What this is built on: Six snapshots of henry.careers' own listing corpus, 17 September to 2 October 2026, covering 2,944 distinct roles.
If you are watching for senior roles, the useful question is not how many exist. It is how quickly the set changes — because that decides whether checking weekly is enough.
We can answer it, because this board keeps every version of itself. Six measurements between 17 September and 2 October 2026 saw 2,944 distinct UK roles advertised at £100,000 a year or more, or £500 a day or more. The board holds about 1,800 at any moment, which means roughly a thousand more roles passed through than were ever visible at once.
About 3% changes every day
Two independent intervals, measured the same way:
| Period | Days | New | Gone | Turnover |
|---|---|---|---|---|
| 19–24 September | 5 | 293 | 301 | 16% |
| 24–28 September | 4 | 243 | 228 | 13% |
That is about 3.2% of the board a day on both readings, arrived at separately. The totals barely move — 1,856 to 1,871 to 1,801 — which is the part worth noticing. A board that looks static is not. Roughly one role in thirty is replaced daily beneath a headline number that stays put.
Over a fortnight that compounds: a third or more of what is listed today will have gone, and a similar volume you have never seen will have arrived.
What that means if you are looking
Checking once a month means seeing a different board each time, with no overlap worth speaking of. Checking weekly means missing roles that appeared and closed between visits — our own measurements are four and five days apart, and roles appear and vanish inside that gap.
We are not going to tell you how often to look. But the arithmetic is the arithmetic: at 3% a day, a week is about a fifth of the board.
What we cannot tell you yet, and why
The obvious next question is how long an individual role stays open. We can see when each listing first appeared and when it last appeared, so it looks answerable — and the answer we would get is wrong.
Our snapshots are four to five days apart. A role that appears and goes between two of them reads as lasting under four days whether it lasted one or four, and nothing can score above fifteen days in a fifteen-day window. Of the 274 roles with both a first and last sighting, 38% landed in the shortest bucket and none in the longest — both of those are the sampling interval, not the market.
Published, “senior roles stay open a median of five days” would look like a finding and would be a description of how often we happen to measure. So it is not published. When ingestion runs daily and the window is months rather than a fortnight, it becomes a real question with a real answer.
Method
Every sweep of this board is committed, so each measurement is a diff between two versions rather than an estimate. Turnover is the mean of roles added and roles removed, over the board size at the start of the period.
One exclusion, because it would have distorted the figures: between 28 September and 2 October, one employer’s job board stopped responding entirely and 82 of its roles left the board at once. That is a source failing, not a market event, and those are not counted as churn. Without that exclusion the 28 September to 2 October period would read as a 22% collapse in private-sector listings, when the honest figure is roughly flat.