A 79-point gap between two cities 160 kilometres apart
London and Birmingham sit in the same labour market. Same statutory floor on flexible-working requests, same tax treatment, same commuting culture, roughly the same industries at the top of the list. On our city pages today, London reports 0% remote-friendly roles and Birmingham reports 79%.
Manchester, between them geographically and industrially, reports 54%. New York reports 0% and Boston 81%. Toronto 0%, Vancouver 44%. Sydney 0%, Melbourne 41%.
No account of remote work explains those pairs. Something other than the labour market is producing them.
| City | Country | Sources read | Remote-friendly |
|---|---|---|---|
| London | United Kingdom | 5 | 0% |
| Manchester | United Kingdom | 10 | 54% |
| Birmingham | United Kingdom | 10 | 79% |
| New York | United States | 7 | 0% |
| Boston | United States | 10 | 81% |
| Austin | United States | 14 | 76% |
| Toronto | Canada | 9 | 0% |
| Vancouver | Canada | 10 | 44% |
| Sydney | Australia | 12 | 0% |
| Melbourne | Australia | 10 | 41% |
The number tracks our source count, not the city
Each city page prints how many sources it was built from. Once you put that column next to the remote share, the pattern is immediate.
Eleven cities are built from seven sources or fewer. Nine of them report exactly 0%. Seventeen cities are built from ten or more. One of them reports 0%. The mean remote share is 7.3% in the thin group and 53.5% in the thick group — a sevenfold difference produced by how much we read, not by where the work is.
London is built from 5 sources. Birmingham from 10. New York from 7, Boston from 10, Austin from 14. The cities that look most remote-hostile are the cities we know least about.
| Sources feeding the page | Cities | Reporting 0% | Mean remote share |
|---|---|---|---|
| 7 or fewer | 11 | 9 | 7.3% |
| 8 to 9 | 9 | 5 | 11.4% |
| 10 or more | 17 | 1 | 53.5% |
Why thin coverage manufactures zeroes rather than noise
A remote flag has to be carried by the listing. Direct employer career pages and the large national boards frequently do not carry one at all: the location field says London and nothing anywhere says the role can be done from elsewhere. Absence of a flag becomes absence of remote work.
When a city page is built from a handful of such sources, every listing arrives unflagged and the page computes zero. Add sources that do carry the flag — smaller specialist boards, remote-first aggregators — and the share jumps, without a single new job existing.
So the number is not measuring the market. It is measuring the intersection of the market with the metadata conventions of whoever we happened to read.
What we are doing about our own column
Two things, and the first is unglamorous: this desk will not publish city-level remote percentages, in the same way it retired city-level advertised medians last week for the same class of reason. A figure that changes when we add a source is a figure about us.
The second is that the column stays on the city pages, with its source count next to it, because a reader who can see both can see the artefact for themselves. Hiding the number would be tidier and less honest.
The broader point is not about us. Every job-market dashboard you have read this year has this problem, and almost none of them print the denominator that would let you catch it. When a chart tells you a city has no remote work, the first question is not what changed in that city. It is how many places the chart looked.
Every figure here is on a page you can open. Read any city at babzituna.com/jobs and check the source count next to the remote share yourself.