The Four-Day Week: What the Trials Measured, and What They Did Not

The four-day week has moved from advocacy to evidence, and the evidence is genuinely more favourable than sceptics expected. It also has a design limitation that almost no coverage mentions, and understanding it is the difference between running a useful pilot and importing someone else's result.

For a product-oriented comparison of productivity vs efficiency, see Monitask.

Comparative labour-market research is available through the OECD employment resources.

What the studies found

The largest controlled study. Published in Nature Human Behaviour in 2025 by Fan, Schor, Kelly and Gu, it tracked 2,896 employees across 141 companies in six countries — Australia, Canada, Ireland, New Zealand, the United Kingdom and the United States — through a six-month trial of reduced hours with no reduction in pay. Researchers tracked stress, burnout, health and work-life balance rather than raw output alone. Across those measures worker wellbeing improved while productivity held steady or rose.

The UK pilot. 61 companies over six months. Revenue stayed broadly consistent, up 1.4% on average, while staff turnover dropped 57% and burnout fell 71%. When the trial ended, 92% of companies chose to continue. It ran between June and December 2022, was the largest trial to that date, and was the first to include in-depth interview research, conducted by a Cambridge team.

Iceland. The earliest large-scale trials ran from 2015 to 2019, putting 2,500 public sector employees on 35–36 hour weeks with no pay cut. Productivity stayed the same or improved, and wellbeing improved across stress, burnout, health and work-life balance. By 2026 roughly 86% of Iceland's workforce has access to shorter hours or is actively working them.

That is a consistent pattern across different countries, sectors and research teams. It is not nothing.

The limitation the coverage skips

The companies volunteered.

Every one of these trials recruited organisations that wanted to try a four-day week, believed it might work for them, and were willing to commit six months to it. That is not a random sample of employers — it is a sample of employers predisposed to succeed at exactly this.

The consequences:

Selection at the organisation level. Firms with obvious structural barriers — thin staffing, rigid coverage requirements, customer-facing hours they cannot compress — mostly did not apply.

Selection at the sector level. Participants skewed toward knowledge work, professional services and smaller organisations. Manufacturing, healthcare, retail and continuous-coverage operations are underrepresented, and those are the settings where the operational problem is hardest.

Committed leadership. A CEO who applied to a trial is not the same as a CEO who was told to implement one.

Some productivity measures are self-reported. The wellbeing findings are measured directly and are strong. Output measures vary in rigour between studies, and self-assessed productivity is a weaker instrument than a revenue line.

Short duration. Six months captures the enthusiasm of a new arrangement. It does not capture year three.

None of this means the findings are wrong. It means the honest statement is: among organisations motivated and able to redesign how they work, a four-day week held output steady and improved wellbeing. That is a different claim from "a four-day week works," and it is the claim the evidence supports.

The distinction that decides the outcome

The single most useful finding, and the one most often collapsed in summaries.

The 100-80-100 model means 100% pay, 80% of typical hours, and 100% output — usually 32 hours across four days. A compressed schedule of four ten-hour days keeps the same 40 hours in fewer days. Research consistently shows that compression does not reduce stress or burnout, because the total workload has not changed. The wellbeing and retention gains come specifically from working fewer hours, not from working fewer days.

If you implement a compressed week and expect the trial results, you will not get them. You will get the same workload in longer days, which is a different intervention with different effects — including on childcare, commuting and fatigue.

Why it works where it works

The mechanism in the successful cases is not that people worked harder in less time. It is that the reduction forced a redesign that was overdue anyway.

In the Nature study, every participating company spent roughly eight weeks preparing before the trial began. That preparation is where most of the effect comes from:

• Meetings cut, shortened, or converted to written updates
• Standing meetings with no decision output eliminated
• Interruption norms established
• Low-value work stopped rather than compressed
• Decision rights clarified so fewer people had to be present

Much of this is available without changing the working week at all. An organisation that cannot cut a meeting is unlikely to succeed at cutting a day.

Where it fails

The second failure point is operational unpreparedness. Teams that simply drop a day without restructuring meetings, workflows or collaboration norms end up cramming the same obligations into fewer hours — with more fatigue and worse outcomes than before. The preparation phase is nonnegotiable.

Other predictable failure modes:

Coverage requirements. Where customers, patients or production require presence five or seven days, a shorter week is a staffing question, not a scheduling one — and the honest answer may be that it costs money.

Work spilling onto the fifth day. People quietly working Fridays defeats the entire mechanism and produces unrecorded hours. For non-exempt staff that is also a wage and hour problem. See off-the-clock work.

Uneven application. Some functions get it, others cannot. This is manageable but must be addressed explicitly rather than left to resentment.

Reduced hours with unchanged targets. The fastest route to burnout, achieved by announcing the benefit without doing the redesign.

The wage and hour dimension

For US employers, two points that the international trial literature does not cover.

Non-exempt employees. Overtime obligations are based on hours in a workweek, not days. A 32-hour week has no overtime implications; a compressed 4×10 schedule may, depending on the state. California pays overtime after eight hours in a day, which makes 4×10 an expensive arrangement there unless an alternative workweek schedule is properly adopted under state procedure.

Exempt employees. Reducing the working week does not permit reducing the salary below the applicable threshold. And a salary reduction tied to hours can undermine the salary basis on which the exemption depends.

Take advice before implementing either.

If you are going to pilot one

A 90-day pilot gives enough data to decide whether to extend; a full year captures seasonality. The minimum viable pilot is 12 weeks with weekly pulse surveys and clear baseline metrics established before the trial begins.

Pilot first — a six-month trial in one department gives you real data without betting the company on a hypothesis.

Practical additions:

Establish the baseline before you announce it. Output, quality, response times, turnover, absence, overtime. Measuring after the announcement measures the announcement.

Do the preparation phase properly. Eight weeks, spent on meetings, workflows and decision rights. If the organisation will not commit to it, do not run the pilot — you will learn nothing except that unprepared teams struggle.

Define output per role in advance. See measuring knowledge work.

Watch for spill. Check system access logs on the fifth day. If people are working, the pilot has failed regardless of what the survey says.

Decide the failure criteria before you start, so that ending the trial is a decision rather than a defeat.

The honest summary

The evidence supports a narrower claim than the headlines: organisations willing and able to redesign how they work can reduce hours without reducing output, and the wellbeing and retention gains are substantial and consistently measured.

Whether your organisation is one of them is a question the trials cannot answer — and the preparation phase, which is where the effect actually comes from, is worth doing whether or not you ever change the week.