Half the ads older than three months were still live in the source the last time our crawler looked. How we can tell, and what the number does not say.
Ghost jobs are one of those arguments nobody can settle, because both sides are guessing. So here is a count from a crawl instead of an opinion.
Of the postings in our corpus that were first published more than 90 days before we measured, 50.8% turned up again at the source in the week before the measurement. Old ads, still being served.
Nothing clever here, just bookkeeping. Every posting we keep carries two dates. One is the publish date the source hands us. The other is the date our own crawler last saw that posting at that source.
Cross the two and you get a small table: old and quiet, old and still moving, new and moving, new and quiet. 17.1% of everything we measured lands in the old-but-still-moving cell.
That cell is the interesting one. A candidate cannot see it, because a candidate sees one snapshot of one ad. Neither can a crawler that visits a site once. You have to come back on a schedule and remember what you saw last time. We do that, which is the only reason we can say anything about this at all.
It does not say the role is filled. We cannot know that from the outside, and neither can anyone reading the ad.
It does not say the ad is fake, either. A requisition that is genuinely hard to fill, a pipelined role, a slow approval chain and a listing somebody forgot to close all look the same from where we sit: published three months ago, still live today.
The phrase "ghost job" carries intent. We can measure age. We cannot measure intent, so we publish the age and label it "still listed" rather than "open". That distinction matters more than it sounds. One of those words we can back up.
We probe employer links one at a time instead of assuming they work, and out of 25,080 of them, 0.62% were dead.
That is a small number, and it kills the simplest explanation. The ads are not broken. They are not dangling pointers to a page that moved on. They are live pages, serving a live posting, months after the posting was written.
Age is information, not a verdict. An ad at 90 days is a different object from an ad posted on Tuesday, even if the text is identical.
The practical move, if you are the one applying, is to ask. Has this been open since June, and is it still going? A recruiter can answer that in one sentence. The ad will not answer it at all, and the silence after you apply is not an answer either.
And if you run hiring: this is worth checking against your own boards. Six postings that were filled in the spring and never taken down look, from the outside, exactly like a company that is growing.
Half of the old ads in our corpus are still being served. That is the finding. What it means for any single application is a question for a human, preferably the one who owns the requisition.
→ Ghost jobs, measured · Ghost index
Our numbers below are ours; these are other people looking at the same question. We paraphrase and link, we do not quote.
Want a cut we have not published? A region, an industry, a role family, a time series, one country against another — we can pull it from the same corpus these posts are measured on.
Tell us what you are looking at: [email protected]. We answer either way.
Source & licence. All figures here come from our own crawl of public job postings, measured against the corpus published in the monthly report and its JSON (CC BY 4.0).
Cite as: DeUnemployment, “The job ads that never come down”, https://deunemployment.com/blog/the-ads-that-never-come-down (data as of 2026-09-20).