The cheapest thing a company can do is nothing, right up until the morning someone good hands in their notice. Then the meter starts. The job gets reposted. A manager spends her afternoons interviewing instead of managing. The leaver's work spreads across people already tired, and stays spread for months while a replacement is found, hired, and brought up to speed. None of it shows on the budget. All of it is real money.
When the UK ran the largest trial of a shorter week to date, 61 companies and around 2,900 workers cutting their hours by a fifth, to about 34 a week, on full pay, the headline that traveled was wellbeing: 71 percent of employees reported less burnout. True, and good. But the figure that should have frightened finance directors was this: over the six months of the trial, resignations fell by 57 percent.
Revenue, meanwhile, held. It rose 1.4 percent on average, weighted by company size. The policy stuck: 92 percent of the firms kept it, and a year on most still had it and half made it permanent. More than 250 British employers are now accredited as standing four-day-week workplaces.
So do the turnover savings alone pay for the fifth day?
Ask what a leaver costs. SHRM, the main US body for HR professionals, puts replacing a salaried worker at somewhere between half and twice their annual pay once you count recruiting, lost output, and ramp-up. Oxford Economics, in an older but much-cited study it ran for the insurer Unum, landed on a harder British figure: 30,614 pounds to replace one employee earning 25,000 or more, most of it the lost output while a replacement climbs to full speed. Now cut your resignations by more than half. That 57 percent came from 61 firms that chose the experiment, measured against the same six months a year earlier, with no control group running alongside, so read it as a strong signal rather than a guaranteed lever. Even discounted, for a mid-sized employer this is not a soft benefit, it is a line item that shrinks.
The fifth day was never a gift the company gave. It was the price of not burning people until they leave, and the old model was paying that price all along, in churn, in reposted jobs, in the knowledge that walked out the door. We just never itemized it.
The honest caveat: these firms skew small, white-collar, knowledge work. Nobody has run this at scale on a night shift or a hospital ward, and a four-day week that rewards desk staff and skips the people who cannot leave their post is a perk dressed as a principle. And the cost of the fifth day here is opportunity cost, not wages. Pay stayed flat. Output stayed flat. What moved was who stayed.
Then the machine walks in. PwC's read of more than a billion job ads finds productivity growing about 40 percent faster at the most AI-exposed firms. If a tool genuinely lifts output per hour, the one precondition that made the four-day week revenue-neutral, the same work in less time, gets easier. The question is who collects the dividend. Handed to workers, it buys back a day. Handed to owners, it buys a smaller payroll. The technology does not decide that. We do.
Ask the three questions of any workplace and you learn who it is really for. Who benefits. Who carries the risk. Who gets to leave, and who cannot. The four-day week answers none of them. It just makes the price of the current answer, finally, legible.
