Return-to-office mandates are loud. Calendars filled with status calls are louder. The argument over office days risks missing the real constraint in 2026: not where work happens, but how it moves when teams are rarely in the same room at the same time. Hybrid is a seating chart. Async is the operating system.
What hybrid actually buys you
A randomized controlled trial inside Trip.com, published in Nature, is unusually clear about the effects of a two-days-at-home schedule. Over roughly 1,600 university-trained employees randomized by birthday parity, resignations fell by about one third for those on hybrid, from 7.2 percent to 4.8 percent, with no detectable difference in performance reviews or promotions over two years (Bloom, Han and Liang, 2024). This is causal evidence, not a survey correlation, although it is one firm and one context. The point is directionally robust: hybrid reduces quits without hurting measured output or career progression within the experiment's window.
The managerial translation from the authors adds what many firms miss. The retention gain turned into profit when paired with three conditions: rigorous performance management, coordinated team or company schedules, and visible support from senior leadership (Bloom, Han and Liang, 2024, HBR). Trip.com estimates the reduction in attrition saved millions in recruiting and training. That number is an internal estimate, but the mechanism is common sense. Keeping experienced people is cheaper than replacing them.
Async is the missing operating system
If hybrid is a policy, async is a practice. The Trip.com study tested days at home, not messaging norms or documentation habits. Yet its three conditions point straight at asynchronous operations. Rigorous performance management means measuring work by outputs, not by time online. Coordinated schedules mean teams set shared rhythms for when real-time interaction is needed and when the written record carries the load. Senior support means leaders model these defaults so they survive the pull of old habits.
There is a cautionary case that shows why process matters. A decade earlier, a randomized experiment at Ctrip assigned call-center staff to work fully from home or stay in the office for nine months. Remote workers were 13 percent more productive, largely because they worked more minutes per shift and enjoyed a quieter environment. They also had higher job satisfaction and lower attrition. Yet their promotion rate, conditional on the same performance, fell during the trial period (Bloom, Liang, Roberts and Ying, 2015). When the company later let everyone choose their arrangement and learn, the gains nearly doubled to 22 percent. The lesson is not that full remote is bad. It is that absent a system that makes output visible and careers legible, out of sight can still mean out of mind.
Async practices supply that system. GitLab, one of the most public all-remote operators, states its mechanism plainly: asynchronous work is unworkable without strong documentation, because every unanswered follow-up blocks a colleague for hours or days. The firm runs a "handbook-first" culture, a living manual that would exceed 2,000 printed pages, so anyone can self-serve answers instead of waiting for a meeting (GitLab handbook, 2026). This is case evidence rather than a controlled trial, and GitLab’s outcomes are self-reported. Still, the operating logic matches what the Trip.com authors prescribe. If performance is managed by outputs, you need a written record to know what was decided, by whom, and against what standard.
Practitioners often describe the goal as Level 4 on Matt Mullenweg’s five-level autonomy scale: work is asynchronous and judged on results, not hours or the process used to get there (Mullenweg, 2020). His framework is opinion, not experimental science, and its claim that the highest level outperforms any in-person firm lacks empirical proof in that essay. Even so, the frame is useful because it maps cleanly to the conditions that stabilized hybrid at Trip.com.
Building an async spine in a hybrid firm
Most companies do not need a manifesto. They need three concrete moves that line up with the evidence.
-
Make output the unit of performance. Translate roles into shipped artifacts, service levels, or quantified milestones. That is the "rigorous performance management" the Trip.com authors call for (Bloom, Han and Liang, 2024, HBR). It reduces the career penalty that fully remote Ctrip employees faced when managers could not easily see their effort, because results are visible by design (Bloom, Liang, Roberts and Ying, 2015).
-
Write it down. Adopt a handbook-first norm for decisions, operating procedures, and status. GitLab’s claim that async collapses without documentation is not a randomized test, but it names the bottleneck that kills most hybrid teams: waiting. A searchable, current record turns many synchronous check-ins into asynchronous updates and gives new hires context on day one (GitLab handbook, 2026).
-
Coordinate time on purpose. The Trip.com study did not say, "Let every team do whatever they want." It highlighted the value of coordinated schedules at the team or company level (Bloom, Han and Liang, 2024, HBR). Decide the few windows each week when a team must be live at the same time, and protect the rest for deep work and asynchronous exchange. Then let individuals choose their two home days within that frame. The Ctrip follow-up shows that enabling informed choice, when the system can absorb it, magnifies gains through selection and learning (Bloom, Liang, Roberts and Ying, 2015).
None of this says that meetings vanish. It says meetings earn their keep. Weekly rituals become crisper when the pre-read is mandatory and the decision record is updated in the open. Performance conversations are easier when a quarter’s worth of output is captured in a shared doc rather than a memory of who spoke up loudest on a Tuesday.
The practical read
The retention dividend from hybrid is real in at least one large field experiment, and Trip.com’s managers converted it into dollars saved (Bloom, Han and Liang, 2024, Nature; Bloom, Han and Liang, 2024, HBR). The operational dividend arrives only when a company upgrades how work travels across time zones, floors, and home offices. That means outputs as the currency of performance, coordination as a design choice rather than a calendar accident, and documentation as infrastructure, not as a chore.
Boards will keep debating badge policies. Executives who want an advantage will treat hybrid as a trust audit and async as the architecture that makes that trust visible. The policy sets where people sit. The operating system decides whether the work ships.
Sources
- Bloom, Han and Liang (2024). Hybrid working from home improves retention without damaging performance, Nature. A randomized trial inside Trip.com where two days at home cut resignations by about one third with no loss to performance or promotions, anchoring the case that hybrid can improve retention without harming output. https://www.nature.com/articles/s41586-024-07500-2
- Bloom, Han and Liang (2024). One Company A/B Tested Hybrid Work. Here's What They Found, Harvard Business Review. Managerial translation identifying three conditions for successful hybrid and reporting Trip.com’s internal estimate that reduced attrition saved millions, guiding implementation. 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, Quarterly Journal of Economics. A randomized trial at Ctrip where fully remote raised performance 13 percent and cut attrition but reduced promotion rates conditional on performance, then doubled gains to 22 percent when employees could choose, underscoring visibility and choice. https://academic.oup.com/qje/article-abstract/130/1/165/2337855
- GitLab (2026). GitLab — handbook-first & asynchronous work (all-remote handbook). Case evidence that async requires strong documentation, exemplified by a public handbook exceeding 2,000 printed pages, illustrating the operating mechanism behind async practices. https://handbook.gitlab.com/handbook/company/culture/all-remote/asynchronous/
- Mullenweg (2020). Five Levels of Autonomy. A practitioner framework where Level 4 defines asynchronous work judged by output, offering a useful lens for designing hybrid operating norms while noting it is not empirical evidence. https://ma.tt/2020/04/five-levels-of-autonomy/