Where the famous number came from
Three facts, each with its source, because this is a subject where almost everything circulating has none.
The widely quoted claim that 75% of resumes are rejected by an ATS traces to a 2012 sales pitch. A company called Preptel used it to sell resume-optimization software. Preptel was out of business by 2013. No methodology was ever published alongside the figure.
Origin: Preptel marketing, 2012. Company defunct 2013.
"Contrary to popular belief, an ATS rarely 'rejects' a candidate automatically."
That is not a critic talking. It is Jobscan, the largest vendor in the ATS-optimization category, on their own guide to how these systems work. They describe an ATS as a search and filtering tool that recruiters drive, through criteria those recruiters set.
Source: Jobscan, "What Is an Applicant Tracking System (ATS)? The Complete 2026 Guide", 10 April 2026.
Of 25 US recruiters interviewed across tech, healthcare, finance, education and retail, 2 said their ATS was configured to reject applications automatically on content. That is 8%. The other 92% relied on human review, guided by knockout questions and optional match scores.
Source: Enhancv, interviews with 25 US recruiters, September to October 2025.
The one exception worth knowing about is the knockout question, and it is not hidden from you: work authorization, a required licence, a location. A recruiter sets those deliberately and you answer them yourself in the application form.
The problem is a queue, not a filter
Here is the number that actually describes the situation.
Applications per open role reached 244, up 111% since 2022. Over the same period, recruiting teams shrank by 55%.
Source: Greenhouse Benchmark Report, March 2026, North America. Based on data from over 6,000 companies.
Put those two together and the picture is not a machine screening people out. It is a person with a list of 244 and an afternoon, working from the top until they have enough candidates to move forward with. The applications below that point were not rejected. They were not reached.
Nobody deleted the application. Somebody had a few seconds and 243 other tabs.
Which is why an ATS score cannot help. Even a perfect one only tells you the parser understood your file. It says nothing about where you landed in the order, and the order is the part that decided the outcome.
What is worth checking instead
Three checks, all of them about the specific posting in front of you.
1. Which of this posting's requirements your experience actually covers
Not a score against an imaginary robot. The literal list: the requirements this posting states, which ones your own confirmed experience covers, which ones it partly covers, and which ones it does not.
3.5/5 · just below your 4.0 apply-bar. Close. Worth it if you specifically want this employer.
- Likely a down-level. Targets Lead, you present as Director.
- No Media & Entertainment experience (+2 more), often a listed qualification.
- 4 unmet must-haves: Experience implementing CI/CD pipelines for data; Proficiency in Java and SQL (+2 more).
Matched experience
Missing experience
A read that only ever goes up is a sales pitch. This one moves in both directions, and the moment it revises down after you answer honestly is the moment it is telling you something. That is the honest read, and it is more useful than a green number.
2. Whether the terms are demonstrated or merely listed
This is the distinction a score out of a hundred structurally cannot make, and it is the strongest thing on this page.
A keyword sitting in a skills line is a claim. The same keyword inside a bullet, attached to what you did with it and what happened, is evidence. Both would satisfy a keyword count. Only one of them survives a recruiter reading the line, and only one of them gives you something to say when they ask about it.
3. The plain text your file becomes
This is the part of the formatting worry that is completely correct. Parsers do mangle multi-column layouts, tables, text boxes, headers and footers. The result is not rejection. It is that your job title lands in the wrong field, or a role arrives as a jumble, so a recruiter searching their own database for someone like you does not find you.
The check costs nothing: look at what your file becomes once the formatting is gone.
Exports from JobShifu are single column with no tables and no text boxes by construction, and the plain text view is there to be read before you send anything. Nothing about that check is gated. It is a free thing everyone should do once, whatever they use to make the file.
How JobShifu builds the file
It proposes changes to your real resume one at a time, and you approve each one. Every line cites the experience it came from. No source, no line: a line that cannot point at something you confirmed fails the build, and no document is produced.
That is the whole mechanism. It is the reason the keyword read above can tell you a term is unsupported instead of quietly adding it.
Common questions
Does an ATS reject my resume?
Almost never on its own. Jobscan, the largest vendor in the ATS-optimization category, writes that "contrary to popular belief, an ATS rarely 'rejects' a candidate automatically". Enhancv interviewed 25 US recruiters in late 2025 and found 2 of them, 8%, had configured any content-based automatic rejection. The other 92% had not.
The common exception is a knockout question a recruiter set up deliberately, such as work authorization or a required licence. You can see those, because you answer them yourself in the form.
Where did the 75% number come from?
A 2012 sales pitch. A company called Preptel used the figure to sell resume-optimization software and was out of business by 2013. No methodology was ever published with it.
The number outlived the company because it is useful to anyone selling a fix for it. It now circulates with no source attached at all, which is part of why it is so hard to argue with.
Does formatting still matter?
Yes, and this is the part of the worry that is genuinely right. Parsers mangle multi-column layouts, text boxes, tables, headers and footers, and graphics.
The consequence is not rejection, it is misfiling: your job title lands in the wrong field, or a whole role comes through as a jumble, so a recruiter searching their database does not find you. The fix is boring and free. A single-column file with real text, checked by looking at the plain text it becomes.
Should I put keywords in white text?
No. A parser reads the text, not the colour, so hidden keywords arrive in the database exactly like visible ones and then sit in the document a human opens. It reads as an attempt to game the process, which is a worse first impression than a missing keyword, and several ATS vendors flag the pattern.
There is a simpler problem with it too. Getting into a search result for a skill you cannot discuss puts you in a conversation you are going to lose.
What should I check instead of an ATS score?
Which requirements in this posting your experience actually covers and which it does not. Whether the posting's terms appear in your bullets, where they are demonstrated, rather than only in a skills line, where they are merely listed. And what the plain text of your file looks like once the formatting is stripped.
All three are about the job in front of you. None of them is a number out of a hundred.
Sources
- Jobscan, "What Is an Applicant Tracking System (ATS)? The Complete 2026 Guide", 10 April 2026. Quoted verbatim above.
- Enhancv, "Does the ATS Reject Your Resume? 25 Recruiters Explain What Really Happens", interviews conducted September to October 2025.
- Greenhouse, Benchmark Report, March 2026, North America. Applications per open role and recruiting team size, 2022 to 2025.
- Preptel, 2012 marketing material. The company ceased operations in 2013 and the figure has circulated without a published methodology since.
Links verified 5 August 2026. This page is published by JobShifu, which sells a resume tool. Every figure on it is sourced above so the argument can be checked rather than taken on trust.