Do ATS systems auto-reject 75% of résumés?

The number with no source

The claim travels fast: "75% of résumés are auto-rejected by an ATS before a human ever sees them." It appears in career coaching content, LinkedIn posts, résumé-tool marketing, and the occasional news feature. It is cited with the confidence of an established fact.

It is not one. The figure traces back to a 2012 sales pitch by Preptel, a company that positioned its product as an ATS optimizer. Preptel shut down in August 2013 — about a year after the claim began circulating. No methodology was published. No underlying study exists. What spread was a marketing line, stripped of its commercial origin, laundered through a 2013 funnel-style article that was then picked up by subsequent pieces including a 2021 Wall Street Journal variant. By the time it reached TikTok and LinkedIn in the mid-2020s, it had accumulated enough repetition to read as a sourced finding.

Christine Assaf, who traced this origin chain in detail, put it plainly: your application was rejected by a human, not a computer. If anything, that framing is generous — most résumés are never opened at all, so genuine rejection is rarer still.

What recruiters actually do

In one study of 25 US recruiters — a small sample that should be read as directional rather than representative — Enhancv found that every recruiter in the group used knockout questions as a hard mechanical filter. These are the binary gates: work authorization, minimum years of experience, geographic location. Answering "no" to a required field does result in automatic removal, and that is real.

What the same study found about content-based filtering is different. Only 8% of the recruiters surveyed — two out of twenty-five — had configured their ATS to do any content-based auto-rejection at all. The remaining ninety-two percent relied on knockout questions, boolean keyword searches, and manual scanning. An ATS score or AI fit summary might appear on a recruiter's dashboard, but more than half of the recruiters in this study said they ignored it; another third used it only as a loose reference point.

The "75% content auto-reject" claim describes a world where the ATS is making nuanced decisions about résumé substance before a recruiter gets involved. The evidence from people who actually run these systems suggests that world is largely fictional. The real hard filter is the knockout question, and it is transparent: you know if you answered no to a binary requirement.

The real filter is volume, not the algorithm

The mechanism that buries most résumés is not algorithmic rejection — it is arithmetic.

Software and technology roles receive approximately 369 applications each on average, of which roughly 5% meet the stated qualifications on the posting. A recruiter working an active requisition faces hundreds of inbound submissions within the first few days. The practical response is not a sophisticated content-scoring pass across all 369 candidates. It is stopping active review after the first wave — the first hundred or hundred-fifty that come in — and moving those to the next stage. Everyone who applied on day four is not rejected. They are simply in a stack that no longer has any reason to be opened.

This distinction matters for how you think about optimizing your application. If the problem were algorithmic content-filtering, the solution would be keyword density, format compliance, and ATS compatibility. Those things are not irrelevant — a résumé that fails to parse cleanly is genuinely disadvantaged. But if the real bottleneck is being visible early enough to land in the first wave before the recruiter's attention moves on, the optimization problem is different. It is about timing, about applying to roles where your qualifications are close enough to the center of the posting that a recruiter scanning quickly can see the match in the first few seconds, and about ensuring your document reads clearly when someone is skimming rather than reading.

The goal is not to beat the bot. It is to rank high enough in the first wave that the bot question never comes up.