The number you keep hearing
You have seen it: something like three in four résumés get thrown out by a machine before a human ever reads them. It gets quoted in LinkedIn posts, in résumé-service ads, and in the pitch for nearly every tool in this category, including tools that compete with us.
We went looking for where that figure comes from. Our research corpus, which pulls from official vendor documentation, parser-vendor engineering posts, and recruiter community reports, does not contain a primary source for it. That is worth sitting with. A statistic repeated everywhere, attached to nothing you can check, is not evidence. It is a slogan. We are not going to publish it as a fact, because we cannot trace it, and this whole product is built on not bluffing about numbers.
Our rule for ourselves: if we cannot point you to the source, we do not print the number. That applies to the scary stat too, even when the scary stat would help us sell.
What the record actually shows
When you go to the sources that do exist, the picture flips. In a recruiter survey referenced in our corpus, 92% of recruiters said Workday, one of the most widely deployed systems in the world, does not auto-reject a candidate for a missing keyword on its own. Rejection came from a failed knockout question or from a parse that left the profile incomplete. That is the opposite of a keyword bot silently trashing your file.
Source: RESEARCH.md, Workday ranking, SHRM finding [HIGH confidence].
Across the six systems we model, base-tier keyword search is just that: search. On Greenhouse, the co-founder confirmed there is no automated ranking at all in the base product; every advance is a person's decision. There is no hidden reject score doing the work people imagine.
Source: RESEARCH.md, Greenhouse ranking, co-founder confirmed [HIGH confidence].
So what does the rejecting?
Two things, mostly, and neither is the formatting bogeyman.
These are the yes/no screening questions on an application: work authorization, a minimum years-of- experience threshold, a required certification. Answer one the wrong way and you can be archived before a human sees you. On Workday they are described as the number-one cause of rejection, and they look identical to the optional preference questions, so people fail them without realizing it. On Taleo they run before the résumé is even read.
Source: RESEARCH.md, Workday and Taleo ranking [HIGH confidence].
Parsing does not reject you. It buries you. If a table scrambles your job titles, or your contact info sits in the header where every system skips it, or a legacy parser drops a whole section because your heading was not on its whitelist, you are still in the database. You are just not findable, or not findable as the person you actually are. On Taleo, missing one required keyword drops you to the "Other" tier that recruiters on a busy role almost never open. On Workday, anything you wrote in your Professional Summary is stored only in the attached PDF and is invisible to recruiter search.
Source: RESEARCH.md, Taleo ACE tiers and Workday parsing [HIGH confidence].
Why the myth is worse than useless
The "a robot rejected you" story is comforting because it takes you off the hook, and it is profitable because it sells a fix for a problem that is not shaped the way it is sold. But if you believe the real blocker is keyword density, you will spend your energy stuffing keywords and never notice that you answered a knockout question wrong, or that your two-column template quietly scrambled your last three roles.
The honest version is less dramatic and more actionable: answer the screening questions carefully, and give the parser a clean, single-column document it can read without guessing. Do that and you are visible. Being visible is the whole job of the résumé at this stage. It is not a guarantee of an interview, and no tool that tells you otherwise is being straight with you.
Independent and not affiliated with or endorsed by the named vendors. This reflects publicly documented behavior and may not match any specific employer's configuration. Trademarks belong to their owners.
Questions people ask
Rarely on formatting or keywords alone. The two documented auto-reject mechanisms are knockout questions, which are a yes/no screen on minimum qualifications applied before anyone reads you, and a parse that comes out so incomplete that you are effectively unsearchable. Recruiters surveyed said Workday, one of the most widespread systems, does not auto-reject for a missing keyword by itself.
There is no primary source behind that figure in our research corpus, and the surveys that do exist point the other way. The number is repeated widely without a citation you can check, which is a reason to distrust it, not to act on it.
Because a bad parse does not reject you, it buries you. On exact-match legacy systems like Taleo, one missing required keyword drops you to a tier recruiters rarely open. On Workday, keywords placed in a non-searchable Summary field never surface in a recruiter's query. You are still in the system, just invisible.
Find out what would actually bury you.
We score your résumé against all six major screeners and show you, per system, what parses cleanly and what does not. No made-up rejection rate.
Run my audit