The Four-Day Week Trials Report 90% Success. Almost None Involved Teams That Owe Continuous Coverage.

Ninety-two per cent of the companies in the UK's 2022 four-day week trial kept the policy after the pilot ended. That figure has been repeated in board decks, HR podcasts and LinkedIn posts so often that it now functions as settled fact: cut the week to four days, hold pay steady, and output doesn't suffer. A 2025 study in Nature Human Behaviour, tracking 2,896 employees across 141 companies in six countries, reported much the same pattern. Burnout fell by roughly 71 per cent on the researchers' scale, staff turnover dropped, and 90 per cent of participating organisations chose to continue past the six-month trial period.

What rarely makes the slide is who was actually in the room. Marketing agencies, consultancies and non-profits dominate the case-study libraries built from these trials. Organisations that owe someone continuous coverage, a support desk that has to answer at 2am, an on-call engineer who gets paged when a payment gateway fails, a hospital ward that cannot simply close on Fridays, are barely represented in the data. Where they do show up, the results look different enough that quoting the headline number at them is closer to wishful thinking than evidence.

Who was actually in the trial

The UK's 2022 trial, organised by 4 Day Week Global, the think tank Autonomy and researchers at Cambridge and Boston College, ran 61 organisations and roughly 2,900 employees from June to December that year. Eight of those firms were in marketing and advertising, seven in professional services such as insurance and asset management, and five in the charity and non-profit sector, close to a third of all participants concentrated in three fields that already operate entirely inside a standard business-hours week. Manufacturing and construction, the sectors most likely to run shift patterns, made up just 11 per cent of the employees in the trial combined. One of the more-quoted participants was a fish-and-chip shop in Norfolk, a charming detail, and also a reminder of how far the trial's centre of gravity sat from anything resembling round-the-clock operations.

The 2025 Nature Human Behaviour study didn't publish as granular a sector breakdown, but it shares a structural weakness with every four-day week study run so far: participation is voluntary. Companies that believe they can absorb a shorter week opt in; companies that already know their coverage model makes it impractical don't apply in the first place. That self-selection is well understood by the researchers themselves, who have noted that the evidence base still lacks a randomised controlled trial precisely because every study to date has drawn from firms that chose themselves in.

The coverage math nobody puts in the deck

Compressing a working week and compressing the hours a business needs someone reachable are two different problems, and the four-day week literature mostly measures the first one. Digital Commerce 360 published a breakdown this month of what round-the-clock retail support actually costs to staff: a single desk that needs covering across nights, weekends and shift handovers typically requires close to four times the headcount of one eight-hour shift, and even a lean operation needs five to six full-time staff before order volume is factored in. At a North American average customer-support salary of roughly $55,000, that is upwards of $250,000 a year before overnight-shift premiums, multilingual coverage or turnover replacement costs are added.

Engineering teams run into the same wall from a different angle. Most organisations treat four engineers as the practical floor for a sustainable weekly on-call rotation; fewer than that and someone carries the pager every second or third week, which is a well-documented driver of attrition on its own. That convention was built around a five-day working week, where being reachable outside work hours is already the exception rather than the rule. A four-day week does not shrink the hours that need covering outside those four days; it squeezes the same coverage window into fewer people's rostered working time, which leaves two options: add headcount, at the cost above, or let someone quietly work the day they were told was theirs.

What a four-day week quietly becomes at an always-on company

In practice, most SaaS and operations teams that advertise a four-day week solve this by not really solving it. The job listing says four days; the on-call rota is untouched, so whoever is on call during their nominal day off is still expected to respond. Others move to a compressed 4x10 schedule, four ten-hour days instead of five eight-hour ones, which keeps total hours worked roughly flat rather than cutting them. That distinction matters more than it sounds: the Nature Human Behaviour researchers found that employees whose weekly hours fell by eight or more saw meaningfully larger wellbeing gains than those whose schedules were merely rearranged. A four-day week built by compressing hours rather than cutting them targets the version of the policy with the smallest measured benefit, while still carrying the coverage risk of the full version.

The exceptions are the most honest data point

The NHS is one of the few continuous-coverage employers to seriously propose a four-day, 32-hour week at the same pay, largely as a retention tool: work-life balance has become the second most commonly cited reason NHS staff give for leaving, behind only retirement, and trusts already pay staffing agencies roughly 20 per cent above standard rates to plug shift gaps. The proposal has real backing. It has also drawn blunt pushback from frontline staff, who have pointed out what the marketing-agency case studies never had to confront: a hospital ward does not have a version of "we'll just pause new client work this week." If there are not enough nurses to fill a rota today, cutting the week does not create more nurses. It creates more gaps.

Hospitality faces a blunter version of the same problem before wellbeing even enters the conversation. More than half of UK and US hotels reported active staffing shortages as of early 2026, with turnover in the 70 to 75 per cent range against 3 to 4 per cent in most other sectors. An industry that already cannot fill its existing shifts has no spare capacity to absorb a shorter week without adding heads, and adding heads is precisely the cost problem the four-day week was pitched as avoiding.

Four ways coverage-bound teams are making it work anyway

None of this makes a four-day week impossible for a support desk, an on-call rotation or a hospital ward. It means the handful of continuous-coverage organisations that have made it work did so by rebuilding the schedule, not by copying a marketing agency's memo.

Stagger the day off across the team. A support or ops team split into two or three cohorts, each with a different fixed day off on a rolling basis, keeps the organisation's total coverage window unchanged even though every individual works one fewer day a week.

Pay on-call as its own shift, not a free extension of the week. Teams that treat pager duty as separately compensated time, in cash or banked leave, stop quietly taxing the four-day week's benefit back out of whoever happens to be on the rota that week.

Split coverage by timezone rather than by day. Distributed teams already running follow-the-sun support, where a ticket opened in London is picked up by a colleague in Bengaluru before the UK day starts, get close to round-the-clock coverage without asking anyone to work unsociable hours. A four-day week fits far more easily inside that model than inside a single-timezone team trying to cover the same 24 hours alone.

Set headcount against the coverage line first. Budgeting for the hours that must be covered, nights, weekends, holidays, before deciding how many people work four days rather than five avoids retrofitting a shorter week onto a headcount that was only ever sized for a five-day, no-off-hours model.

The four-day week's 90-plus per cent success rate is real, but it was measured almost entirely on organisations that were never asked to answer the phone at 3am in the first place. NHS trusts are expected to begin their own pilots within the next couple of years, which will be the first dataset of any size testing the policy against a coverage model that does not get to close for the weekend. Until that data exists, treating a marketing agency's results as a preview for a support desk is optimism dressed up as evidence.