TLDR
A six-month randomised trial at Trip.com found two days working from home per week cut quit rates from 7.2 to 4.8 per cent with no effect on performance or promotions. Separate research shows junior engineers benefit measurably from sitting near teammates, pointing to real trade-offs for early-career staff.
KEY TAKEAWAYS
The number that should end the argument
Quit rates tell you something pay surveys cannot: how much staff actually want to stay. At Trip.com, the Chinese travel platform, a properly designed randomised trial found that letting engineers and other graduate employees work from home two days a week pushed that number down by a third. The quit rate among hybrid workers fell from 7.2 to 4.8 per cent over six months, a statistically significant drop that the researchers attribute directly to the schedule change.[1] That is not a marginal rounding difference. It is the difference between a team that holds together and one that is quietly bleeding out.
The trial, published in June 2024, involved 1,612 employees randomly assigned to either a full five-day office schedule or a hybrid arrangement of three days in and two at home.[1] Randomisation matters here enormously. Most studies of remote work suffer from selection bias: the people who work from home are already different from those who do not, in ways that confound any comparison. This trial removed that problem by design.
What the data actually showed
Beyond turnover, the hybrid group reported meaningfully higher job satisfaction, scoring a mean of 8.19 on a ten-point scale against 7.84 for the office-only group.[1] The gap is not enormous, but it is consistent and statistically strong. People preferred the arrangement and said so.
The headline finding for sceptical executives is what the trial found on performance. Researchers ran null equivalence tests across two years of follow-up data and found no detectable difference in performance grades or promotion rates between the two groups.[1] The careers of hybrid workers did not stall. Their ratings did not slip. Stanford economist Nicholas Bloom, one of the study's authors, said "these results indicate that a hybrid schedule with two days a week working from home does not damage performance."[1]
Managers were wrong, and then they changed their minds
One of the quieter findings in the paper deserves its own paragraph. Before the trial began, managers at Trip.com estimated that hybrid working would reduce productivity by around 2.6 per cent. After participating in the trial, that estimate reversed: the same managers reported a perceived gain of 1.0 per cent.[1]
That shift is not trivial. It suggests the instinct driving return-to-office mandates at many large firms is not grounded in observed evidence from within those firms. It is grounded in prior assumption, the kind that evaporates when you actually measure what happens. Managers who believed flexibility was leaking productivity changed their assessment once they had data instead of intuition.
The Trip.com result does not prove that every manager at every company is wrong about every flexibility arrangement. It does establish that the presumption of harm is not a safe default.
Where office time genuinely matters
The evidence for hybrid is not a clean sweep, and presenting it as one would be dishonest. A separate line of research points to real costs for early-career workers when physical proximity disappears entirely.
Research examining coding teams found that sitting near teammates increases coding feedback by 18.3 per cent and improves code quality, with gains concentrated among less-tenured and younger engineers.[4] The study authors also found the opposite effect for experienced engineers, who wrote less code when seated near colleagues, likely because proximity generated more interruptions and mentoring demands on their time. The authors said "we find that sitting near teammates increases coding feedback by 18.3% and improves code quality," and also said "there is a tradeoff: experienced engineers write less code when sitting near colleagues."[4]
Proximity transfers skill from senior to junior staff. Remove it entirely and you break something that does not show up in quarterly performance reviews but compounds over years. The same research found that in US national data, young college graduates in remotable jobs showed persistently elevated unemployment rates relative to older cohorts, a signal that the labour-market entry costs of remote work may fall disproportionately on the people least able to absorb them.[4]
How widespread is hybrid work, really?
The policy debate often proceeds as if hybrid work is either ubiquitous or fringe, depending on who is making the argument. Survey data suggests something more fragmented. In 2023, 35 per cent of employed Americans did some or all of their work at home on the days they worked.[2] That figure captures a wide range of arrangements, from a single afternoon at home to full remote schedules.
More granular data from a survey of 150,000 US firms conducted between November 2024 and January 2025 puts the average at one remote day per week across the economy.[3] That average conceals extreme variation: workers in information industries averaged 2.8 remote days per week, while those in accommodation and food services averaged just 0.1 days.[3] The headline number is almost meaningless without the industry breakdown.
The Trip.com trial, while methodologically strong, describes one kind of firm: a large technology-adjacent company with graduate-level employees doing work that can, at least partly, be completed at home without loss. Manufacturing, retail, logistics and hospitality are different categories entirely, and extrapolating the quit-rate finding to those sectors would be a mistake.
What the trial cannot tell you
The Trip.com study is unusually rigorous for workplace research, but it has limits worth naming. It ran during a specific period in China's post-pandemic labour market, covered engineers and managers at a single travel platform, and does not speak to new-hire onboarding, team formation across unfamiliar colleagues, or creative collaboration that depends on unplanned conversation.
The proximity research on junior engineers suggests a two-day hybrid arrangement might still be enough in-office time to sustain mentoring relationships, but the Trip.com paper does not directly measure that. The honest synthesis from the available evidence is this: two days from home appears to retain staff and leave measured performance intact at firms like Trip.com, while full remote arrangements carry documented costs for early-career skill development that do not yet show up clearly in individual performance scores.
For HR teams and executives working through flexibility policy right now, the trial offers a defensible baseline: a structured hybrid schedule, properly designed, is not the productivity threat that managerial intuition often assumes it to be. Whether it suits a specific organisation depends on how much of its value creation runs through proximity, mentoring and the kind of learning that happens when a junior employee watches someone more experienced work through a problem in real time. Nicholas Bloom and his co-authors published the full trial results in Nature in June 2024.
SOURCES & CITATIONS
FREQUENTLY ASKED QUESTIONS
What did the Trip.com hybrid work trial find?
Did hybrid work affect performance ratings at Trip.com?
Are there downsides to remote or hybrid work?
How common is hybrid work across US industries?
Claire Bennett writes about work and workplace culture. She is interested in the gap between how organisations describe themselves and what it feels like to work inside them.
