The Day Off Was Never the Point

The largest controlled study of the four-day week ran across 141 companies in six countries and reported the headline everyone wanted: people worked one day less, stayed just as productive, and were happier for it. Published in 2025, it is about as close to proof as this field gets.

Buried in the method is the part almost nobody quotes. Before a single worker dropped a single hour, each company spent roughly eight weeks tearing its own week apart, cutting meetings, killing status updates, trimming approval chains, deciding which emails actually needed answering. The shorter week came second. The audit came first.

Hold that order in your head, because it quietly rewrites what these trials are really about.

The results themselves are not in doubt, and it’s worth saying so plainly before poking at them. Britain ran the biggest pilot in 2022: 61 companies, around 2,900 workers. When it ended, 56 of the 61 kept the four-day week; burnout had fallen sharply, revenue held steady, and the number of people quitting dropped by more than half. Iceland went further years ago and never really went back, so that today the large majority of its workforce is on shorter hours with no cut in pay. The 2025 six-country study found the same shape again: output flat, revenue steady, nine in ten companies choosing to carry on. Something real is happening.

But look at what actually produced it, because it isn’t what the slogan implies. Nobody in these trials worked harder or faster in any heroic sense. What happened is older and less flattering, and it has a name: Parkinson’s Law, work expands to fill the time available. Give a decision a sixty-minute meeting slot and it takes sixty minutes. Take the slot away and the same decision gets made in twenty, or in a message nobody has to attend. The extra day came out of slack that existed only because the time was there to fill.

Which turns the four-day week into something more like a confession than a breakthrough. A large slice of the standard week was never output in the first place. It was padding. It was the daily stand-up that could have been three lines of text, the report no one reads, the open-plan visibility, the fast reply sent to prove you were at your desk. A great deal of professional work is the performance of working, presence and responsiveness staged for whoever might be watching. Pull that scaffolding away and the building doesn’t fall, because the scaffolding was never holding it up.

And that is exactly why the four-day week works beautifully for a software team and not at all for a hospital ward.

The usual reading has it backwards. We talk as though frontline work, the nurse, the cashier, the bus driver, is the awkward case, too rigid to modernise. In truth those jobs are the honest ones. A nurse cannot discharge a ward twenty percent faster without someone getting worse care. A bus route is a bus route. A till needs a body behind it during opening hours. There is no slack to cut because the work was already almost pure output, tightly bound to time. Knowledge work can shed a day precisely because it was bloated; frontline work can’t precisely because it wasn’t. So we hand the padded jobs a reward for their padding, and tell the lean ones the perk isn’t for them.

The unfairness sharpens when you notice who those lean jobs belong to. The people who would gain the most from a shorter week, the lower-paid, the physically worn, the ones stitching a job around childcare, are the ones whose work can’t be compressed and who have the least power to demand the only real fix, which is hiring more staff to cover the hours. For them a four-day week isn’t a clever workflow tweak. It’s a bigger payroll and a political decision about who pays for it. Left purely to the market, the four-day week drifts to the people who already have autonomy and comfort, and glides straight past the ones carrying the physical load.

Step back far enough and the deepest thing these trials expose is that the forty-hour week was never engineered around human output at all. It’s an industrial-era inheritance, standardised for factories and then frozen into law and habit, and it survives mostly because hours are an easy thing to count when the real thing you care about, actual output, is hard to measure. So managers count time and call it commitment. The four-day trials “succeed” by finally forcing the process audit that a serious manager should have run anyway. The lesson hiding inside them isn’t “give everyone Friday.” It’s “stop measuring the chair and start measuring the work,” which is harder, which is exactly why the old habit endures.

I don’t want to oversell any of this. The firms that volunteer for these trials are self-selecting, already employee-friendly, already reasonably run; the outfits that couldn’t afford the disruption simply never signed up. The pilots are short, and people behave better when they know they’re being studied and have just been handed a gift. And there’s a new variable scrambling the picture: AI is now compressing exactly the kind of knowledge work these trials relied on, by some estimates five to twenty-five percent in areas like support and coding. That could make a shorter week easier to give, or it could quietly delete the padded jobs that made the four-day week look free. I genuinely don’t know which.

What I’d take from it isn’t a policy to wait for. It’s the audit itself. Look at your own week the way those 141 companies were made to before they earned the day: which meeting is really a message, which report dies unread, which hour is presence rather than product. You may not get Friday off. But you can claw some of it back by cutting your own theatre, and in the process you’ll run straight into the uncomfortable thing the trials actually proved, which is how much of the week was never work to begin with. And maybe spare a thought for the people on the other side of it, whose week is all work and no slack, and who are told that shorter hours can’t be for them for the very reason that they never had any padding to give.