Most collaboration never reaches a calendar

What the research found

Microsoft's 2025 Work Trend Index special report combines a survey of 31,000 knowledge workers across 31 countries with anonymised usage telemetry.

Fifty seven percent of meetings are ad hoc, arranged without a calendar invitation. Employees are interrupted roughly every two minutes during core hours, which works out at around 275 interruptions in a day. Thirty percent of meetings now span time zones.

What we see happening

Every workplace planning model in common use is built on the calendar. Room ratios, desk ratios, peak day forecasting, catering, cleaning schedules: all of it derives from what people booked.

If more than half of collaboration never enters that system, then more than half of the demand your building serves is invisible to the tools you plan with. That is not a small measurement error. It is a structural blind spot, and it explains a pattern almost every facility manager will recognise. Meeting rooms show as fully booked while corridors, kitchens and empty desks fill with the conversations that could not find a room.

We expect this to be the defining workplace measurement problem of the next few years, and we expect the answer to come from the building rather than from the calendar. You cannot ask people to book the unbookable. A three minute conversation that happens because two people passed each other is the single most valuable thing an office produces, and the moment you require it to be scheduled, it stops happening.

So the instrument has to change. Instead of counting reservations, count presence. Instead of asking what was planned, observe what occurred. The organisations getting this right are not tightening booking policy. They are measuring occupancy directly and designing for the traffic they find.

The space type shift visible across the industry follows the same logic. Demand is moving toward phone booths, focus rooms and small meeting rooms precisely because that is where unscheduled collaboration lands.

The interruption figure points somewhere else, and it is worth holding both findings at once. If people are interrupted roughly every two minutes during core hours, then the same building has to serve two incompatible needs on the same floor: frictionless spontaneous contact, and genuine protection from it. Most offices are designed for one and accidentally deliver neither. The open floor makes interruption easy and concentration impossible, while the bookable meeting rooms sit empty because a two minute question does not justify a reservation.

The buildings that resolve this are not the ones with the most collaboration space. They are the ones where the distance between a desk and a door that closes is short enough that people actually use it. That is a floor plan question, and answering it requires knowing where unscheduled demand is landing rather than where you hoped it would.

What this means for how you run a workplace

Two moves make ad hoc collaboration visible and then usable.

First, measure it. Room sensors and people count sensors feeding iotspot Insights record what actually happened inside a space, including the meetings nobody booked. That turns "we need more meeting rooms" into a specific question about which type, on which floor, at which hour.

Second, make spontaneous use frictionless. Meeting room displays and NFC points let someone claim a room by tapping it as they walk in, which captures the booking without asking anyone to plan ahead. iotspot Room Booking then releases rooms automatically when a reservation goes unused, so the ad hoc half of the week can reclaim the space the scheduled half did not need.

And for the collaboration that is worth planning, iotspot Team and Co-working gets the right people in on the same day, which is what turns a corridor conversation into something more than a coincidence.

The honest caveat

Microsoft's telemetry component is drawn from its own product usage and excludes EU tenants, so the interruption figures reflect one software ecosystem rather than all working life. The survey component is self-reported. What the study is good for is establishing that a large share of collaboration is unscheduled, which matches what occupancy sensors show independently. Treat the 57 percent as a strong signal and your own measured data as the number.