Everybody in this market repeats the same advice, apply early, and nobody says how early, because almost nobody has the data to say. Job boards know what is open. Very few of them keep a record of what closed, and when.
This one does, as a side effect of how it works. Every employer's board is re-read on the hour. A posting that was on Monday's read and is missing from Tuesday's gets stamped closed. That stamp, across 26,102 postings, is a reasonably honest measure of how long a European tech role is actually open for.
Here is what it says.
Half of all roles are gone in 10 days
The middle half sits between 4 days and 20 days. So the shape of the thing is: a fortnight, give or take, and a long tail of roles that sit open for months because somebody forgot to take them down or the req keeps getting re-opened.
10 days sounds like plenty of time. It is not, and the reason is in the next number.
11.2% never see a second morning
11.2% of closed postings were gone within 48 hours of appearing. 35.9% were gone within a week.
Think about what a two-day posting actually is. It is not a role that got filled in two days. It is almost always a req that was already spoken for, an internal candidate, a referral in flight, an agency shortlist. Posted because the process requires it to be posted, and pulled the moment the paperwork allowed. If you found it on day three, you were never in that race. You could not have been.
The ones you can win are in the 35.9% that lasted a week or the 62% that lasted a fortnight. And in those, the day you arrive decides almost everything, because of how applications arrive.
Applications are front-loaded, and postings are not
Application volume on a public tech posting is brutally front-loaded. The first day carries a large share of the total, the first three days carry most of it, and after the first week the trickle barely moves the pile. The posting stays open for 10 days; the shortlist does not.
So the two curves are misaligned, and that gap is the entire problem:
The role is open for 10 days. The decision about whose CV gets read properly is made in the first two or three.
This is why "there are 15,000 jobs on this site" is a number that helps nobody. Of the roles open right now that you would be a serious candidate for, the ones you can still win are the ones that went up in the last day or two. Everything else is either already shortlisted or was never real.
What arrives, and how fast
Roughly 3,566 new European roles a week, on a board of 20,151 open ones. That ratio is the useful one to hold onto: at any moment, the slice of the board that is actually contestable is a few percent of it.
Which is a much less impressive number to put on a homepage, and a much more useful one to organise your week around.
What to do with this
Three things, and none of them is "apply to more jobs".
Narrow hard, then watch. A search you can check in ninety seconds every morning beats a search that returns 400 results and gets checked on Sunday. Pick the role, the countries and the two or three things that actually rule a job in or out for you. Remote or not, sponsorship or not, and let everything else go.
Treat anything older than a week as a long shot. Not a non-shot. Apply if it is genuinely yours. But know that you are arriving after the shortlist and price your effort accordingly, a tailored application on a nine-day-old posting is usually worse spent than the same hour on something that went up this morning.
Get told, do not go looking. Checking a board is a behaviour that decays; you do it daily for a week, then twice, then when you remember. An alert does not decay. This is the entire reason this site mails people rather than hoping they come back.
How this was measured, and what is wrong with it
Honesty about method, because the numbers above are only worth as much as the method under them.
The clock starts at first_seen_at, the hour the posting first
appeared on this board, and not at the employer's own publish date. Two
reasons: a large share of postings carry no publish date at all in the ATS
feed, and when they do carry one it is frequently the date the req was
created rather than the date it went live. First-seen is the moment the role
became findable by a member of the public, which is the reader's clock and
the one the question is about.
The known biases, stated plainly:
- A newly-added employer looks like a burst. When a company is added to the board, its whole open board arrives at once and those postings' clocks all start that hour, even though some had been live for months. That pushes the measured lifetimes down. The effect shrinks as an employer's history accumulates, and the medians here are dominated by employers who have been tracked for a long time.
- A closed posting is not necessarily a filled one. Reqs get cancelled, frozen, re-scoped and re-posted. "Gone from the board" is what is measured; "hired somebody" is not knowable from outside.
- Re-posts count twice. A role pulled and re-listed a week later appears as two short postings rather than one long one, which again pushes the measure down.
- This is one slice of the market. European roles at 382 mostly well-known technology employers, reading their own ATS endpoints. Agency-heavy corners of the market behave differently and are not represented here.
The direction of all four biases is the same, which is worth saying: they make roles look shorter-lived than they are. So read 10 days as a floor rather than as the truth. It does not change the advice. If anything the front-loading of applications matters more, not less, on a posting that survives longer.
If you want the live version of all of this rather than the prose: the board is here, sortable by date, and every row prints how old it is. The employer table is the other half of the same dataset.
