How many job postings are actually fake

Nobody official measures it. That sentence is not a hedge, it is a quotation: the Congressional Research Service, in a briefing for members of Congress published on 25 April 2025, states that "there are no official statistics on the magnitude of ghost jobs."

If you are here because forty applications produced two replies, the practical version of the answer is this. The estimates that actually counted postings rather than surveying people run from about one in seven to about one in five, so most of what you applied to was a real opening, and being ignored is a different problem from being deceived. Now the numbers.

The figures you have been reading do not measure the same thing as each other. Four of them count four different populations: hiring managers describing their own behaviour, recruiters describing their employers, one applicant-tracking system counting its own postings, and one company that actually went and counted job ads. The last of those found the lowest number in the set.

The numbers in circulation, and what each one counted

The third column is the one that decides what each number means. A survey of hiring managers who admit posting a fake listing tells you what share of managers did it, not what share of postings are fake. One manager with a hundred live ads and one manager with two count the same in that survey.

The row that answers the question as asked is the one that counted job ads. Clarify Capital scraped 176,268 Indeed listings in February 2026 and flagged those still active after 30 days: about 14%, one in seven. It is also the most careful of the set about what it claims, describing those listings as "flagged as potential ghost jobs" rather than confirmed. Four percent had been live for more than four months. Wholesale-sector listings came in at 51% and senior-level postings at 21%.

One awkward detail that belongs here rather than in a footnote. Clarify Capital also ran a survey in 2022, using the same 30-day threshold, and found 68% of managers with a posting open that long. Same company, same threshold, self-report against direct measurement, roughly a fivefold gap. The company does not reconcile them, and neither can this article; the plausible explanations are that people over-report, that the denominator changed, or that the market did. Nobody has shown which.

What a job opening is, officially

The other half of this question is a government statistic that is constantly cited and rarely read.

The Bureau of Labor Statistics measures job openings in the Job Openings and Labor Turnover Survey, and it counts an opening only where three conditions hold. A specific position exists and there is work available for it, full or part time. The job could start within 30 days. And the employer is actively recruiting outside workers.

The Congressional Research Service, reading that definition, draws the obvious conclusion: "if employers answer the JOLTS questionnaire honestly, the job openings measured in JOLTS are unlikely to be ghost jobs." It gives a reason beyond good faith, too. JOLTS responses are confidential, so there is nothing to signal to investors or employees by inflating them, and BLS monitors its own data for discrepancies between openings and hires. The one ambiguity CRS does flag is what counts as active recruiting.

JOLTS is also not scraped from job boards. It is an establishment survey that collects data directly from sampled employers, which is the structural reason it cannot inherit a job board's duplicate and stale listings. BLS tested the overcounting theory directly and published the result in August 2026: only 0.27% of JOLTS reports were identified as potential overcounts, accounting for 0.31% of weighted employment, and the conclusion was that "the sharp rise in JOLTS job openings during the pandemic is reflective of real economic change."

Where the headline number comes from, and how to check it in one line

The figure at the top of this search result, and the one Google's AI Overview repeated when we read it on 4 September 2026, is 30%. Its provenance is short and it is worth following to the end.

Forbes published it on 18 November 2025 and says where it got it: "One in three job postings never results in a hire, according to a MyPerfectResume analysis of Bureau of Labor Statistics (BLS) data."

MyPerfectResume publishes the arithmetic. "According to the June 2026 JOLTS report (based on April 2026 data), employers reported roughly 7.6 million job openings but made just 5.1 million hires, leaving more than 2.5 million jobs that never materialized in a single month. That means roughly 1 in 3 job listings lead nowhere."

Now the line from BLS's own release, sitting directly above the numbers being subtracted: "Job openings include all positions that are open on the last business day of the month. Hires and separations include all changes to the payroll during the entire month."

Openings are a count taken on one day. Hires are a count taken across thirty. Subtracting the second from the first does not produce the number of postings that went nowhere; it produces the number of positions still open when the month ended, and next month most of them will be filled by someone. Every search that runs longer than four weeks lands in that gap. So does every role posted on the 28th.

The July 2026 release, published on 1 September 2026, reports 7.3 million openings and 5.1 million hires. Run the same subtraction and it returns 30% again. Which is why the same source can also say the rate "has remained between 28% and 32% for years." A statistic that returns the same answer in a hiring boom and a hiring freeze is not measuring what its name says.

That release also reports total separations of 5.1 million, exactly matching hires. People were leaving payrolls as fast as they were joining them. A standing stock of open positions is what that looks like from the outside.

Every agency and contractor placement sits inside that gap

The same BLS release defines what hires exclude, and the list is specific: transfers or promotions within the location, employees returning from strike, and "employees of temporary help agencies, employee leasing companies, outside contractors, or consultants."

A role filled through a staffing agency does not appear as a hire at the company where the person now works. Neither does a contractor, or a consultant. Every one of those placements sits inside the openings-minus-hires gap, counted as a job that supposedly went nowhere.

If you place people for a living, the number being used to prove that employers are lying about hiring is partly a measurement of your own industry.

CRS has an uncomfortable line for agencies in the other direction, too, and it is fair to print both: "Staffing agencies may also invent and post job openings to show employers they could recruit talented people."

The conflation, in an official document

On 18 June 2026, Senator Ruben Gallego wrote to the BLS, the FTC and the Department of Labor about ghost jobs. The letter cites the ResumeBuilder finding that "40 percent of employers admitted to posting job listings with no intention of filing them," and separately states that job openings have outnumbered hirings "by a staggering 2.2 million a month."

Those two facts come from different universes. The first is a survey of 649 hiring managers about their own conduct. The second is a stock minus a flow in an establishment survey that, by its own definition, excludes positions that do not exist. Placed in one paragraph they read as corroborating each other.

This is a primary record of the error rather than evidence about ghost jobs, and it is the clearest available demonstration of why the population column in that first table matters more than the percentage column.

Why the definition decides the answer

There is no shared definition, and most sources do not say which one they are using.

CRS defines a ghost posting as one "for positions that do not exist, or that employers are not planning to fill immediately." Note the "or": two different situations, one label. Greenhouse says "positions advertised with no intent to hire." MyPerfectResume's recruiters were asked about roles that "either don't exist or are already filled," which folds in a posting left up after a successful hire. Clarify Capital's public definition covers a listing that "stays live long after the role has been filled, or was never real to begin with."

Then there are the operational rules, which make no claim about intent at all. Clarify Capital counted listings active for 30 or more days. ResumeUp.AI applied the same 30-day threshold to LinkedIn and reported 27.4%, reached in this research only through a secondary write-up. Ashby counted postings that closed without a hire, which happens for reasons that have nothing to do with deception: a budget freeze, a reorganisation, a declined offer.

And ResumeBuilder's survey asked hiring managers whether they had posted a fake listing without publishing what a respondent should treat as fake. Each of the 649 people answered against their own private definition, and the answers were then aggregated as one measure.

Employment scams are a fourth thing again, and they are the one genuine fraud category in this whole subject. CRS reports that job and employment agency scam complaints in the FTC's Consumer Sentinel database "nearly tripled from 2020 to 2024." A fake company harvesting personal data is not the same phenomenon as a real company leaving a filled role on its careers page, and the two get counted together constantly.

CRS lists why real employers do it, without deception in most of the cases: to collect resumes for later, to look like they are growing, to show a diverse applicant pool while planning to promote internally, or to see whether someone exceptional turns up. Some are not deliberate at all. Job boards copy postings automatically, so an employer can take its own ad down and find a copy still live somewhere else.

What can honestly be said

The share of US job postings that are not real vacancies has never been measured officially, and every unofficial figure carries a population you have to read before the number means anything.

If you want the closest available answer to the question as most people ask it, one company scraped 176,268 Indeed listings in February 2026 and found about 14% still active after 30 days, which it called potential rather than confirmed. Two applicant-tracking systems, looking only at their own customers, report 18% to 22%. Surveys of hiring managers and recruiters report 40% and 81%, and both are measuring people rather than postings. And the 30% at the top of the page is a subtraction that BLS's own release tells you not to perform.

For anyone reading a posting as a signal, the useful move is not to pick a percentage. It is to read the individual ad, which is what our piece on how to spot a ghost job posting is for, and to know which companies cannot be won at all before spending a week on one.

Every figure in this article was read at source: the Congressional Research Service product IF12977 at congress.gov, BLS's Job Openings and Labor Turnover Survey release of 1 September 2026 and its August 2026 Monthly Labor Review article, Senator Gallego's letter of 18 June 2026, and the published pages of ResumeBuilder, MyPerfectResume, Greenhouse, Ashby, Clarify Capital and Revelio Labs. Where a sample size is not disclosed, this article says so.