The Four-Day Workweek Experiments: What Actually Happened

Real pilots have tested three different arrangements — fewer hours at full pay, the same hours squeezed into four days, or fewer hours for less pay — and the results get lumped into one headline. Here is what each trial actually measured, where it fell apart, and what would have to be true at your company for any of it to transfer.

For three years, a specific claim has circulated through founder group chats and startup Twitter: a four-day week that raised productivity, held pay steady, and made employees measurably less burnt out, all at once. It is the kind of result that sounds too convenient to be real, which is exactly why it is worth checking with a reporter's scepticism instead of a marketer's enthusiasm.

The honest answer is messier than the pull-quote version, and more useful to an actual founder. Several well-documented trials really did happen, with real payroll records and real output measurements, at scales ranging from a single 240-person New Zealand trust company to more than 2,500 Icelandic public-sector workers to 61 companies across the UK. Some of what they found holds up under scrutiny. Some of it does not generalise past the companies that volunteered for it. And at least one large, well-funded four-day experiment — a US state government's statewide switch for its own workforce — was reversed after three years, for reasons worth understanding before anyone else repeats them.

Three things get called a "four-day week," and they are not the same bet

Before comparing outcomes across trials, it matters which of three distinct arrangements each one actually tested. Casual coverage collapses them into one idea, but they make very different promises to employees and carry very different risk for an employer.

Nearly every headline result you have heard — higher output, lower burnout, better retention — comes from Model 1. Nearly every widely reported failure comes from attempts at Model 2 done at large, mandatory scale, or from Model 1 pilots run in the wrong kind of workplace. That distinction does most of the explanatory work in what follows.

What the well-documented Model 1 trials actually measured

Iceland, 2015–2019. Reykjavík City Council began trialling shorter hours with no pay cut in 2015; the Icelandic national government followed in 2017. Combined, the trials eventually covered more than 2,500 workers — a genuinely large sample, and notable because it was overwhelmingly public-sector and included plenty of jobs with rigid schedules: hospitals, preschools, police stations, not just office staff. Researchers Gudmundur D. Haraldsson and Jack Kellam, publishing through the UK think tank Autonomy and the Icelandic association Alda, tracked the results in a 2021 report titled "Going Public." They found productivity and service levels held steady or improved across most participating workplaces, and self-reported stress and burnout fell. By their count, 86% of Iceland's entire working population had, by 2021, either moved to shorter hours or won the contractual right to. That is a rare thing: a shorter-hours result that reached genuine national scale over several years, not a single press cycle.

The honest caveat: Iceland is a small, highly unionised country of roughly 370,000 people, and the trial workplaces were converted individually, with dedicated implementation support, over years — not flipped overnight. What worked there was as much about careful, funded change management as it was about the shorter week itself.

New Zealand, Perpetual Guardian, 2018. The estate-planning and trust company, under founder Andrew Barnes, ran an eight-week trial of 100-80-100 with its roughly 240 employees. University of Auckland researcher Helen Delaney and AUT's Jarrod Haar conducted an independent academic evaluation: self-reported stress fell from 45% to 38%, and satisfaction with work-life balance rose from 54% to 78%. The company made the policy permanent, and Barnes went on to co-found 4 Day Week Global to help other employers run their own version. This is the trial that effectively started the current wave — but it is one company, in one sector, with output monitored informally by managers rather than through a rigorous, independently audited productivity metric.

The UK, 2022 — the largest trial to date. Coordinated by 4 Day Week Global, the UK think tank Autonomy, and the 4 Day Week UK Campaign, with academic input from researchers at Cambridge, Boston College, and Oxford, this six-month pilot ran from June to December 2022 with 61 companies and roughly 2,900 employees, almost all on the 100-80-100 model. The February 2023 report, "The Results Are In," found that 56 of the 61 companies continued the four-day week once the trial ended, with 18 confirming it as a permanent policy outright. Company revenue over the trial period came in roughly flat to slightly up compared with the equivalent period a year earlier. Resignation rates fell, sick days fell, and self-reported burnout and anxiety dropped — with the largest improvements concentrated among the employees who reported the worst symptoms going in.

That selection effect is not a minor footnote. Autonomy and 4 Day Week Global say so themselves in their own reporting: companies that sign up for a public four-day-week pilot are, by definition, already receptive to the idea, often already trimming meeting load and process waste before the clock officially starts. That does not make the results fake. It does mean a founder should read "56 of 61 kept it" as "56 of 61 motivated, self-selected employers kept it" — a different, narrower claim.

US and Ireland, 2022. A parallel cohort of 33 companies and about 903 employees ran the same six-month, 100-80-100 model, with academic evaluation led by Boston College sociologist Juliet Schor, a longtime researcher of shorter-hours policy and author of the 2024 book Four Days a Week. Reported results: average revenue across the group rose roughly 8% versus the same period the prior year, though individual company results varied widely; resignations and sick days fell; self-reported burnout dropped substantially. Same caveat as the UK cohort — opted-in employers, not a random sample of businesses.

Microsoft Japan, 2019 — the outlier worth flagging clearly. Every Friday in August 2019, Microsoft's Japan office closed, affecting about 2,300 employees, in what the company called its "Work-Life Choice Challenge." Microsoft Japan's own published results: sales per employee up roughly 40% versus August 2018, electricity use down 23%, and pages printed down 59%. This is the most-cited statistic in every trend piece about the four-day week, and it deserves the most scrutiny: it is one company, one month, self-reported by the company itself with no independent academic evaluation, measured during a traditionally slow August in Japan with an unclear amount of control for seasonal effects. It is a real data point. It is not proof of anything at the scale people usually cite it for.

Where it fell apart: Utah's compressed week

Not every well-resourced four-day experiment is a success story quietly waiting to be discovered, and the clearest counterexample is a government, not a startup. In 2008, under Governor Jon Huntsman Jr., Utah became the first US state to move nearly its entire executive-branch workforce — about 18,000 employees — onto a mandatory four-day, 10-hour compressed schedule, Monday through Thursday. This was Model 2, not Model 1: total hours did not drop, they were compressed, and the goal was never employee wellbeing measurement — it was cutting energy costs by closing state buildings on Fridays.

The policy ran for three years. Reported results were modest and mixed: the state's own review found energy savings well below the multimillion-dollar figure originally projected, and the visible cost to the public was government offices being closed on the fifth business day, which drew complaints from residents and businesses that needed services on Fridays. In 2011, Governor Gary Herbert reversed the policy and returned most state offices to a standard five-day week.

Utah's experiment never tested whether four days of work could produce five days of output — nobody's hours were cut, so there was no productivity claim to prove or disprove. What it tested was whether the public was willing to trade Friday office access for whatever savings a compressed calendar produced, and the answer, over three years, was no. It is the four-day week's most visible large-scale reversal, and it is a compressed-hours story, not a reduced-hours one — worth remembering before assuming every four-day pilot is testing the same thing Iceland or the UK tested.

What actually separated the pilots that stuck from the ones that didn't

Reading across Iceland, New Zealand, the UK, and Utah together, a few conditions repeat on the side of the trials that held up — and their absence shows up consistently on the side of the ones that struggled or reversed.

They cut the actual work before cutting the calendar. Companies in the UK and NZ pilots commonly ran a meeting audit — trimming standing meetings and low-value process — before flipping to four days, rather than assuming the fifth day's work would simply vanish. Utah did the opposite: it compressed the calendar without first reducing the total volume of work or public-facing hours government offices needed to cover.

They had a coverage plan for anything client-facing. The pilots that struggled hardest, in the UK cohort and elsewhere, were disproportionately roles where someone has to be reachable on a fixed schedule. Trials that pre-planned staggered days off, or accepted higher headcount to preserve coverage, held up better than ones that assumed Friday would sort itself out.

They set an output metric before starting, not after. Iceland's multi-year rollout and the UK's academically monitored pilot both had baseline productivity and service-level measures defined before the trial began. Microsoft Japan's one-month test, by contrast, has no comparable baseline discipline — which is exactly why its headline number travels so much further than its evidentiary weight should allow.

They gave it a fair trial window. Six months, the length used in both the UK and US/Ireland cohorts, was long enough to get past the novelty period and see whether output actually held. One month is not.

They stayed willing to reverse it. The most credible large-scale example in this article, ironically, is the one that ended: Utah built in the option to review and reverse, and used it when the numbers said to. That is a healthier posture than announcing a four-day week as a permanent, unconditional company value on day one.

The honest gut-check before piloting this for real

None of the above is an argument against trying a shorter week. It is an argument against treating it as a hiring perk you can bolt on without doing the same homework the actual trials did. A few questions are worth answering honestly first.

Is the real bottleneck hours, or unmanaged process? If a team is drowning in status meetings and reflexive Slack threads, a fourth day off treats a symptom without touching the cause — cut the meeting load and message noise first, and you may find the team barely needed the extra day off to begin with, or that cutting a day becomes much easier once the real time sink is gone.

Do you have real coverage for anything client-facing? If Friday just means "nobody available," you have not built the four-day week the successful trials built — you have built the version that made Utah's residents complain.

Can you actually measure output right now? Companies in the UK and US/Ireland cohorts that already tracked output — tickets closed, revenue booked, deals closed — could tell within six months whether the pilot worked. A team with no baseline metric today will have no way to know after, either.

Is your team small enough to reverse this fast if it doesn't work? Under roughly 20 people, a schedule change is genuinely reversible within a month with minimal cost to morale. At a few hundred people, you are closer to Utah's position: a public policy that is hard to walk back without it reading as a retreat.

None of this makes the four-day week a bad idea for a growing team. It means it is not a costume you put on for the careers page — it is an operational bet, tested by real employers who kept careful enough records to know when it stopped working. Treat it with the same rigor they did, and the decision gets much easier to make honestly.