Companies adopt asynchronous work the way they adopt any tool, by installing it. They pick the messaging app, write a policy about response times, and declare themselves async. Then the meetings creep back, people apologize for slow replies, and everyone concludes that async does not suit their business. The diagnosis is almost always wrong. The problem is rarely the tooling. It is that the firm never learned to define good work precisely enough to stop watching people produce it.
Async is a measurement claim in disguise
Strip asynchronous work to its core and it makes a demanding assumption. If a colleague is not online when you are, you have to be able to evaluate what they delivered on its own terms, without the reassurance of having seen them do it. That is a claim about measurement before it is a claim about calendars. Matt Mullenweg, who runs the distributed company Automattic, makes the dependency explicit in his ladder of distributed-work maturity, where the decisive rung is the shift to judging people on output rather than hours or process (Mullenweg, 2020). His account is a practitioner's opinion rather than a study, and his boldest claim, that a fully distributed firm outperforms any in-person one, comes without supporting evidence. But the structural point stands. Async is what becomes possible once output is legible.
The randomized trials give that idea empirical teeth. The reason the Trip.com study could conclude that hybrid work did not damage performance is that the firm already measured performance in a way it trusted. Formal equivalence tests found no difference in review scores across four semiannual cycles (Bloom, Han and Liang, 2024). Those tests were only meaningful because the underlying metric was real. In their managerial write-up, the authors list rigorous performance management first among the conditions the result depended on (Bloom, Han and Liang, 2024, Harvard Business Review). A firm without that measurement could not have run the analysis, let alone trusted the answer.
Documentation is how the definition scales
Defining good work is necessary but not sufficient. The definition has to be written down and available, or every worker still has to interrupt someone to learn what done looks like. This is where the all-remote company GitLab locates the whole discipline. It runs handbook-first, documenting its operations in a public handbook that would exceed 2,000 printed pages, and it states flatly that asynchronous work is impossible without strong documentation, because without it a single follow-up question blocks a colleague for hours or days (GitLab handbook, 2026). This is a self-reported case, not a measured outcome, and deserves that label. But the mechanism it describes is exactly the measurement problem in operational form. If expectations, decisions, and standards live only in someone's head or in a meeting, then async work stalls the moment that person logs off. Writing the definition down is what lets people work on different clocks without colliding.
Notice how this reframes the usual async failure. A team that keeps falling back to synchronous meetings is not undisciplined about calendars. It is compensating for an absence. The meetings are where the undocumented definition of good work gets negotiated in real time. Remove the meetings without first writing the definition down and you have not made the team async, you have made it blocked.
The order of operations
The practical sequence follows directly, and most firms attempt it backwards. They start with the schedule change and hope the measurement catches up. The evidence points the other way. First, define what good output is in each role, concretely enough that a competent colleague could evaluate it cold. Second, write those definitions and the decisions around them down, so the answer to a routine question is a document rather than a person. Only then does turning off the requirement to be online at the same time become safe rather than reckless.
Two cautions keep this grounded. The Trip.com evidence comes from a single Chinese firm with graduate employees, and it studied hybrid rather than fully asynchronous work, so it supports the measurement argument without settling every async question. And GitLab's handbook culture is a story the company tells about itself, which recent employee reviews suggest may have strained under later leadership. Neither undercuts the logic. Both illustrate it.
The practical read
When an async experiment fails, the honest post-mortem should not start with the tools. It should ask whether the firm could actually say, for each role, what a good week of output looks like without watching anyone produce it. If the answer is no, the firm was never ready to go async, whatever its software allowed. The good news buried in this is that the work of getting ready pays off even for teams that never fully leave synchronous work behind. Defining output clearly and writing it down improves onboarding, reduces the interruptions that fragment a manager's day, and makes performance reviews defensible, all of which held true at the firms whose results this evidence rests on (Bloom, Han and Liang, 2024; GitLab handbook, 2026). Async is not a permission you grant your calendar. It is a capability you earn by learning to measure the work, and the earning is worth it whether or not you ever switch off the meetings entirely.
Sources
- Mullenweg (2020). Five Levels of Autonomy. Practitioner taxonomy whose decisive rung is judging work by output rather than hours, framing async as an output-measurement capability. https://ma.tt/2020/04/five-levels-of-autonomy/
- Bloom, Han and Liang (2024). Hybrid working from home improves retention without damaging performance (Nature 630). The equivalence tests on performance ratings were only possible because the firm already measured output well, grounding the measurement argument. 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). Lists rigorous performance management as the first success condition, supporting the claim that async depends on defined output. https://hbr.org/2024/10/one-company-a-b-tested-hybrid-work-heres-what-they-found
- GitLab handbook (2026). Handbook-first and asynchronous work. Self-reported case that async is impossible without strong documentation, illustrating how a defined standard must be written to scale. https://handbook.gitlab.com/handbook/company/culture/all-remote/asynchronous/