Reporters, researchers and regulators on ghost ads and pay silence, set beside 7.4 million indexed postings: listing age, re-crawl presence, pay disclosure.
Every argument about the job market is really an argument about who counted what. That is why this beat has names attached to it, and why a citation tells you more than a number on its own.
Below is the strongest public reporting and research on ghost ads and on employers who keep quiet about pay, set beside what our own crawl measures. We paraphrase and link; we do not quote.
Lauren Weber at The Wall Street Journal mapped the politics: ghost ads, the anger they produce, and the bills in New York and Pennsylvania that answer it. Emma Burleigh at Fortune followed employer ghosting to a three-year high. Jeff Cox at CNBC put the government's openings figure next to the government's hiring figure and let the two speak. Graig Graziosi at The Independent and Ben Kesslen at Quartz both worked through Greenhouse's one-in-five claim. Tim Paradis at Business Insider reported the survey side of the same question. Darian Woods and Wailin Wong at NPR's The Indicator took it to an economist. Lisa Feist at Indeed's Hiring Lab is the person to read on salary disclosure in Europe.
The names matter for a plain reason. The claims are large, the methods differ, and only a citation keeps them straight. A survey of 1,600 hiring managers and a platform's own logs are not the same kind of evidence, and they should never be printed as if they were.
Surveys ask people what happened. ResumeBuilder surveyed hiring managers and found 40% of companies had posted a fake listing in the past year, with 3 in 10 holding active fake listings when they answered. Criteria Corp, reported by Fortune, found 53% of job seekers were ghosted by an employer in the past year — a three-year high, up from 48%. Greenhouse's platform work puts 18–22% of the postings on its own system in the ghost category, and higher in some sectors.
Platform data sees one vendor's slice. Greenhouse sees employers who run hiring on Greenhouse. Revelio Labs, in work done with Bloomberg, matched postings to hires and found the ratio fall from 0.75 hires per posting in 2018 to below 0.5 — every other posting ending without a hire. That is the sharpest version of the claim, and it is a ratio inside their matched sample, not a census.
Academic work tries to classify intent at scale. Hunter Ng's paper ran a language model over Glassdoor postings and put up to 21% of ads in the ghost category, then tied the pattern to the recent disconnect in the Beveridge curve. It is careful work with a visible method. It is still a classifier's opinion about an employer's intent.
Official statistics count the gap, not the cause. Since the start of 2024, openings have outnumbered hires by more than 2.2 million a month, as CNBC reported from the Bureau of Labor Statistics. The gap is real. What fills it is not measured. The Congressional Research Service says as much in plain language: no government statistics agency publishes a ghost-job count.
DeUnemployment crawls public job postings — company career pages and applicant-tracking systems that publish openly. As of September 20, 2026: 7,415,865 postings indexed, 3,624,739 with a full description, across 82 countries.
We do not publish a ghost-job count. An employer's motive is not visible from outside, and anyone selling you a motive from outside is inferring. What we publish instead is two facts about each ad: how old it says it is, and whether our crawler still found it being served at the source.
Median listing age is 27 days, and 29.8% of postings are older than 90 days, measured over 5,983,556 dated postings. In the crawl archive — 6,564,432 postings, 2026-09 edition — 33.6% are older than 90 days. Of those, 50.79% were still being served by the employer's own page within the last seven days. Old and still served at once: 17.07% of everything measured. → /data/ghost-jobs · /data/ghost-index
The cheap explanation — that old listings are just broken links — fails against the data. We probe employer URLs one address at a time: 0.62% of 25,080 were dead, and 0.8% sat behind a bot wall, which we record as unknown rather than dead.
That is where we stop. An old ad still being served is a measurement. Calling it a ghost is a claim about someone's intent, and we do not have it.
Only 12.4% of postings state any pay figure. Of the ones that do, 7% give a clean min–max range, 11.1% of those ranges are wider than $50,000, and 5.9% answer with "competitive" or DOE.
The country gap is wider than the legal gap: 24.36% in the UK, 0.47% in Germany, 0.37% in Switzerland. → /data/pay-transparency
Regulators are moving on both continents. Article 5 of the EU Pay Transparency Directive, Directive (EU) 2023/970, gives applicants the right to receive pay information from the prospective employer before an interview, and member states have to transpose it by June 2026. Ontario went further and faster. Since 1 January 2026, employers there with 25 or more staff must state expected pay, say whether the vacancy is real, disclose any use of AI in hiring, and reply to interviewed candidates within 45 days. New York's S8877, passed on 2 June 2026 and awaiting signature, would require fill timelines and removal within two weeks of a hire. Pennsylvania has bills in play.
Read Ontario's list once more. Two of those four duties are already fields in our crawl: pay stated, and how long a listing survives. The third, whether the vacancy is real, is a field nobody publishes at scale yet. If that clause spreads, the ghost-job argument stops being a question of survey design and becomes a string in an upload feed.
Indeed's Hiring Lab found that in Europe salary ranges dominate among postings that disclose anything at all. Transparency is improving inside the group that was already transparent. That is the easier half of the problem, and it is worth saying so.
The platform matters as much as the employer. Workday postings have a median listing age of 12 days — but 47.76% of them arrive with no publish date at all, so they cannot be dated and are excluded from that median. Greenhouse postings sit at 58 days, Paycom at 69, Lever at 101. Same market, different behaviour, depending on whose software the employer uses. → /data/hiring-platforms
Two mechanics matter whenever someone quotes a count of open roles. 9.3% of postings appear on more than one board, so a raw count of listings overstates the number of distinct roles before a single ghost enters. And the market is atomised: 408,086 distinct employers, with the top 20 holding 6.3%. Duplication is real, and it is far too small to explain a gap of 2.2 million a month. We will not stretch it to fit. → /data/hiring-market
Quote us for: how long ads stay listed, how many are still being served at the source, how often pay is stated, how platforms differ, how often the same role is posted twice. Every one of those figures carries a sample size and a written definition.
Do not quote us for how many jobs are fake. We measure ads. We do not measure intentions, and neither does anyone else who is being straight with you.
Every number here comes from one index, rebuilt daily, and each figure has a page with its sample attached. Start at /data/key-figures, or take the raw files — /report/data.json and /report/data.csv — under CC BY 4.0. The method is at /about-data. What the corpus can and cannot vouch for, including link liveness and coverage, is at /data-health.
Working a different angle? If you are reporting on ghost ads, pay disclosure or a specific market, we can cut the numbers for you: a country, a role, a platform, a time window, or the same measure across two markets. Write to [email protected] and say what you need. We pull it, publish it with its method, and hand you the figures. All we ask is a citation.
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, “Ghost jobs, silent salaries and the laws arriving — what the reporting says, and what a crawl can check”, https://deunemployment.com/blog/ghost-jobs-what-the-reporting-says (data as of 2026-09-20).