What a booking can tell you, and what it cannot

Almost every organisation that introduced desk booking now has a number for how busy its offices are. Very few of them have a number they should act on, because the thing being counted is a reservation, and a reservation is a statement about somebody's intention on the day they made it.

The gap has a shape, and it is predictable.

People book optimistically. They book on Monday for a Thursday they are not certain about, because a desk is free to hold and awkward to find later. They book recurring slots for a pattern that changed two months ago. They book a room for six because six were invited, and four turn up. None of this is bad behaviour; it is the rational response to a system where holding costs nothing and missing out costs something.

The result is a consistent overstatement of demand, and it is not a small one. It is common to find a third of reservations unattended, and the pattern is worst on the days leadership cares most about, because those are the days people hedge.

What the overstatement actually costs.

The first cost is space. A floor that reads as full at eleven on a Wednesday justifies keeping a floor that is half empty. Property decisions run on multi-year commitments, so an error in the occupancy figure is not corrected next quarter; it is paid for until the lease ends.

The second cost is the experience of the people who are in. If the map says the quiet zone is full, they sit somewhere else, or they go to another building, or they stop coming in on that day. They are being displaced by reservations rather than by colleagues, and they have no way of knowing that.

The third cost is credibility. Once people notice that the booking system disagrees with the room they are standing in, they stop trusting it, and then they stop using it, and the data gets worse in a way that is difficult to reverse.

Confirming attendance is the whole fix, and it has to be trivial.

The correction is simple to describe: a booking counts once somebody confirms they are there. Scanning a code on the desk, tapping a tag, or pressing one button in an app all work. What does not work is anything that takes more than a couple of seconds or requires remembering, because compliance falls off a cliff and you end up measuring who is diligent rather than who is present.

The second half is release. A booking that is not confirmed within a grace period should return to the pool automatically, with the person told. That single mechanism does three things at once: it makes the floor plan true, it gives the desk to somebody who wants it, and it teaches people not to over-book without anybody having to send a policy email.

Choose the grace period deliberately.

Too short and you punish a delayed train. Too long and a morning is lost before the desk comes back. Fifteen to thirty minutes suits most offices, but the number should follow the commute and the working pattern rather than a default. Rooms usually want a shorter window than desks, because a room being held is more visible and more resented.

Whatever it is, say it out loud. A release that surprises people reads as the system losing their booking. The same release, announced with a warning message first, reads as the system working.

What to measure once the data is honest.

Utilisation against capacity by day and by floor, which is the number the property conversation needs. The gap between booked and attended, which tells you whether the behaviour is improving. Peak hour, because the estate is sized for the peak rather than the average. And demand by team, because the useful decision is usually about where groups sit rather than about how many desks exist.

One warning: do not report attendance by individual, and be seen not to. The moment people believe a booking system is an attendance register, they will defeat it, and you will lose the estate data as collateral damage. The value is in the aggregate, and the aggregate is available without ever naming anybody.

The uncomfortable part.

Honest occupancy data usually shows less demand than the organisation believed, and the first reaction is often to question the measurement. That reaction is worth taking seriously exactly once: check the confirmation mechanism, check that visitors and meeting rooms are counted the way you think, and check that whole teams have not been excluded.

After that, the number is the number. It is the only version of it that a property decision can safely be made on, and finding out that a floor is not needed is a better outcome than continuing to pay for one on the strength of reservations nobody kept.