We watched 1,691 UK job adverts from the hour they appeared. Two thirds never reached 10 applications.

Every piece of job search advice tells you to apply early. Set an alert, check at 6am, get in before the queue builds. So we built a collector that photographed every live advert in ten UK occupations once an hour for 13 days, and watched 1,691 of them from the hour they appeared. The queue that everyone is racing usually does not exist.

Scrapes of job boards are common. What they give you is a stock number: this advert has 46 applications, that one has 3. It tells you nothing about when those applications arrived, which is the only thing that matters if you are deciding whether to fire off an application tonight or write a proper one tomorrow.

An hourly panel answers that. Take a snapshot of the same adverts every hour, keep only the ones that appeared while you were watching, and you have a growth curve per advert instead of a single number.

What the curves say

Two thirds of the adverts we watched never got 10 people to apply. Not in the first hour, not in the first week, not at all while we were watching. That is the number to hold on to, because almost every piece of advice about applying fast is written as though the opposite were true.

Where a queue did build, it built slowly. The tenth application arrived after a median of 28 hours. A quarter of those adverts took more than three days to get there. And for adverts with a genuine queue that we watched for at least four days, a median of 24% of the applications landed on the first day, so three quarters of your competition turns up after you have gone to bed.

The occupation decides everything

The averages above hide the thing that actually predicts your odds. Here is the same analysis per occupation. The panel is ten occupations picked because they are small enough to enumerate completely, so this is not a cross-section of the UK market, but the spread inside it is the point.

Receptionist adverts gained a median of 15 applications and 144 of 247 reached ten of them, typically inside a day. Staff nurse adverts gained a median of nothing. Three of 187 reached ten. Same country, same fortnight, same job board, and a candidate applying for one of those is in a completely different market from a candidate applying for the other.

This matches what we found in a much larger cross-section of adverts last month, where a receptionist advert drew 45 applications and a nurse advert drew 2. What the hourly panel adds is the shape of the growth rather than the final total, and it shows the gap is there from the first day.

So should you still apply early?

Yes in a few occupations, and the data tells you which ones. Library assistant adverts that reached ten applications did it in a median of 14 hours. Receptionist adverts did it in 20.5. If you are applying for busy front-of-house or admin work, the first day is worth having, because that is where the queue genuinely forms fast.

Everywhere else the urgency is manufactured. Occupational therapist adverts took a median of 73 hours to reach ten applications and almost none of them got there. Optical assistant, 56 hours. Paralegal, 53. In those markets an advert can sit for a week with two people in the queue, and the hour you press send is worth nothing next to whether your CV answers the advert.

The practical rule: check the queue if the board shows you one, and spend the time you were going to spend refreshing job alerts on the application itself. A tailored application sent on day three beats a generic one sent in the first hour, in every occupation in this table.

What it means if you are hiring

Two things fall out of this for anyone posting adverts. Repost timing is mostly wasted effort in slow occupations, because the applications are not arriving in a burst you can catch. And a first day with nobody in the queue is not a signal that the advert has failed. Three quarters of the eventual applicants had not arrived yet.

The harder read is for clinical hiring. When 184 of 187 nursing adverts never see ten applications, that is not an advert problem. No amount of rewriting moves a market where the candidates do not exist locally, and the money spent on refreshing and reposting would go further almost anywhere else.

Method

A GitHub Actions workflow ran a collector against the Reed API once an hour from 18 to 31 July 2026, taking 168 snapshots across ten occupations and recording the live application count published on each advert. Adverts that were already live when the experiment started are excluded, leaving 1,691 that appeared while we were watching. Each one is measured from its own first sighting, so applications are counted as growth rather than as a stock figure.

Ages are computed per advert, so "by 24 hours" means the last snapshot taken before that advert turned a day old. The first-day share is only computed for adverts that gained at least five applications and were tracked for at least 96 hours, which is 528 adverts, because a share of two applications is noise.

Limits worth stating. The window is 13 days, so an advert that appeared on the last day was watched for hours rather than weeks, which pulls the total-gained column down and is why the age-anchored columns are the ones to quote. The cadence was hourly by design but ran closer to one snapshot every 90 minutes in practice, so times to the tenth application are accurate to a couple of hours, not to the minute. Application counts are what one job board reported. Reed does not publish applications made elsewhere, so an advert cross-posted to the employer's own site may have a longer queue than we can see.

The wider corpus behind this work, 15,291 UK adverts, is described in ten findings from 15,291 job adverts. We also tested 16 things that supposedly make an advert competitive, and 14 of them did nothing.

Frequently asked questions