How many days do you actually overlap with a remote team?
Every timezone overlap tool answers half the question. The other half costs a London–Tokyo team the better part of a month a year.
By Adam Simmons-Spaans · Updated 2026-09-23
Overlap is measured in hours and lost in days
Pick any timezone tool and it will tell you that London and Tokyo share a narrow window in the early London morning. True, useful, and only half of it. The other half is how many days in the year both sides are actually working — because a public holiday on one side is a day nobody ships on, however well the clocks line up.
Counted properly for 2026: the UK and Japan share 241 working days out of 261 weekdays. 20 weekdays are lost because one side is off — the better part of a month, invisible to every tool that measures only hours.
The US and Germany fare better at 247 shared days of 261, losing 14 — still most of a working fortnight.
Why the number is bigger than people expect
The spread in public holidays is wide. Japan has 16 nationwide public holidays; the United Kingdom has 5. Two calendars that barely coincide do not simply average out — they compound, because a day is lost when *either* side is off, not when both are.
Weekend holidays are excluded from that count, since they cost nobody a working day, and regional holidays are counted separately: a day observed in some German states but not others is a real day off for one person and not a national shutdown.
The asymmetry nobody plans for
Holiday loss is not shared evenly, and that is what makes it awkward socially rather than just arithmetically. The side with fewer public holidays spends more days waiting, and experiences the gap as the other team being permanently unavailable; the side with more spends more days coming back to a backlog built without them.
Clusters make it worse than the totals suggest. A country with sixteen holidays spread evenly costs a partner sixteen scattered days. A country that takes most of them in one stretch costs a working week in which nothing moves at all, and no monthly average will show you that.
The fix is not to avoid the pairing. It is to know the shape before committing: which weeks are structurally dead, and whether the overlap you are counting on survives them.
What to do with it
If you are hiring across timezones, the shared-working-days figure belongs in the decision alongside the overlap hours. A pairing with three hours of overlap and aligned calendars can easily beat one with five hours and calendars that miss.
And if you are the one moving, it is worth knowing which holidays you are inheriting. The local calendar is part of the compensation, and it is one of the few parts nobody negotiates because nobody thinks to price it. A statutory week is a week whatever your salary is.
There is a caveat worth stating, because it cuts against the headline. Public holidays are not the same as leave: some countries with short statutory holiday lists have generous annual-leave entitlements that more than close the gap, and a few substitute a weekday whenever a fixed date falls at a weekend while others simply lose it. Counting only the public calendar answers the coordination question — which days is the other side unavailable — and not the quality-of-life one.
Nor does any of this survive a company that ignores it. Plenty of distributed teams operate on their headquarters' calendar regardless of where people live, which resolves the coordination problem by pushing the cost entirely onto whoever is furthest from head office. Worth knowing which arrangement you are joining.
Check your own pairing
Every figure above is computed from published data and recomputed on each build. Run it against your own numbers rather than taking the example: check your own pairing .
About the author
Adam Simmons-Spaans, Founder, Geo-Parity. Adam builds Geo-Parity single-handedly: the engines, the datasets and the fetchers that refresh them. Every figure on the site comes from a published source — World Bank, IMF, OECD, Eurostat, WHO, HUD, Cloudflare Radar — and where a number is modelled rather than measured, the page says so. More about the site.