On 8 September 2026 Flint had 49,916 live UK job listings in its pool, pulled directly from 9,719 employers and about 120 boards, ATS feeds and aggregators. I've been meaning to sit down with that table for months. This week I did, and the numbers were more lopsided than I expected.
Quick note on the data before anyone asks. These are listings we fetch twice a day from the source (Greenhouse, Lever, Workday, NHS Jobs, Civil Service Jobs, Reed, Totaljobs, TES and so on), deduplicated across boards, and marked inactive when the source drops them. I've excluded anything with a non-UK location. Salaries only count if the employer wrote one; Adzuna's "estimated" salaries are thrown out, and I'll come back to why. Everything below is a snapshot of one day, not a survey.
Pay transparency runs backwards
38.7% of live UK listings state a salary. That's the headline number, and on its own it's neither shocking nor useful. The split underneath it is the interesting part.
Social care roles state pay 70% of the time. Clinical roles, 74%. Teaching, 51%. Corporate finance, 47%. Then it falls off a cliff: marketing 26%, data engineering 24%, software engineering 17%, machine learning 15%, product management 14%.
Line that up against what those jobs pay when they do say. The median advertised range for social care is £30,288 to £34,016. Teaching, £32,916 to £42,000. Software engineering, £65,000 to £80,000. Machine learning, £78,700 to £95,000.
So the rule, at least in this pool, is that the higher the salary, the less likely you are to be told it. The jobs paying £30k print the number on the advert. The jobs paying £80k make you get to the second interview to find out.
Part of that is structural. NHS and school roles sit on published pay bands, so there's nothing to hide. But most of it is a choice, and you can see who's making it. Listings that come straight from an employer's own applicant tracking system (Greenhouse, Lever, Ashby, Workday, the VC portfolio boards) state a salary 4.7% of the time. Listings on UK job boards do it 53% of the time. Reed, 65%. The same company will put a salary on Reed because Reed nags them for one, and leave it off their own careers page because nobody does.
The big direct employers in the pool make the point on their own. Airwallex has 508 live UK listings and a salary on none of them. Capco, 504, none. Stripe, 334 UK listings, 0.9%. Databricks, 153, 0.7%. The outlier is Anthropic, which puts a range on 87% of its 55 UK roles. So it can be done, and one of the best-paying employers in the pool is the one doing it.
And when a range is given, it's narrow. The median advertised range is 16.7% wide. £40,000 to £46,700, that sort of thing. Wide "£50k to £90k depending on experience" bands exist but they're the minority, no?
One more thing on salaries, because it catches people out. Adzuna carries a salary on most of its UK listings, and 100% of the ones in our pool are flagged by Adzuna itself as estimates. Reed's are real (0.4% estimated). Google for Jobs sits in between at 12%. If you're comparing offers against "the market rate" you saw on an aggregator, check whether the number was written by the employer or by a model.
Six senior roles for every junior one
I ran the titles through a regex. 15.1% of live UK listings have "senior" in the title. 11.9% have lead, head, principal or director. 2.8% have junior, graduate, entry level, trainee or apprentice.
7,524 senior titles. 1,379 junior ones. Five and a half to one across the whole pool, and that's being generous to the junior side, because "graduate scheme" listings tend to cover dozens of hires while a senior listing covers one.
In software it's worse. 31.7% of live software engineering listings are senior. 1.6% are junior or graduate. Twenty to one. Data engineering is twelve to one. The only field where the ratio flips is teaching, where 3.7% of listings are trainee or early-career and 1.3% are senior, and that's because schools hire for classroom roles and "senior" is a management grade.
I wrote about the collapsing middle and the missing bottom rung back in April, mostly from other people's data. This is our own pool saying the same thing, five months on. If you're 22 with a CS degree, the market is hiring your future manager and not you.
Two in five "live" jobs are over a month old
40.5% of live UK listings were posted more than 30 days ago. 22.6% were posted more than 60 days ago.
"Live" here means the source still shows the job. It hasn't been withdrawn, closed or filled, as far as the board knows. But a listing that's been up for two months is a different thing from one posted on Tuesday. Either the role is hard to fill, the employer stopped looking and forgot to take it down, or it was never really a role in the first place. I put the ghost-job research in an earlier post; the short version is that Greenhouse's own audit found 18 to 22% of postings per quarter never produce a hire.
How long a listing lives depends a lot on where it lives. Since July, a job on an employer's own ATS stayed up for a median of 17 days, and a quarter of them were gone within a week. NHS Jobs, 13 days. Civil Service Jobs, 16 days, and only 2% disappear within a week because they run on fixed closing dates. Reed, 35 days, which is roughly Reed's default posting period. So a Reed listing you found today has, on average, a fortnight longer left in it than one on a Greenhouse board. Apply to the ATS ones first.
The rest of the sheet
Some smaller numbers that didn't earn their own section.
55.5% of listings don't say whether the job is remote, hybrid or in an office. Of the ones that do, 59% are on site, 25% remote, 16% hybrid. By field, remote is a tech and sales thing: 40.6% of software listings with a stated pattern are remote, 40.4% of machine learning, 38.5% of sales, 37% of product and data. Clinical roles, 2.5%. Teaching, 3.6%. Fair enough.
21.7% of listings with a named city are in London. Manchester is next with about an eighth of London's count, then Birmingham and Bristol at half of Manchester's.
August was busier than July. On the sources we ran continuously through the summer, 56,092 new listings appeared in August against 50,600 in July, up 10.9%. Marketing was up 49%, data engineering 25%, machine learning 19.5%, software 15%. Legal and civil engineering were both down 10%. I don't think August normally does that.. worth watching whether September holds it.
6.5% of listings go up at the weekend. Friday is the busiest posting day at 18.2%, and I have no explanation for that one.
To find the 86,000 new jobs we added over the last 30 days, the ingestion read 1.6 million listing rows. Most of what a job board shows you is what it showed you yesterday.
What I'd do with this
If you're job hunting in the UK right now, three things fall out of the numbers.
Don't read a missing salary as a bad salary. In tech it's the default, and the roles that hide pay are, on median, the best-paid ones in the pool. Ask in the first call and move on if they won't say.
Weight the age of a listing. Over a month old and still up on a company's own careers page is a yellow flag. Over a month old on Reed is normal.
And if you're early-career, stop applying to "senior" roles with a covering note about how fast you learn. It's not you. There are twenty of them for every one of yours, and the fix is finding the one, not persuading the twenty.
Flint does that scoring for you, if you'd rather not run the regex yourself. It's the reason I had this table to look at.
I'll rerun these numbers in a month. If the August bump was real, it'll show up as more junior listings by October. That's the number I'm actually watching.
Data: Flint UK job listings snapshot, 8 September 2026. Published under CC BY 4.0. Quote any number here with a link back to this page. If you want the query set or a cut by discipline that isn't above, email chris@getflint.work.