ATS Guide · 2026-08-13
10 ATS Myths, Fact-Checked with Primary Sources (2026)
Quick answer
The biggest ATS myths are that 75% of resumes are automatically rejected, every system uses one score, PDFs are unreadable, and a public checker reveals the employer’s result. None is universally true. ATS products differ: they may parse, search, filter, match, rank, or automate employer-defined criteria. The useful strategy is clean parsing, truthful job alignment, and evidence—not gaming a magic number.
ATS advice is unusually vulnerable to repetition. A vendor publishes a percentage, a blog removes the caveat, and a decade later the number is presented as an industry fact. This page checks ten common claims against original reporting and current documentation from Greenhouse, Workable, Workday, Jobscan, and Harvard Business School.
Our rating scale is simple: False means the claim conflicts with available evidence; Misleading means it contains a real risk but turns it into a universal rule; Partly true means it applies to some systems or workflows, not the whole ATS category.
The ten ATS myths at a glance
| Claim | Verdict | What the evidence says |
|---|---|---|
| ATS rejects 75% of resumes | False | No universal study supports the number |
| A bot always decides who gets rejected | Misleading | Automation depends on product and employer configuration |
| Every ATS assigns a resume score | False | Some score; others search and filter without one |
| 70%, 75%, or 80% is the universal pass mark | False | Threshold advice belongs to a particular checker, not the industry |
| ATS only understands exact keywords | Partly true | Exact search exists, but newer matching also evaluates criteria and related skills |
| ATS cannot read PDF resumes | False | Text-based PDFs are accepted; layout and text layer matter |
| Columns automatically reject a resume | Misleading | Columns can damage parsing; a parse problem is not an automatic rejection rule |
| More keywords always produce a better resume | False | Repetition can raise one tool’s score while weakening credibility |
| A public checker shows the employer’s score | False | The checker and employer run different systems |
| ATS-friendly and human-friendly are opposites | False | Clear structure and concrete evidence help both passes |
Myth 1: “ATS rejects 75% of resumes before a human sees them”
Verdict: False as a universal statistic.
The traceable source is a 2012 Computerworld article that attributed the figure to Preptel, a resume-optimization service. The article presents no sample size, study design, or calculation for the percentage[1]. Repetition does not turn that vendor claim into measured industry data.
A stronger source tells a more precise story. Harvard Business School and Accenture surveyed 2,250 executives and found that more than 90% used recruiting systems to filter or rank candidates; 88% agreed that qualified candidates were being screened out because their profiles did not match the exact criteria used[2]. That is a serious filtering problem, but it is not proof that one universal bot automatically rejects exactly 75% of submitted resumes.
Myth 2: “A bot always decides who gets rejected”
Verdict: Misleading.
Greenhouse now documents AI-generated match categories, but states that Talent Matching is assistive: it does not automatically advance or reject candidates, and recruiters remain responsible for the decision[3]. Workday likewise says its AI provides qualification-match insights while customers retain control and human oversight[10]. Those statements disprove the claim as a universal description of ATS.
Automation does exist in some workflows. Workable documents AI matching degrees and recruiter sorting, while its broader agent product can act on employer-defined criteria when the employer enables those actions[5]. Screening-question answers can also trigger configured disqualification rules. The accurate answer is therefore system-specific: do not assume either “a bot always rejects” or “automation never rejects.”
Myth 3: “Every ATS assigns a resume score”
Verdict: False.
Workable's Screening Assistant can calculate how many selected job criteria a candidate meets and let recruiters sort by best match[5]. Greenhouse offers optional Talent Matching categories, while its separate Talent Filtering workflow uses exact keyword searches, required and preferred terms, and filters such as location or scorecard status[3][4]. A score is one possible, configurable interface—not a defining feature of every ATS or every employer account.
Myth 4: “70%, 75%, or 80% is the universal pass mark”
Verdict: False.
A percentage only has meaning inside the model that produced it. Jobscan recommends its users aim for a particular range inside Jobscan's own Match Rate[8], and its support documentation lists the inputs in that proprietary result[7]. Another checker may grade different factors, and an employer's private configuration may not produce a comparable score at all.
This is why there is no cross-product conversion between a 76 in one checker and an 89 in another. Read the honest answer to “what is a good ATS score?”and compare revisions within one documented rubric.
Get a category breakdown—not a mythical pass/fail verdict
Free scan · no signup · your resume never leaves your browser
Check my resume free →Myth 5: “ATS only understands exact keywords”
Verdict: Partly true.
Exact terminology still matters in workflows that use literal search. Greenhouse states that a searched keyword must exactly match the application and lets recruiters combine required and preferred terms[4]. If the posting says “PostgreSQL” and your resume only says “databases,” a literal search can miss you.
But exact keyword counting is not the entire market. Workable's current screening tools evaluate selected criteria across education, experience, skills, and other requirements[5]. Use the employer's terminology where it is accurate, but support each term with evidence. Do not assume repeating one phrase simulates every matching system.
Myth 6: “ATS cannot read PDF resumes”
Verdict: False.
Greenhouse's own parsing documentation treats .docx and .pdf as normal documents and distinguishes them from image uploads. The documented failure risks include files over its size limit, images instead of documents, complex formatting, graphics, headers, footers, text boxes, and columns[6]. The useful distinction is not “PDF bad, Word good”; it is extractable text and predictable reading order versus a visually complex or image-only file.
Follow the detailed PDF vs Word decision guideand always obey the file types listed by the application portal.
Myth 7: “Columns automatically reject a resume”
Verdict: Misleading.
Columns are a parsing risk, not a universal rejection command. Greenhouse lists columned layouts among formats that can produce an unsuccessful or partial parse[6]. A damaged parse may put dates, titles, or skills into the wrong fields and make search less reliable. That is enough reason to prefer one column; there is no need to invent an automatic-rejection rule.
Myth 8: “More keywords always produce a better resume”
Verdict: False.
Keyword coverage can improve a job-specific match, but repetition is not evidence. Jobscan itself warns about over-optimization and recommends presenting experience and skills naturally rather than stuffing terms[8]. A useful edit adds a missing skill you genuinely have inside a bullet that explains what you did and what changed.
Use the resume keyword workflowto separate must-haves from nice-to-haves and avoid adding claims you cannot defend in an interview.
Myth 9: “A public ATS checker shows the employer’s score”
Verdict: False.
A public checker runs outside the employer's account and applies its own rules. Jobscan describes its Match Rate as a visualization that helps a user measure resume-to-job alignment and explicitly notes that the result is not an employer score[9]. The employer may use another match model, manual search, screening questions, or no numeric score.
See the full resume score vs job match score comparisonbefore treating any percentage as an external verdict.
Myth 10: “ATS-friendly and human-friendly are opposites”
Verdict: False.
The shared middle is a clean document with standard sections, explicit skills, readable bullets, and specific outcomes. Those choices make extraction more predictable and make a fast human review easier. Ladders' eye-tracking study found that an initial recruiter screen lasts only seconds[11], so clarity is not a concession to software—it is part of persuading the person after the software.
What to do instead of trying to “beat the ATS”
- Use a text-based PDF or DOCX. Follow the portal's explicit format request.
- Keep one reading order. Use standard section names and avoid putting essential facts in graphics, headers, footers, or text boxes.
- Tailor to one real vacancy. Mirror accurate terms for must-have skills, tools, credentials, and scope.
- Prove the terms. Put important skills inside bullets that state an action, context, and outcome.
- Check categories, not only the headline score. Fix parsing, structure, evidence, and missing job terms separately.
- Answer application questions carefully. Employer-configured knockout criteria can matter more than resume keyword density.
Frequently asked questions
Is it true that ATS rejects 75% of resumes?
No published study establishes a universal 75% ATS rejection rate. The traceable 2012 source is a Computerworld article attributing the claim to resume-optimization vendor Preptel without presenting a sample or methodology. Real systems and employer configurations vary widely.
Do all applicant tracking systems score resumes?
No. Some products provide AI matching or ranking, while others rely on recruiter searches, filters, screening questions, and human scorecards. Even where a score exists, its inputs and employer configuration are product-specific.
Can an ATS read a PDF resume?
Yes. Modern systems commonly accept text-based PDFs. The real risk is a missing text layer, excessive file size, or complex formatting such as columns, graphics, headers, footers, and text boxes that causes a partial or failed parse.
Is a public ATS checker score the score an employer sees?
No. A public checker applies its own rubric outside the employer’s account. The employer may use a different matching model, manual keyword filters, screening questions, or no numeric score at all.
Sources
- [1] Computerworld — the 2012 article that attributed the “75%” claim to Preptel
- [2] Harvard Business School & Accenture — Hidden Workers: Untapped Talent (2021)
- [3] Greenhouse — Talent Matching (official product documentation, 2026)
- [4] Greenhouse — Talent Filtering (official product documentation)
- [5] Workable — Using the AI Screening Assistant (official product documentation)
- [6] Greenhouse — Unsuccessful resume parse (official product documentation, 2026)
- [7] Jobscan — What exactly is being checked? (official support, 2026)
- [8] Jobscan — What Match Rate should I aim for? (product-specific guidance)
- [9] Jobscan — Resume Scanner and Match Rate explanation
- [10] Workday — Demystifying AI in Hiring (official product guidance)
- [11] Ladders — recruiter eye-tracking study update (2018)
For the rules behind our own result, read the public ATSGrader methodology. For the practical next step, use the pre-application ATS checklist.