We Measured the Hidden Job Market. It's About 18%.

Every career guide you have ever read contains a version of the same sentence: 70, or 80, or 85 percent of jobs are never advertised. It is always stated as settled fact, and it never has a footnote. Follow it far enough and you land on articles citing articles citing a study nobody can produce — the trail runs back through 1980s outplacement advice to a survey that, if it existed, nobody has seen in forty years. College Recruiter went looking for the source of the 80% version and came back empty-handed. Our own take on how these numbers get made is in a previous post about applying early.

The annoying part is that the underlying idea is not wrong. There really are roles you will not find on the aggregators. We just had no idea how many, so we counted the ones we can see.

The answer, on 20 August 2026, is about 18%.

What we actually counted

We track 303,112 open roles at 21,836 companies. Every row carries a source label saying where we read it — which vendor feed, or which careers page, or which government service. That label is the whole reason this number can be checked rather than believed.

Of those roles, 55,818 — 18.4% — at 12,978 companies do not come from a vendor applicant tracking system feed. They come from one of two places:

Company careers pages we read directly. Not a feed, not an API, just the page an employer publishes and updates themselves.

The national employment services of Sweden, Germany and France — Arbetsförmedlingen, the Bundesagentur für Arbeit, and France Travail. Public job registries, run by the state, that the big aggregators largely do not ingest.

13,211 of those 55,818 appeared in the last 24 hours of the snapshot. This is not an archive of stale postings. It is a live slice of the market that a board-first search does not reach.

One more number that matters more than it looks: we have read 52,009 of the 65,344 careers pages we know about, which is 80% coverage. The remaining 20% is not empty. So 18% is a floor, not a ceiling — the real figure is higher, and we will not know how much higher until the crawl finishes.

What this number does not prove

This is the part most articles about the hidden job market get wrong, so let us be exact about what we measured.

Our data proves these roles do not flow through the vendor ATS feeds that job boards and aggregators ingest. That is a claim about a distribution channel, and it is the only claim the data supports.

It does not prove those roles are absent from LinkedIn or Indeed. Nothing stops a company from posting a job by hand to a board while its careers page stays the only machine-readable source, and plenty of companies do exactly that. Which of these 55,818 roles are missing from any particular board is a question our data cannot answer — and if we answered it anyway, you would be right to stop trusting the rest of the number.

So the honest version is narrower and still useful: if your entire search runs through aggregators, there is a large, live set of roles you are structurally unlikely to see — not because they are secret, but because the pipe that fills your search does not carry them. Whether any individual role also got posted somewhere by hand is a coin flip we cannot call for you.

We are also only counting what we can reach. Roles filled purely through referral, before anything is written down anywhere, are invisible to us — as they are to everyone else with a crawler. If a version of the hidden job market exists that is genuinely 80%, it lives there, and no one has measured it, including the people who keep quoting the figure.

Who the hidden roles belong to

The 55,818 skew small. Not exclusively, but strongly, and for reasons that make sense once you see them.

Running a Greenhouse or Workable subscription is a line item. A company hiring four people a year does not have one. Its openings live on a careers page somebody edits by hand, or in a Notion doc, or as a paragraph on the About page — and nothing about that setup produces a feed for an aggregator to consume. The postings are public. They are just not machine-readable by anyone who has not gone looking.

Then there is the state-registry habit, which is regional rather than about size. In Germany, Sweden and France, posting to the national employment service is normal practice — sometimes near-universal for public sector and mid-sized employers. A German engineering firm may put a role on the Bundesagentur and its own site and consider the job done. It is thoroughly advertised. It is just not advertised anywhere a US-centric aggregator is looking.

This is why the hidden job market is not one thing. It is a small company with no recruiting budget, and a Swedish municipality following a national norm, and those two have nothing in common except that your job board never heard of either.

This is the reason we started reading careers pages directly instead of aggregating boards, and the resulting index is free and public on GitHub — you can check the source label on every row yourself.

Where to look instead

None of this requires a product. It requires knowing where the pipe does not reach.

Go to careers pages directly. Pick fifteen companies you would actually work for and bookmark their jobs pages. This is tedious and it is also the single highest-yield thing in this article, because it puts you in front of the posting on day zero rather than whenever a board picks it up. The 48-hour post covers why that gap matters so much.

Use the national employment service where you are looking. If you are job hunting in or into Germany, Sweden or France, these are large public registries and most people searching in English never open them:

• Germany — Bundesagentur für Arbeit
• Sweden — Arbetsförmedlingen
• France — France Travail

They are clunky, they are mostly in the local language, and a browser translation handles them fine. The listings are real and frequently not duplicated anywhere you are already looking.

Expect fewer applicants, not easier hiring. A role that never hit an aggregator has a smaller pile in front of it. That is a genuine advantage and it is not the same as an easy job to get — the bar is whatever the bar is. You are buying a better queue position, not a lower standard.

Questions people ask

Is the hidden job market real?

Yes, in the specific sense that a large share of live openings never enters the feeds aggregators consume — we measure that share at about 18%, and rising as our coverage grows. No, in the sense the phrase is usually used, where it means 80% of jobs are secret and only networking can reach them. That version has never been measured by anyone.

What percentage of jobs are never advertised?

Nobody credibly knows, because "never advertised" is unmeasurable — you cannot count what was never written down. What is measurable is the share that skips the aggregator pipeline, which in our index is 18.4% as of 20 August 2026.

Do these jobs show up on LinkedIn?

Some certainly do, posted by hand. Our data cannot tell you which, and any source claiming a precise figure for that is guessing. What we can say is that they are not arriving there through the automated feeds that fill most search results.

Is it worth checking government job sites?

In Germany, Sweden and France, yes, and more than most English-speaking job seekers assume. In countries without a strong state registry habit, much less so. It is a regional answer, not a universal one.

---

The numbers above are a snapshot from 20 August 2026 and will have moved by the time you read this. The index they come from is published daily and free, with the source label intact on every row, at github.com/trylynceus/jobs — go and disagree with our arithmetic. If you would rather search it conversationally than read a CSV, that is what Lynceus is for.