Every honest pilot carries the possibility of a no. That is what makes it a pilot rather than an announcement. The uncomfortable truth from the published record is that the organisations which damaged themselves were rarely the ones that stopped. They were the ones that stopped badly: quietly, late, or with an explanation nobody believed.
Ending a pilot well is a design task, and most of the work happens before the pilot begins.
Why rollback hurts more than expected
A day off is not like a cancelled software subscription. Within weeks, staff build lives around it: childcare arrangements end, a course starts, a caring rota with siblings is agreed, a second small commitment is taken on. Remove the day, and you are not removing a perk; you are breaking arrangements your policy encouraged people to make. That is why a poorly handled ending costs more trust than never starting, and why “we’ll see how it goes” is not a sufficient plan.
The documented reversals are instructive. Alter Agents, a US research firm, reverted and was candid that staff felt they were doing 40 hours of work in 32; the lesson widely drawn from coverage of its decision is that an honest, explained exit at least left the conversation intact. Contrast that with the quiet drift leaders sometimes confess to us in scoping calls: the pilot “paused for a busy quarter” two years ago, and nobody has mentioned it since. Staff have mentioned it. Every Friday.
Pre-commit, in writing
Before launch, publish four things to everyone in the pilot:
- The decision criteria. Quantified floors for customer results and quality, people measures (including actual hours and a wellbeing measure), and the outputs that define success for each team.
- The decision date. A named month in which the decision will be taken and communicated. Pilots without decision dates do not end; they rot.
- Who decides, and what the options are. Continue, extend (with a new end date), adjust (for example to a nine-day fortnight), or stop. “Extend indefinitely” is not an option; it is an abdication.
- What happens in each case. Especially: if we stop, how much notice, what transition, and what is preserved (flexible start times? meeting norms?) so the work of the pilot is not entirely discarded.
That document does more than protect you at the end. It disciplines the middle: teams measure what was agreed, rather than ransacking the data for a favourable number in month five.
Running the ending itself
If the decision is to stop, treat it as a change programme, not a memo.
- Announce to staff before clients, with the results attached — including the measures that moved the wrong way. If staff learn of the end from a client, you have lost them twice.
- Give at least a month’s transition, longer if childcare or care arrangements were built on the day. Offer a defined taper (for example, a nine-day fortnight for a quarter) where the business can bear it.
- Separate the workload finding from the people finding. “Our scorecard says customer resolution breached the floor for two consecutive months, and demand modelling shows we need X” is a business argument. “People weren’t trying hard enough” is an accusation, and in most failed pilots it is false: the work was never triaged.
- Keep the gains that were real. Meeting discipline, async norms, the stop-doing list — ending the day off does not require restoring the noise. Say so explicitly.
- Publish the follow-up. A short internal (and, where credible, external) post six months later — what we tried, what we measured, why we stopped — converts a disappointment into a reputation for straight dealing. Future change becomes easier, not harder.
The honest alternative to ending
Sometimes the data support neither a confident yes nor a clean no. A time-boxed extension with tightened conditions — a different day pattern, extra staffing at peak, one team continuing while another reverts — is legitimate, provided it has its own decision date. What destroys trust is not caution; it is the suspicion that the decision is being avoided because nobody wants to own it.
A pilot you might end is not a weaker promise. Prefaced with clear criteria and ended with transparency, it is a stronger one.
Sources
- UK four-day week pilot results and continuation rates — Autonomy, University of Cambridge, Boston College and 4 Day Week Global, February 2023.
- Alter Agents’ decision to revert, and the workload factors it cited — Fast Company reporting, 2022.
- Germany four-day week pilot continuation and return rates — University of Münster and Intraprenör, October 2024.
Day5Group is a consultancy that assists in the transition to a 4 day work week. More about Day5Group · All blog articles