelled bookings and sessions longer than 24 hours are excluded.

Use-case and day-of-week splits come from Croissant's weekly company usage snapshots and cover company-attributed usage only. Use case is self-reported at booking, so it reflects employees' stated reason rather than an observed one. Finally, this measures workspace booked through Croissant: employees who also have a company office available to them may be using it on the days they do not appear here, which means these figures describe demand for flexible workspace rather than total workplace attendance.