The commute came back

Card-swipe data from 1,240 office buildings across nine cities shows the same shape in every one: a collapse in 2020, a long flat trough, and then eleven consecutive quarters of increase that has now taken average weekday occupancy to 81 per cent of its 2019 level.

The obvious reading is that remote work was a phase and the mandate won. The data does not support that, and the reason is a composition effect that almost every write-up of these numbers has missed.

Who is in the building

The occupancy recovery is driven overwhelmingly by roles that were never remote-capable — facilities, security, reception, hospitality, lab and hardware work, and the enormous services layer that exists to support an occupied building. As offices reopened, those jobs came back at 100 per cent, because they were always at 100 per cent when the lights were on.

Filter to desk-based roles only and the picture inverts. Average attendance among people who could plausibly do their job from a laptop is 2.7 days a week and has moved by less than a third of a day since early 2024. The mandate did not win. It negotiated, and it settled around three days.

Two very different populations are being averaged into one number, and only one of them ever had a choice.

What three days does to a city

A three-day week is not a smaller five-day week. It is a different city. Tuesday to Thursday now carries traffic and footfall that exceed 2019 peaks, while Monday and Friday sit twenty points below. Every business calibrated for a flat five-day rhythm — canteens, transit, cleaning contracts, the sandwich shop on the corner — is now running two-thirds capacity on a five-day cost base.

That is the actual policy problem, and it is not going to be solved by anybody’s return-to-office memo. The peaks are the point. A city that fills up three days a week needs infrastructure sized for the peak and priced for the average, and nobody has worked out how to pay for that.