Your policy says 32 hours. Your people are working some other number. The gap between the two is the most important measurement in a 4-day pilot, because it decides whether your results are real — and whether the people producing them will still be with you next year.
Policy hours, rostered hours, actual hours
Three numbers get confused in pilot reporting. Policy hours are what the contract now says. Rostered hours are what the schedule shows. Actual hours are what human beings worked, including the early logins, the lunch-hour inbox checks and the Friday-morning “just finishing this.” Only the third number connects to wellbeing, and only the third tells you whether output held because work was redesigned or because the working week quietly refused to shrink.
The UK 2022 pilot illustrates the point usefully. Contracted time fell by a fifth; measured average time fell from roughly 38 to roughly 34 hours a week. That is still a substantial, health-relevant reduction, and the wellbeing results — lower burnout, less stress, fewer sick days — tracked it. But a firm that read only its own policy would have overstated the change by six hours a week. In your pilot, the realised figure is the honest one to report to the board and to staff.
Where hidden overtime hides
Start with the day off itself: logins, sent messages, edited documents, aggregated at team level. A healthy pilot shows Friday activity falling sharply after the first fortnight and staying low. A struggling one shows a sawtooth — quiet in the honeymoon weeks, creeping back as unreformed processes reassert themselves.
Then look at the edges of the working days. Earlier starts, later finishes and skipped lunches in the four remaining days are the classic compensation pattern. After-hours messages are the easiest signal to count and the hardest to explain away: if the volume of evening and weekend communication did not fall when hours did, the week did not really shrink — it spread out.
Overtime claims, where your workforce records them, are gold. Rising recorded overtime alongside falling contracted hours is intensification with a paper trail, and it is far better to read it in a dashboard than in a grievance.
Tripwires, not surveillance
The purpose of counting is redesign, not discipline, and the design of your measurement has to prove that. Report at team level, never as named individual lists. Agree with staff representatives what will be counted and why before you count it. And attach a tripwire that helps people: for example, if a team’s day-off activity or after-hours volume sits above the agreed threshold for two consecutive weeks, the response is a workload review with authority to stop or reassign work — not an email telling staff to try harder at resting.
Managers need the same protection in the data. If team leaders’ own hours are invisible, their Friday working becomes the organisation’s unspoken standard. Include leadership activity, in aggregate, in the same report. Nothing builds trust in the count like the senior team publishing its own numbers first.
A monthly rhythm that takes twenty minutes
Once a month, put three charts in front of the pilot group: actual versus contracted hours by team; day-off and after-hours activity; and the intensity pulse from your staff survey — can people complete their work within their hours at a sustainable pace? Where hours are genuinely down and pace is sustainable, celebrate precisely. Where they are not, leave the meeting with one named piece of work stopped, moved or re-staffed. Measurement that never changes a decision is decoration, and staff learn to ignore it within a quarter.
Practical takeaways
- Report actual hours worked, not the contract figure — the realised reduction is the one linked to wellbeing.
- Count day-off activity, day-edge creep and after-hours messages at team level, never to name individuals.
- Attach a tripwire: sustained hidden overtime triggers a workload review with power to stop work.
- Include leaders’ own after-hours activity in the same report, and publish it first.
- Review three charts monthly and leave with one concrete workload change for any team over threshold.
Sources
- Autonomy, University of Cambridge and Boston College, UK four-day week pilot results, reported February 2023 — https://www.sciencedaily.com/releases/2023/02/230221113132.htm
- Autonomy, UK pilot one-year follow-up, reported February 2024 — https://www.hrdconnect.com/2024/02/26/one-year-on-uks-4-day-work-week-trial-has-been-a-success/
- Portugal four-day week pilot, Birkbeck and Henley Business School, reported by Personnel Today, 14 December 2023 — https://www.personneltoday.com/hr/portugals-four-day-week-reduces-anxiety-and-fatigue/
Day5Group is a consultancy that assists in the transition to a 4 day work week. More about Day5Group · All blog articles