Methodology
How an ATS resume score is calculated
No universal ATS score exists. Nobody outside Workday, Taleo or Greenhouse knows their thresholds, so this one measures something narrower and checkable: how much of your resume a parser recovers. Below is what that number means, which habits cost you it, and what it never counts against you.
What it measures
What the number is
How much of your resume a parser recovers, and nothing else. Not a prediction of whether you get the interview, and not a percentage any vendor publishes. A high score means the software read your name, your contact details, your job titles and your dates back correctly. A low one means some of that went missing between your file and the database a recruiter searches, and the report names which parts.
01
The ceiling is 98
Not 100. No vendor publishes its parser rules, so a perfect score would be a claim about software nobody outside those companies can read.
02
Some files score zero
A resume nothing can be read out of is worth nothing to the software: no text layer, no text, or a file that yields no text for any other reason.
03
Blockers sit outside it
A file the portal will not accept never reaches a parser, so it is shown above the score rather than inside it. The score still computes: the advice survives a re-export.
04
The same file scores the same
No sampling, no model, no luck. Run the same file twice and you get the same findings. The number moves only when the checker itself changes, as it did on 30 September 2026.
What a number means
What counts as a good ATS score
A score from one tool cannot be compared with a score from another, and anyone publishing a target across all of them is guessing. Every checker weighs different things, and none of them publish the weights. So the only honest version of this answer is a narrow one: here is what a number means here, on the scale this page describes, out of 98.
| Score | Band | What to expect |
|---|---|---|
| 88–98 | Parses clean | Should come through intact. |
| 70–87 | Minor issues | Will parse, with some detail lost. |
| 40–69 | At risk | Expect fields to go missing. |
| 0–39 | Breaks | Most of this will not survive. |
Those boundaries are a judgment about what each band should mean, not a curve fitted to a corpus of real resumes, because no such corpus exists to fit one to. They are set so that parses clean means nothing structural is wrong, and a single table or one page header is enough to drop a document out of it.
The more useful reading is not the band at all. A 70 whose only deduction is a missing portfolio link and a 70 whose contact details are stuck in a page header are the same number and not the same problem. The report names the deductions in order, heaviest first, and that list is what to act on. The number is only there to tell you whether to bother reading it.
What matters
What a parser reads cleanly, and what it loses
Almost every low score comes from the same short list of layout habits, and none of them are about how good the writing is. A plain resume that a parser reads completely will outscore a beautiful one it reads in pieces, every time.
Reads cleanly
One column, top to bottom. Real selectable text rather than a picture of text. Ordinary section headings: Experience, Education, Skills. Your name, email and phone in the body of the first page. One date format used throughout, with an ongoing role marked Present. A .docx or a text-based PDF, comfortably under a couple of megabytes.
Gets lost
Two columns, because the text is read straight across and your job titles end up interleaved with your skills. Tables used to align things. Contact details in a header, a footer or a text box, which some parsers never open. Scanned or exported images of text. Letter-spaced words, where s p a c i n g breaks one word into several. Invented headings, so the section under them maps to nothing.
Costs you nothing
Length, as long as the content earns it. Color and typeface, within reason. A photo, which is expected in much of Europe and Asia. Gaps between roles. A career change. None of these are penalized here, and a checker that marks you down for them is scoring taste rather than machine readability.
Fix in this order
Anything that stops the file being read at all, then anything that loses your contact details, then the structure of your work history, then the rest. Your report is already sorted this way: the top finding is the one worth an hour, and the bottom one is worth a minute.
Sources
What these checks rest on
Where a vendor has documented something, it is here with a link to what they said. Where nobody publishes anything, the call is ours, and your report says so on the finding itself rather than dressing it up as research.
Greenhouse Support, Unsuccessful resume parse
Greenhouse lists graphics, tables, headers and footers, contact details in a text box, multi-column layouts and inconsistent section formatting as causes of a failed or partial parse.
Hidden Workers: Untapped Talent, Harvard Business School and Accenture (2021)
88% of employers say qualified candidates are screened out because they do not match the exact criteria in the job description; over 90% use a recruiting system to make the first cut.
Layout-Aware Parsing Meets Efficient LLMs, arXiv 2510.09722 (2025)
Dates and named entities are extracted at F1 0.95 to 0.97 when present, so a missing date is a gap in the document rather than a parsing risk.
Greenhouse Support, supported formats and parse limits
Greenhouse cannot parse a resume larger than 2.5MB, and a resume uploaded as an image rather than a document will not parse at all.
Layout-Aware Parsing Meets Efficient LLMs, arXiv 2510.09722 (2025)
Around 20% of real resumes use a non-linear multi-column layout; removing layout-aware reordering costs 10+ accuracy points, and long descriptions fall to F1 0.55 without it.
Not scored
What never enters the score
Keyword coverage
Reported only when you paste a job posting, and never folded in: a resume is not worse for missing keywords from a job you have not applied to. The terms a posting asks for most often are ranked against what your resume already says, and common filler words are ignored. Repeating a term far more than the posting does is flagged rather than rewarded. Abbreviations show as near misses: an ATS matches literal text, so counting K8s as Kubernetes would promise a match it will not make.
Writing quality
Bullet length, weak verbs and missing metrics are reported for the human reader and kept out of the parse figure. Mixing “a machine can read this” with “this is well written” is what makes a single blended score meaningless.
Employment gaps
Surfaced to you, never scored and never framed as a defect. Gaps are normal, and a tool that marks people down for them is doing harm rather than analysis.
Photos
Detected and reported at zero cost. A photo is expected on a resume across much of Europe and Asia, so charging points would mark a document down for following local convention. The risk is information that exists only inside the image, and the finding says so.