What eight months of fully remote work actually taught us

The argument about remote work has been stuck in the same loop for years. One side produces a study showing engineers ship more from home. The other produces a study showing junior staff learn less. Both are correct, and neither is the interesting part.

What we found, after eight months of running a distributed team with no office to fall back on, is that location barely predicts anything. Documentation does.

Meetings are a symptom

When a team goes remote, meeting load usually spikes. The instinct is to treat this as a scheduling problem and to solve it with policy — no-meeting Wednesdays, a thirty-minute default, an agenda requirement.

Those help at the margins. But most recurring meetings exist to transmit information that nobody wrote down. If the decision, the reasoning and the current state all live in someone's head, the only retrieval mechanism is a conversation. The meeting is the interface to an undocumented system.

Every recurring status meeting is a document that someone decided not to write.

The three-hour rule

The most useful convention we adopted was deliberately unglamorous: any decision that affects more than one person gets written down within three hours, in the place where that kind of decision lives, with the reasoning included.

Not a summary. The reasoning. Six months later the summary tells you what was decided; only the reasoning tells you whether the decision still applies.

• Decisions go in the project doc, not the chat thread.
• Chat threads get a link to the doc, not a paraphrase.
• If nobody can find it in two minutes, it was not written down.

Where remote genuinely struggles

Two things did get harder and we never fully solved them. The first is apprenticeship. A junior engineer learns an enormous amount from ambient exposure — overheard debugging, watching someone senior make a judgement call in real time. Screen sharing recovers maybe half of that.

The second is the low-stakes conversation that turns into a good idea. You cannot schedule serendipity, and the attempts to do so — randomised coffee pairings, virtual hangouts — mostly produce polite small talk.

What we would do differently

We would invest in written onboarding far earlier, and we would be much more deliberate about pairing juniors with seniors on real work rather than on training exercises. We would also stop pretending that four hours of overlapping time zones is the same thing as being asynchronous.

The teams that thrive remotely are not the ones with the best video conferencing. They are the ones where a new hire can reconstruct the last two years of reasoning from the archive.