Most resume anxiety about ATS software is aimed at the wrong thing. Here's what actually happens between uploading a file and a person opening it.
The journey
1. You upload. Usually a PDF, sometimes into a form that also asks you questions directly.
2. The text is extracted. The system pulls the words out of your file. If the words aren't real text — if they're inside an image — there's nothing to pull.
3. It's structured. The system tries to identify your name, contact details, employers, titles, dates, education and skills, and file them into fields. This is the step that goes wrong. If your job title ends up in the employer field, or a date range is misread, the record built from your resume is subtly wrong from then on.
4. It's indexed and ranked. Recruiters search and filter that pool — by title, by skill, by location, by years of experience. Your record surfaces or it doesn't.
5. A person opens it. They read the actual document, not the parsed record.
Where the real risk is
Not a robot rejecting you. Two duller things:
A bad parse. Your resume is filed wrongly, so you don't surface in searches you should have.
A weak rank. You parse perfectly and sit below better-matched candidates.
Both are addressed by the same two moves: clean structure, and genuine alignment with the posting.
Knockout questions are the actual gate
Where applications are filtered before human review, it's usually not from your resume. It's from questions you answered in the form — work authorization, a required license, a minimum years-of-experience threshold, a location requirement.
Those are typed answers, not parsed text. Which means filling in the application form carefully matters as much as the resume attached to it, and no amount of resume optimization compensates for a knockout answer.
Research on 25 recruiters across ten-plus ATS platforms found the large majority don't configure content-based auto-rejection at all — the platforms rank and sort. That study is small and we couldn't find a full published method, so treat it as directional. But it matches how these systems are documented to work, and it doesn't match the folklore.
The "75% get auto-rejected" claim
It's false and it's traceable to a company that shut down over a decade ago. There's no study underneath it. It has been recycled through job-search content ever since because it's frightening and it sells things.
We're not going to repeat it. See resume advice we won't give you.
What genuinely helps a parse
- Real text. Every PDF we produce is extractable text, not a picture.
- Standard section headings the system has seen before.
- Contact details in the body, not in a page header or footer.
- Dates in a consistent, ordinary format.
- A single column when you're uploading into a portal.
- Plain bullet glyphs, not icons or symbols.
You'll notice these aren't choices you have to get right — they're built into the templates, and they're what our ATS-strict designs guarantee. See ATS-strict vs design-forward.
What the ATS Ready dimension is actually measuring
It's about your content, not a rating of your template's aesthetics. Are the pieces a parser needs present and unambiguous — contact details, dated roles with titles, recognizable sections, real bullets?
A gorgeous design with no dates on your roles has a content problem. A plain one with everything filled in doesn't.
And then a human reads it
Worth ending here, because it's the part that gets forgotten. Once you surface, the whole thing is decided by a person reading a document for tens of seconds.
Which is why we don't optimize purely for machines. The formatting that survives parsing is largely the same formatting that survives a fast human read — clear sections, real white space, one column of thought at a time. See the research behind how we score.