For most of the twentieth century the office was a container. The work was there — the files, the machines, the phone, the people who knew things. You went to the office because that is where the work physically was.
That stopped being true, and most thinking about offices has not caught up. The building is now a coordination device: a mechanism for putting particular people in the same place at the same time. It is useful for what coordination is useful for, and expensive for everything else.
For an applied reference, see Monitask’s how Microsoft Teams tracks activity overview.
What follows from the reframe
If the office is a container, attendance is the point and the question is how many days.
If the office is a coordination device, attendance is a means and the question is what happens when people are there.
Almost every unproductive argument about office policy comes from the two sides using different definitions without noticing.
What coordination is genuinely good for
Work that requires many people to converge quickly. Workshops, planning, working through something contested. High-bandwidth, iterative, hard to do well asynchronously.
Establishing relationships between people who do not have them. New joiners, teams that have merged, cross-functional groups who only meet through tickets. Weak ties form far more readily in person and are difficult to create deliberately at a distance.
Repair. Difficult conversations, conflict, anything where tone carries most of the meaning.
Apprenticeship. Watching someone work is how a great deal of judgment transfers, and it is one of the things distance genuinely degrades. See how learning at work actually happens.
Anything requiring the place itself — equipment, labs, physical materials, secure environments.
What it is bad for
Concentrated individual work. An open floor is a worse environment for it than most homes. See what open plan actually did.
Video calls. A person who commutes to sit in a room on a video call has paid the cost of coordination and received none of the benefit. This is the single most common failure of hybrid policy, and it is visible from any office on a Wednesday.
Being seen. Presence as a proxy for effort is a measurement failure, not a use of the building.
The practical implication
Coordinate the days.
If people are in on different days, they have paid the commute and received the office's least valuable property — the desk — while missing its only valuable one, which is other people.
This is not a small effect. A hybrid policy of "three days, your choice" produces an office that is moderately full every day and rarely contains the specific combination of people any given person needed. A policy of "Tuesday and Wednesday, everyone in this group" produces two days of genuine coordination and three days of quiet.
The second costs the same and delivers something. Most organisations run the first because it is easier to announce.
Design for it
If the office is a coordination device, the floor should look different from an office designed as a container.
More space for groups, less for individuals. The desk is the least valuable thing there.
Rooms sized for real meetings, and enough of them. On coordination days the constraint is always rooms.
Genuinely enclosed space for the calls people cannot avoid. Not two booths for eighty people.
Somewhere to be that is not a desk. Most of the value of a coordination day happens in the unstructured time around the scheduled part, and that requires somewhere to stand and talk that is not in front of somebody working.
Acoustic treatment, which is unglamorous and is the highest-return spend on most existing floors.
Measure the right thing
Occupancy is the metric most organisations have, and it measures the container.
Better questions, if the office is a coordination device:
On the days people came in, how much of their time was spent with other people? If the answer is mostly video calls, the policy is not working regardless of occupancy.
Did the people who needed to be together actually overlap? Coordination is about specific combinations, not headcount.
What happened that would not have happened otherwise? Hard to measure and the only question that matters.
Room availability on peak days. If people cannot find a room to collaborate in, the building is failing at its one job.
What this does not settle
It does not tell you how many days. That depends on the work, the team, the stage, and how much of the job is genuinely coordination-dependent — and reasonable organisations reach different answers.
What it does settle is the framing. A policy expressed as a number of days, with no statement of what those days are for, is a container policy. It will be experienced as an attendance requirement, because that is what it is, and it will produce attendance rather than coordination.
The question worth asking before setting a policy
What specifically do we want to happen when people are in the same room, and does our current arrangement make that likely?
If the honest answer is that people come in and do the same work they would have done at home, in a noisier environment, then the policy is producing cost and the building is being used as a container for work that is no longer kept there.
For broader public guidance and background, consult the National Institute for Occupational Safety and Health.