A ghost job is an advertised role that an employer does not intend to fill. But the term is also used for a role that never existed, one already filled, one frozen after approval, or one kept open to collect resumes for later. Those are not the same problem. Treating them as one thing is a major reason the estimates vary so widely.
Something changed between postings and hires. But no published estimate audits a representative sample against confirmed hires. Every number is a proxy.
How common are ghost jobs?
The leading estimates use different methods. The largest sample has an unpublished method. The independent research counted interview reviews. The repeated survey number is a screener.
| Source | What was measured | Data and date | Result |
|---|---|---|---|
| Greenhouse | Its own classification of postings as advertised without intent to hire | Postings on its customer platform, reported December 10, 2024 | Greenhouse reported that 18 to 22 percent of postings in a quarter met its definition, and that nearly 70 percent of customer companies had at least one in the second quarter of 2024. |
| Hunter Ng | Interview accounts classified for signals associated with ghost hiring | Roughly 270,000 Glassdoor interview reviews, in an arXiv preprint dated October 29, 2024 | Hunter Ng of Baruch College, City University of New York, projected that up to 21 percent of job ads may be ghost jobs. |
| Clarify Capital | Posting age and hiring managers' stated intentions | 1,045 hiring managers surveyed August 31 to September 1, 2022 | Clarify Capital found that 68 percent had postings active beyond 30 days, while 43 percent said they were not actively hiring. |
| Clarify Capital follow-up | Posting age and employers' stated reasons | 1,000 employers and 200 job seekers; survey and publication dates not given | Clarify Capital found that 33 percent had postings active for at least 30 days, and 75 percent of the job seekers suspected ghost jobs existed. |
| ResumeBuilder | A screening question asking whether a company had posted a fake job | 1,641 hiring managers entered a Pollfish survey launched May 22, 2024 | ResumeBuilder reported that 649 respondents passed the screen and continued, which is 39.5 percent of those who entered. |
| Greenhouse candidate survey | Candidate suspicion, not confirmed incidence | 2,500 workers in the US, UK, and Germany surveyed from April to November 2024 | Greenhouse found that three in five candidates suspected they had encountered one. |
| Revelio Labs with Bloomberg | Postings matched to positions starting within six months | Company, role, seniority, and state matches, published October 31, 2023 | Stephanie Hao and Lisa K. Simon reported that hires per posting fell from about 0.75 in 2018 to below 0.5 by 2023. |
| Indeed | Stated platform policy and employer engagement, with no prevalence estimate | Indeed newsroom statement dated May 29, 2026 | Indeed said it does not allow ghost jobs and that a large majority of postings show ongoing employer engagement, but it published no percentage. |
The famous 40 percent is a screener
ResumeBuilder's June 18, 2024 article is often read as finding that 40 percent of companies posted a fake job. Its Pollfish survey did something narrower. A total of 1,641 hiring managers entered. Only the 649 who said their company had posted a fake listing in the prior year continued. Divide 649 by 1,641 and the result is 39.5 percent, which rounds to 40 percent. That is the pass rate for a screening question on an opt-in mobile panel. It is not a measured rate among US employers or job postings.
Later ResumeBuilder results use those 649 self-identified posters. ResumeBuilder found that 67 percent of that selected group cited appearing open to outside talent. That is not 67 percent of all employers.
The academic estimate does not count postings
Hunter Ng's arXiv preprint 2410.21771, dated October 29, 2024, is the only independent, non-vendor estimate here. Ng, at Baruch College in the City University of New York, classified roughly 270,000 Glassdoor interview reviews with a language model. Markers included no competence questions, salary clarification, or resolution. Ng projected up to 21 percent onto job ads. The observed unit was an interview review, not a posting. This remains the best independent work available.
The largest estimate is the least transparent
Greenhouse, measuring postings on its own platform, reported on December 10, 2024 that 18 to 22 percent in a given quarter were advertised with no intent to hire. Greenhouse says it has more than 7,500 customers, giving this estimate the largest base in the literature. It has not published its classification algorithm, activity threshold, posting sample size, or the quarters included. The sample is broad. The method cannot be inspected.
Why the numbers disagree
Platform analysis counts postings under a private definition. Academic work infers intent from interview language. Employer surveys count admissions. Candidate surveys count suspicion. Posting-to-hire ratios count outcomes without motive. These answer different questions.
Suspicion is not measurement. Greenhouse's survey of 2,500 workers across three countries from April to November 2024 found that three in five suspected an encounter. It cannot establish how many postings were unreal.
Stephanie Hao and Lisa K. Simon, working with Bloomberg on October 31, 2023, matched postings to positions starting within six months. Hires per posting fell from about 0.75 in 2018 to below 0.5 by 2023. Something changed. Revelio Labs said this does not prove intent. Slower hiring, uncertainty, more ads per role, and more applicants can cause the same decline.
Why companies leave postings up
Documented reasons
The Congressional Research Service report by Elizabeth Weber Handwerker and Alexander H. Pepper, dated April 25, 2025, documents several reasons. Employers may build a future talent pool, signal growth, search publicly while planning an internal hire, or wait for an exceptional candidate. Some interview without an immediate plan to hire. Others leave a post behind when circumstances change.
The same report says copies can remain after the source closes. Staffing agencies can advertise openings to demonstrate capacity. Outright employment fraud also occurs, but it is separate from a legitimate employer leaving a stale requisition online.
Survey responses add detail. Clarify Capital's 2022 survey of 1,045 hiring managers found 50 percent kept postings open because they were always open to people. Its undated follow-up of 1,000 employers found 20 percent cited pressure to look active, 20 percent cited cutting costs, and 20 percent cited automated posting for inactive roles. These are self-reports, not verified behavior.
Practitioner explanations, not measured findings
Practitioners quoted by SHRM on October 31, 2024 point to internal hires, mid-process freezes, manual removal that never happens, and copied ads surviving the original. These are expert opinions, not measured shares.
Ng's October 2024 preprint makes the cost argument. Another posting costs an employer close to nothing. An application costs the candidate time and supplies personal data. That imbalance makes pipeline collection easy without an immediate vacancy.
The case that ghost jobs are overstated
Toby Dayton of LinkUp made the strongest published rebuttal on April 9, 2023. Using openings collected from company career sites, he found that the average posting stayed live for 45.6 days. In the same analysis, 950,000 of 5.4 million US openings were at least six months old. They were concentrated in high-turnover fields, including 136,000 retail openings and 49,000 accommodation and food service openings. His argument is that many long-running ads are real, continuously open roles.
SHRM's 2025 recruiting benchmark, based on about 2,300 member responses and published October 20, 2025, put median time to fill at roughly 45 days. Put that beside LinkUp's 45.6-day average posting duration and the consequence is plain: the 30-day threshold used by many ghost-job detectors would classify the median normal hire as fake.
Dayton published before the later Greenhouse and Ng estimates, and LinkUp sells posting data. Still, career-site duration directly challenges the assumption that age equals deception. Evergreen hiring in high-churn work can be genuine.
The April 2025 Congressional Research Service report adds a macroeconomic tension. The federal Job Openings and Labor Turnover Survey counts a specific position with work available, a possible start within 30 days, and active external recruiting. Responses are confidential, so investor signaling cannot work there. If employers answer honestly, those openings are unlikely to be ghosts. Ng uses posting data for a Beveridge Curve argument, while that curve is built from the confidential federal survey.
Even Clarify Capital's own surveys resist a simple escalation story. Its 2022 survey of 1,045 hiring managers found 68 percent had ads active beyond 30 days. Its undated follow-up of 1,000 employers found 33 percent did. The samples are weak and not directly comparable, but the reported direction is down, not up.
What is being done
Ontario has the only binding rule identified here that lets you inspect a posting itself. As of January 1, 2026, the Employment Standards Act 2000, as amended by the Working for Workers Four Act 2024 and implemented through Ontario Regulation 476/24, requires employers with 25 or more staff to state on every public external posting whether an existing vacancy exists. Covered employers must also tell interviewed candidates the outcome within 45 days of the interview or final interview.
New York bill S8877 passed the legislature on June 2, 2026. Sources through June 11 said it awaited the governor's signature. It would require vacancy and timing disclosures. Bills in New Jersey, Kentucky, California, and Pennsylvania have been introduced but not passed. No US federal statute specifically governs ghost postings. The April 2025 Congressional Research Service report says proving intent under general deception law would be difficult.
How to tell if a posting is stale
- Check the employer's careers page. A role absent at the source may be a copy that survived closure. The April 2025 Congressional Research Service report documents this.
- Read Ontario's vacancy disclosure. Since January 1, 2026, a covered posting must say whether an existing vacancy exists. A missing statement is checkable.
- Treat age as a hint.At 30 days, a posting is normal against SHRM's roughly 45-day 2025 median and LinkUp's 45.6-day 2023 average. Even past 60 days, age does not prove intent.
- Treat repeated reposting as a hint. An unchanged role returning for months may be stale. It may also be a real hard-to-fill or high-turnover position.
- Look for operational detail. A location, shift, pay range, department, and start timing suggest a defined requisition. Vague language can indicate pipeline hiring, but real evergreen jobs can also be broad.
- Ask a human. This is the only step that can resolve the question rather than estimate it.
What to actually do about it
Detection is less useful than verification. Find the hiring manager and ask one direct question: is this role approved and actively being filled? You do not need to accuse anyone of posting a ghost job. You need a yes, a no, or a concrete timeline.
A posting nobody will answer a basic question about tells you enough. Move to a role with a reachable person and a current need. If you are not getting interviews, separate that diagnosis from the ghost-job question. Do not let an unverified posting consume more effort.