Return-to-office mandates claim to fix culture and speed, yet the best field experiments say the office is not the lever that moves performance. The real bottleneck is coordination. If your teams still need a meeting to know what to do next, hybrid will stall. If the work is written, measurable, and findable, hybrid gains compound.
The experiments remove the alibi
Two randomized trials inside the same company set the floor for this debate. In a large hybrid experiment at Trip.com, employees assigned to work from home two days a week resigned a third less than those in the office full time. There was no detectable hit to performance reviews or promotions across two years of follow-up (Bloom, Han and Liang, 2024). This is causal evidence, not a survey, and it closes the argument that hybrid inevitably harms output or careers. It did not in this setting.
A decade earlier, in a different unit and a different design, the same research team ran a nine-month trial at Ctrip where volunteers worked fully from home or stayed in the office. Home workers were 13 percent more productive and attrition halved, but their promotion rate, conditional on the same performance, fell during the experiment (Bloom, Liang, Roberts and Ying, 2015). When the company later let everyone choose their arrangement, the overall performance gains nearly doubled, largely from people selecting the setup that suited them best (Bloom, Liang, Roberts and Ying, 2015). The lesson is not that home always wins. It is that choice and measurement matter, and that visibility into work becomes a career variable if managers still evaluate proximity.
The managerial translation of the Trip.com hybrid study makes the implementation conditions explicit: successful hybrid models rest on rigorous performance management, coordinated team schedules, and visible top-leadership support. The company also estimates that the fall in attrition saved millions in hiring and training costs, a figure presented as an internal estimate rather than an audited number (Bloom, Han and Liang, 2024 HBR). Treat those three conditions as the scaffolding. Hybrid is not a location perk. It is an operating model that needs measurement, coordination, and sponsorship.
Documentation is the real-time office
Async is often sold as etiquette. It is not. It is infrastructure. GitLab, one of the largest all-remote firms, runs a “handbook-first” model in which the new default is to document decisions and how-to’s in a public manual that would exceed 2,000 printed pages. The company states the mechanism plainly: asynchronous work is impossible without strong documentation, because otherwise a single follow-up question can block a colleague for hours or days. The point is not the size of the handbook. It is the idea that people can self-serve the next step without waiting for someone to come online (GitLab handbook, 2026). This is a case study and the claims are self-reported, not an outcome experiment, but the operating logic aligns with what the Trip.com experiment required: clear expectations, consistent schedules, and a system that makes work visible.
If you cannot see the work, you manage by presence. If you can see the work in a shared, searchable record, you manage by progress. That distinction helps explain the tension between the 2015 and 2024 results. In 2015, fully remote staff delivered higher output yet saw fewer promotions at the same performance. The paper does not establish why, so any explanation is interpretive. A reasonable reading is that proximity and ad hoc visibility still guided advancement, which an async, written workflow could mitigate by shifting what counts as visible. The 2024 hybrid trial reports no difference in promotions, which is consistent with a model where schedules are coordinated and performance is tracked with discipline (Bloom, Han and Liang, 2024).
Manage by outputs, not attendance
The management playbook embedded in the Trip.com analysis is concrete. First, tighten performance management. That means defining outcomes at the role level and reviewing them on a cadence that runs regardless of location. Second, coordinate schedules at the team or company level so people overlap meaningfully when they do meet, rather than scattering days at home (Bloom, Han and Liang, 2024 HBR). Both moves require a shift in what managers pay attention to: documented goals, written decisions, and artifacts that travel through time zones without a meeting.
Practitioners have long framed this as a maturity curve. Matt Mullenweg, who leads Automattic, describes five levels of autonomy in distributed work. Level four centers on asynchronous work evaluated by outputs rather than hours or process, which matches the performance-management spine documented in the Trip.com guidance. His claim that a level-five distributed firm outperforms any in-person organization is an opinion, not backed by causal evidence in that post, but the taxonomy captures the managerial choice at stake: do you measure what was shipped or who was seen (Mullenweg, 2020)?
Put differently, async is not an HR policy. It is a commitment to an audit trail: decisions written down, tasks scoped clearly, owners named, deadlines stated, and status visible without a meeting. When that record exists, hybrid keeps the retention benefit the Trip.com trial measured, while protecting against the career penalties that can arise when contributions disappear into unrecorded conversations (Bloom, Han and Liang, 2024; Bloom, Liang, Roberts and Ying, 2015).
A practical audit for leaders
You do not need to bet the company. Run the A/B test the researchers recommend. Start with one function for 90 days.
- Define three to five measurable outputs per role. Tie reviews to those and publish the rubric internally. This is the “rigorous performance management” condition from the Trip.com analysis (Bloom, Han and Liang, 2024 HBR).
- Set team-level anchor days in the office and quiet hours for deep work. Treat schedule coordination as a system constraint, not a personal preference (Bloom, Han and Liang, 2024 HBR).
- Move decisions and how-to’s into a shared handbook. Make “link or it did not happen” your norm. This mirrors the GitLab mechanism that lets work continue without waiting for replies (GitLab handbook, 2026).
- Track resignations, performance-review distributions, and promotion decisions relative to the control group. The 2024 experiment’s benchmark is fewer quits without a drop in performance or promotions. Treat that as the bar to clear (Bloom, Han and Liang, 2024).
The close is simple. Hybrid will buy you retention. Async will determine whether that gain translates into a durable way of working. If it is not written, it is not hybrid. It is just the old meeting culture spread across more kitchens and commute lines.
Sources
- Bloom, Han and Liang (2024). Hybrid working from home improves retention without damaging performance, Nature. A randomized controlled trial at Trip.com where two days at home cut quits by roughly a third with no loss to performance or promotions. 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 of the Trip.com RCT identifying three conditions for success and reporting internal estimates of savings from reduced attrition. 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 full-time home work raised performance and cut attrition but reduced promotion odds conditional on performance. https://academic.oup.com/qje/article-abstract/130/1/165/2337855
- GitLab handbook (2026). Handbook-first and asynchronous work. A practitioner case stating that async is impossible without strong documentation and describing the handbook-first mechanism. https://handbook.gitlab.com/handbook/company/culture/all-remote/asynchronous/
- Mullenweg (2020). Five Levels of Autonomy. An opinion framework positioning async work as output-based performance at higher maturity levels, used here as practitioner context. https://ma.tt/2020/04/five-levels-of-autonomy/