What actually goes wrong with a distributed team?
Rarely the people, and rarely the tooling. It is almost always handovers and decision latency. Work that would move in an afternoon in one room waits for an overlap window; context that would be overheard has to be asked for; and a decision nobody wrote down gets made again a fortnight later by someone who was not on the call.
The symptom a board sees is delivery slipping, so the diagnosis usually comes back as a capacity problem and the answer is more people — which, in a team that is already losing time to handovers, makes it worse.
Is hybrid easier than fully remote?
No. It is consistently the harder of the two, and it is the one most companies are now in. A fully remote team has one mode and everybody is in it. A hybrid team runs two companies at once: one that overhears things and one that does not, with the room setting the default and the screen catching up afterwards.
Fixing it is not a policy about days in the office. It is deciding which things happen in one mode for everybody — and then actually holding that, because the failure is gradual and nobody reports it.
Isn’t this a management problem rather than a technology one?
Both, and the mechanics are technical. How work is decomposed, where state lives, what a pull request has to explain, whether the build tells you something is broken before a colleague in another time zone does — those are engineering decisions, and they determine how much management a distributed team needs. Teams that are expensive to run remotely are usually running a process designed for a room.
What about time zones that barely overlap?
Three hours of overlap is a different design problem from eleven. Beyond about four hours apart, synchronous working stops being the default and anything that still depends on it becomes a queue. That is survivable and sometimes an advantage — work genuinely can be handed round the clock — but only if the handover carries enough context to be picked up cold, which is a writing problem before it is a scheduling one.
| Signal | Working | Not working |
|---|---|---|
| Decision latency | Decided the same day, by someone named | Waits for the overlap window, then for the next one |
| Handovers | Work moves with its context attached | Re-explained every morning, usually by the same person |
| Written record | A decision can be found six months later | It lived in a call nobody recorded |
| Overlap hours | Deliberate, protected, and used for the things that need them | Accidental, and spent on status |
| Hybrid meetings | One mode for everyone, held to | The room talks and the screen listens |
| Onboarding | A new starter ships something in the first week | Months, and nobody can say why |