So how many resumes are actually rejected by ATS?

Nobody knows, and you should be suspicious of anyone who gives you a precise number. Applicant tracking systems are private software used differently at every company; there is no public dataset covering them. The honest answer is that the widely quoted figures are not measurements.

Do ATS systems reject resumes at all?

They can filter, and companies configure filters differently. A recruiter can search by keyword, set a minimum score, or simply sort a list. What does not happen is an automatic rejection notice — you are usually just never surfaced, which feels identical from the outside.

What actually does get resumes discarded?

The mechanical failures are the ones you can control: unparseable layouts, missing the exact terms the posting uses, and dates or titles that cannot be matched to anything. Beyond that it is human judgement about fit, which no formatting trick changes.

Should I optimise my resume for ATS then?

Yes, but for the boring reasons: clean single-column structure, standard section names, and the posting's own vocabulary where it is accurate. That work is worth doing whether or not any particular statistic is true.

Why does the 75% figure keep appearing?

Because it is useful. It creates urgency, it is short, and every site that repeats it links to another site that repeats it rather than to a study. That is how a marketing claim becomes quoted as common knowledge.

Do ATS Systems Really Reject 75% of Resumes?

No — and the number does not come from a study. Here is what it traces to, what is actually true about automated screening, and why the distinction matters for how you write your resume.

The claim, and where it comes from

You have probably read that 75% of resumes are rejected by applicant tracking systems before a human sees them. It appears in career blogs, university career-centre handouts, resume-tool marketing pages and — today — in AI answers.

The figure does not come from research. Investigative work tracing its citation chain — summarised in this 2026 compilation of ATS statistics and an earlier investigation by Christine Assaf that was cited by Ask a Manager, HiringThing and HR.com — points to a small ATS vendor that used the number in sales material around 2012. That company stopped operating in 2013 and never published a study. Every later citation, however confident, traces back to that.

We are naming the mechanism rather than repeating the number: a marketing claim became quoted as common knowledge, and each retelling cites another retelling instead of a source.

What we can and cannot tell you

We run an ATS keyword check, so it would be convenient for us if this statistic were true. It is still not something we will repeat as fact. Here is the honest split:

We do not know the real proportion of resumes filtered out automatically. Applicant tracking systems are private software, configured differently at every company. No public dataset covers them. Any precise figure you see — including a smaller, more reasonable-sounding one — is unsourced.
We do know that filters exist and are used. Recruiters can search by keyword, sort by score, or set minimum criteria. From an applicant's side this looks the same as rejection: you are simply never surfaced.
We do know that some failures are mechanical and therefore fixable — and those are worth your attention regardless of any statistic.

If a tool tells you "your resume will be rejected by 75% of ATS systems" or gives you a precise rejection probability, treat it as a sales technique rather than information.

The failure modes that are actually real

Strip away the invented numbers and there is a short, useful list. These are the things that genuinely cause a resume to be filtered or misread:

Content the parser cannot read. Two-column layouts, tables used for positioning, text inside images, contact details in a header or footer. The words may be correct and still not exist as far as the system is concerned.
Vocabulary that does not match the posting. "Postgres" and "PostgreSQL" are different strings to a matcher. This is a matching problem, not a quality problem — and it is the one most worth fixing.
Dates and titles that cannot be normalised. Mixed date formats, abbreviations, or invented job titles make it impossible to compute tenure or fit you to a role family.
No evidence attached to the claims. This one is human, not mechanical: a bullet with no outcome reads as a job description rather than a contribution.

Notice that none of these depend on the myth being true. That is the point — the 75% figure is doing marketing work, while these four items are doing actual work.

What to do instead of worrying about the number

Put your resume and the posting side by side and check the terms, spelling included.
Confirm the file parses: select all, paste into a plain-text editor, and see whether the order survives.
Use one column and standard section names. Start from the free ATS-friendly resume template if you are rebuilding.
Write the first two lines deliberately — the summary examples page covers what those lines should do.
Re-check after editing. Compare the pattern of missing keywords rather than chasing a score: our own score varies between runs, and we say so on the page.

Common questions

So how many resumes are actually rejected by ATS?: Nobody knows, and you should be suspicious of anyone who gives you a precise number. Applicant tracking systems are private software used differently at every company; there is no public dataset covering them. The honest answer is that the widely quoted figures are not measurements.
Do ATS systems reject resumes at all?: They can filter, and companies configure filters differently. A recruiter can search by keyword, set a minimum score, or simply sort a list. What does not happen is an automatic rejection notice — you are usually just never surfaced, which feels identical from the outside.
What actually does get resumes discarded?: The mechanical failures are the ones you can control: unparseable layouts, missing the exact terms the posting uses, and dates or titles that cannot be matched to anything. Beyond that it is human judgement about fit, which no formatting trick changes.
Should I optimise my resume for ATS then?: Yes, but for the boring reasons: clean single-column structure, standard section names, and the posting's own vocabulary where it is accurate. That work is worth doing whether or not any particular statistic is true.
Why does the 75% figure keep appearing?: Because it is useful. It creates urgency, it is short, and every site that repeats it links to another site that repeats it rather than to a study. That is how a marketing claim becomes quoted as common knowledge.