Research
Ghost jobs, job scams and ATS filtering. Every figure below names who published it and when, and says what it does not support — because most of these numbers are quoted far beyond what their source actually claims.
Last reviewed · Compiled by the inteller.ai Research Team · how we work
These are not our numbers. They are third-party figures we cite. We do not publish first-party statistics from our own scan data yet, because our volume is not large enough for that to be honest. Our methodology page explains what we can and cannot measure.
of job postings may be ghost jobs — listings with no role behind them, or none the employer intends to fill.
Source: Resume Builder; Clarify Capital; ZipRecruiter (2024–2025)
What it does and doesn’t mean: This is a survey figure, not a census. It comes from employers self-reporting that they keep listings up without intent to hire, so it measures admitted behaviour rather than audited postings. Treat it as the right order of magnitude, not a precise rate.
of resumes are filtered out before a human sees them.
Source: Harvard Business School, “Hidden Workers: Untapped Talent” (2021)
What it does and doesn’t mean: The most-cited number in this category and the one most often distorted. The report describes how automated screening excludes capable candidates; it is not a claim that three quarters of resumes are auto-rejected by a keyword score. You will also see 77% quoted for this — we have not found a source for that figure and do not use it.
of Fortune 500 companies use an applicant tracking system to filter resumes.
Source: Jobscan (2024)
What it does and doesn’t mean: Reliable for large employers and routinely over-extended to all employers. Small companies frequently review manually, so ATS-first advice matters much more the larger the company you are applying to.
of hiring managers say they can recognise an over-optimised, keyword-stuffed resume.
Source: Resume Builder survey (2025)
What it does and doesn’t mean: This is about human detection of keyword stuffing. It is not a statement about how many resumes an ATS rejects — the two are regularly conflated because the numbers are close.
of hiring managers say they can identify an AI-generated cover letter.
Source: ResumeGenius employer survey (2025)
What it does and doesn’t mean: Self-reported confidence, not measured accuracy. Nobody has tested these managers against a controlled set, so read it as the risk of sounding machine-written, not as a detection rate.
of job seekers used AI tools to write application materials in 2025.
Source: Resume Builder survey (2025)
What it does and doesn’t mean: The reason generic AI output no longer differentiates anyone. If half the pile is AI-written, sounding like the pile is the failure mode.
of job seekers say never hearing back is their single biggest frustration.
Source: ZipRecruiter Research (2026)
What it does and doesn’t mean: Relevant to ghost listings: silence is the experience a ghost job produces, and candidates cannot tell it apart from ordinary rejection.
increase in reported job scam losses between 2020 and 2023.
Source: US Federal Trade Commission; FBI Internet Crime Complaint Center (IC3) (2020–2023)
What it does and doesn’t mean: Based on reported losses, so it is a floor rather than a total — most victims never file. Rising reporting rates account for some of the increase.
Research from Resume Builder, Clarify Capital and ZipRecruiter across 2024–2025 puts it at roughly 43% of postings being ghost jobs — listings with no role behind them or none the employer intends to fill. That figure comes from employers self-reporting the practice, so treat it as the right order of magnitude rather than a precise rate. A posting being a ghost job is not the same as it being a scam: most ghost jobs are real companies collecting resumes, not criminals.
75% is the figure with a source behind it — Harvard Business School's 2021 "Hidden Workers" report. The 77% version circulates widely but we have not found an origin for it, and it appears to be a collision with a separate 2025 Resume Builder finding that 77% of hiring managers can spot a keyword-stuffed resume. We use 75% and attribute it.
No. Every figure on this page is third-party and attributed to the organisation that published it. We do not yet publish first-party statistics from our own scan data, because our scan volume is not large enough for that to be honest. We will say so on the day that changes.
It was last reviewed on 3 September 2026. We revisit it when a cited source publishes new data, and we remove figures rather than carry them forward once we cannot stand behind them.
Paste it into the free checker and see which ghost-job and scam patterns it actually matches. No signup.
Check a job postingCiting this page? Please link to it rather than reproducing the table, so readers reach the caveats attached to each figure. If you believe one of these numbers is wrong, or you have a better source, tell us and we will correct it.