Every return-to-office memo rests on a forecast. It says that people out of sight will coast, that collaboration will decay, that output will quietly slide. The claim is made with great confidence and almost no evidence. The most useful fact about the strongest study of hybrid work is not only what it measured, but who was most surprised by the result. The managers were.
What the managers expected
When Nicholas Bloom, Ruobing Han and James Liang ran a randomized controlled trial inside the Chinese travel company Trip.com, they were testing the exact arrangement most firms now argue about. More than 1,600 graduate employees were assigned, by the parity of their birthdays, either to work from home two days a week or to stay in the office five (Bloom, Han and Liang, 2024). Randomizing by birthday is not a gimmick. It removes the chance that keen or coasting workers sorted themselves into one group, which is what makes the comparison causal rather than anecdotal.
Going in, the expectation among the company's own managers was that hybrid work would cost the firm output. In Stanford's account of the experiment, that prior was widespread and confident (Stanford, 2024). It is the same instinct behind most RTO mandates. If I cannot watch the work, the thinking goes, I cannot trust it.
What the data said instead
The forecast was wrong on its own terms. Voluntary resignations among hybrid workers fell by roughly a third, from 7.2 percent to 4.8 percent, a difference unlikely to be chance (Bloom, Han and Liang, 2024). Performance did not suffer. Across four consecutive semiannual review cycles, formal equivalence tests found no gap in performance ratings between the hybrid and fully office-based groups. Over the following two years there was no detectable difference in promotions either. Two days at home bought a large retention gain and cost the firm nothing it could measure.
Confronted with those numbers, many of the managers who had predicted a productivity hit changed their minds (Stanford, 2024). That reversal is the part worth dwelling on. Writing later in Harvard Business Review, the authors drew the obvious managerial lesson: do not rely on intuition about how your people work, run the test and read the result (Bloom, Han and Liang, 2024, Harvard Business Review).
Why intuition fails at exactly this question
Presence is a proxy, and a poor one. A manager watching a full office infers effort from attendance, then mistakes the inference for a measurement. The experimental record shows how unreliable that inference is. An earlier trial by the same lead author, at the call center of Ctrip before it became Trip.com, found that fully remote staff were about 13 percent more productive than their office peers, most of it from working more minutes per shift and the rest from a quieter environment (Bloom, Liang, Roberts and Ying, 2015). Anyone forecasting from gut feel would have bet the other way.
The same study is a warning against forecasting in the opposite direction too. Those fully remote workers, despite higher measured output, were promoted less often for the same performance (Bloom, Liang, Roberts and Ying, 2015). Intuition could not have told you that either. The point is not that remote work is always good or that the office is always suspect. It is that this is a domain where seasoned judgment is systematically miscalibrated, and where the cost of a confident wrong guess falls on the whole workforce.
Two cautions keep the reading honest. Both trials ran inside single Chinese firms with graduate or call-center staff, so the exact figures will not transfer cleanly to every industry. And the hybrid result, two days at home, is a different animal from the fully remote arrangement that carried the promotion penalty. Neither caveat rescues the original forecast. It was still wrong.
The practical read
The managers at Trip.com had something most executives issuing RTO mandates do not: an experiment, and the humility to update when it contradicted them. That is a reproducible move. A firm does not have to bet its culture on a memo. It can assign teams to different arrangements, hold performance measurement constant, and compare attrition and output after two review cycles. The authors recommend exactly this, that companies A/B test their own management practices rather than import a policy on faith (Bloom, Han and Liang, 2024, Harvard Business Review).
The uncomfortable implication is that a leader who refuses to run the test is protecting a forecast, not the business. The cost of that protection is not evenly shared. A wrong guess about location is imposed on the entire workforce, while the guesser keeps the authority that made the call. An experiment redistributes that risk, forcing the policy to answer to evidence rather than to seniority. The evidence that hybrid can hold performance while cutting turnover is now unusually strong. Anyone is free to disagree with it. What no longer holds up is disagreeing with it while declining to look, then presenting the refusal as a productivity decision.
Sources
- Bloom, Han and Liang (2024). Hybrid working from home improves retention without damaging performance (Nature 630). RCT of 1,612 Trip.com staff: two days at home cut quits by a third with no measured loss to performance or promotion, contradicting the managers' forecast. https://www.nature.com/articles/s41586-024-07500-2
- Stanford (2024). Hybrid work is a win-win-win for companies and workers. Reports that managers who expected hybrid to hurt output revised their views after seeing the trial results, which anchors the article's argument about miscalibrated intuition. https://news.stanford.edu/stories/2024/06/hybrid-work-is-a-win-win-win-for-companies-workers
- Bloom, Han and Liang (2024). One Company A/B Tested Hybrid Work. Here's What They Found (Harvard Business Review). The authors' own recommendation to A/B test management practices rather than trust intuition supplies the piece's practical close. https://hbr.org/2024/10/one-company-a-b-tested-hybrid-work-heres-what-they-found
- Bloom, Liang, Roberts and Ying (2015). Does Working from Home Work? Evidence from a Chinese Experiment (QJE 130). The +13% productivity result and the promotion penalty show intuition failing in both directions. https://academic.oup.com/qje/article-abstract/130/1/165/2337855