The claim, and what stands behind it
In February 2019, CNBC Make It wrote that "Three-fourths of all resumes never even get seen by human eyes, according to a study from job search services firm Preptel." The article gave no link to the study and no methodology, such as sample size or how resumes were tracked.
That is where the trail goes cold. Christine Assaf, who writes at HRTact and says she works in organizational psychology, searched for a full hour in 2020 and found no research behind the 75% figure. She found other claims, such as a widely repeated claim, attributed in that post to professor John Sullivan, that ATS software screens out half of resumes, but no study. She traced the figure through a TopResume infographic to a Forbes article that, in her words, "doesn't cite, well, anyone." She also reports that Preptel went out of business in August 2013, so nobody can ask for the study. Her conclusion: "ATS systems screen applicants – they do not reject them."
Treat her work as one researcher's reconstruction. But the burden sits with the number: a statistic printed without a method, from a company selling resume optimization, is a marketing claim, not a finding.
What an ATS does
An ATS is software that stores applications and lets recruiters work through them. Three functions matter to a job seeker.
Parsing. The software converts your file into structured fields: name, contact details, jobs, dates. Workday's administrator guide says "Resume parsing results can vary based on resume format and order of words," and recommends resumes without images or image-based styles. Greenhouse lists what breaks its parser: graphics, photos or word art; tables, headers and footers; and a name and contact block placed in a header, footer or text box. When a parse fails, Greenhouse says the system only attaches the file, and the recruiter must enter your details by hand.
Automatic rejection. The clearest documented case is the knockout question. Greenhouse says applicants who answer a question a certain way "will automatically be rejected," limits the feature to Yes/No, single-select and multi-select questions, and offers it on its Plus and Pro tiers. Its candidate blog says the rules "are set and controlled directly by recruiters," for example a "no" to a work-eligibility question.
Rejection based on a score or skills threshold also exists. Enhancv's recruiter interviews, discussed below, found it was rare and "only appears where someone has deliberately turned it on." Two of 25 recruiters said theirs does.
Search, ranking and scoring. Some systems rank or score applicants; some do not. Greenhouse says that "AI doesn't score or rank applications" in its product, and that "by and large, applications in Greenhouse are reviewed by real people, one by one, in the order received." It adds that "a few common reasons" can keep a resume from being seen, including that when a role is popular, "recruiters may filter the queue."
What recruiters report
Enhancv interviewed 25 U.S. recruiters in September and October 2025. Enhancv sells resume software, the sample is small, and the page does not explain how recruiters were selected, so this is color, not a national statistic. Within those limits: 23 of 25 (92%) said their systems do not auto-reject resumes for formatting, content or design. 44% said their ATS includes an AI or "fit" score, and 36% of all 25 used such scores as a guide only. Charkin Whitehead of Allegis Global Solutions told Enhancv: "The ATS systems that I've worked with don't automatically disposition people—we have to go in and do it."
Where qualified people do get filtered out
The strongest evidence is a 2021 report, "Hidden Workers: Untapped Talent," from Harvard Business School's Project on Managing the Future of Work and Accenture, by Joseph Fuller, Manjari Raman, Eva Sage-Gavin and Kristen Hines. Accenture's public-sector edition says the research surveyed "more than 8,000 hidden workers and more than 2,250 executives" across the US, UK and Germany. Its summary report estimates 27.4 million hidden workers in the United States, meaning unemployed or underemployed people.
On screening, Fuller wrote in HBS Working Knowledge that roughly two-thirds of surveyed employers (63 percent) used a recruitment management system, and "more than 90 percent of employers we surveyed that used an RMS relied on it to make a first cut or rank potential middle-skills (94 percent) and high-skills (92 percent) applicants." Of the hidden workers studied, he wrote, "only one in five made the first cut." He wrote that such systems rely on "negative" filters and gave examples: filters that "exclude candidates without college degrees" or "rank candidates who have not been employed for a six-month period."
The report concerns employer criteria that exclude people, such as gaps and degree requirements. It does not say 75% of anyone is rejected, and it does not say a keyword tweak fixes the problem.
Tools that can help
If a misread resume is the common failure, testing the parse is the useful step. Three options follow.
A manual test, free. Open your resume file, select all text, and paste it into a plain-text editor. If your name and contact details are missing, dates are detached from jobs, or columns are interleaved, a parser may stumble the same way. Greenhouse points to headers, footers, text boxes and tables as usual causes. Also check file size: Greenhouse cannot parse resumes over 2.5 MB.
Jobscan offers a "free resume score" on its homepage. The page does not say how many free scans you get, and its scan buttons link to a sign-up page, so an account may be required.
Disclosure: Oppositioner and ATSGrader are owned by the same company, NexusVibe S.L. ATSGrader's resume parser test is a free browser tool that, per its page, checks the text layer, reading order, contact signals, section labels and DOCX header and footer risk, and processes the file in your browser instead of uploading it. Its methodology page says it is "not any vendor's real algorithm" and that no independent checker can reproduce every employer's private configuration.
That caveat applies to every tool here: a clean result shows your file is readable, not that you will be hired.
"ATS systems screen applicants – they do not reject them."
