Is Your ATS Score Accurate? What a 0–100 Match Score Actually Measures

A number with nothing behind it

Paste a resume and a job description into almost any resume checker and you'll get a score. Sixty-four. Seventy-eight. Ninety-one. It shows up in a big circle or a progress bar, colored red, yellow, or green depending on how you did, and it reads like a grade: official, measured, the kind of number a real system somewhere calculated about you.

It isn't that. No applicant tracking platform, the actual hiring software companies use to collect and search applications, generates a percentage like this and hands it to a candidate. Workday doesn't do it. Greenhouse doesn't do it. Taleo, Lever, iCIMS: none of them. The 0–100 score you're looking at was built entirely by the checker tool you're using, from scratch, using a formula that tool's developers chose and that no other tool is obligated to agree with.

That's worth sitting with, because the framing matters. "Your resume scored 71% against this job" sounds like a report on something that already happened inside the hiring process. What it actually is: a separate piece of software made up a number, after the fact, based on comparing your text to a job posting's text, with no connection to how any real applicant tracking system will treat your file.

What the score is actually counting

Strip the presentation away and most of these scores reduce to a version of the same idea: count how many words and phrases in the job description also show up in your resume, weight some of them more than others (a required skill mentioned three times probably counts for more than one mentioned once), and convert that overlap into a number. Some tools add a layer of synonym-matching, so "managed a team" might get partial credit against a posting that says "led a team." Some don't, and reward exact phrasing only.

None of that is inherently useless as a concept. Checking whether your resume's language overlaps with a posting's language is a real, reasonable thing to want to know before you apply. The problem is the presentation, not the underlying idea: a loose approximation of keyword overlap, dressed up as a precise percentage, implies a confidence the method doesn't have. A phrase-matching algorithm comparing two blocks of text is not measuring your chances of an interview, your fit for the role, or anything a hiring manager is actually weighing when they read your resume. It's measuring word overlap, nothing more, and then presenting that one narrow measurement as if it were the whole picture.

Why the same resume gets a different score everywhere

If you've ever run one resume through two different checkers against the same job posting and gotten two different scores, that's not a bug in one of them. It's the expected result of there being no shared standard. Each tool picks its own list of what counts as a meaningful keyword, its own weighting, its own threshold for what counts as a "match" versus a near-miss. One tool might flag "cross-functional" as missing and tank your score over it; another might not consider that phrase significant at all.

That inconsistency is itself informative. If a score were measuring something real and stable about your resume, running it through a second tool should land somewhere close to the first result. When it doesn't, consistently, across tools and across users, that's a sign the number is an artifact of each tool's particular formula rather than a property of the resume being measured.

The part that's genuinely worth your attention

None of this means keyword overlap is a waste of time to think about. It means the useful part isn't the score, it's the specific list of terms underneath it. If two or three different checkers all flag the same missing keyword, like a specific tool, certification, or skill named directly in the posting, that's a reasonable thing to go add, assuming it's actually true of your background. A score of "73%" tells you almost nothing actionable. "This posting says 'SQL' five times and your resume doesn't say it once" tells you exactly what to go check and, if it's accurate, fix.

The honest way to use a keyword checker, if you use one at all, is to ignore the number at the top and read the list it's built from, cross-checking against a second tool when you can, and treating anything only one tool flags with some skepticism.

What a score can't tell you at all

Here's the bigger gap: even a perfectly calculated keyword-overlap percentage says nothing about whether your resume file is being read correctly in the first place. A match score assumes your resume's text has already made it successfully into the system as clean, structured data. That assumption can be wrong in ways a keyword score has no way of detecting, because it's working from whatever text you pasted in, not from your actual file.

A resume built with a two-column layout can paste cleanly into a checker's text box (you're copying text, not re-running the file through a parser) and still scramble badly when the real file gets opened by an actual tracking system reading positions on a page. A scanned or flattened PDF has no text layer at all, which a keyword checker never discovers because you typed or pasted text in manually instead of letting the tool extract it from the file itself. A phone number sitting in a header might extract fine in one system and vanish in another, something no amount of keyword tuning touches, because the number was never missing from your resume, it was missing from what the parser pulled out of your file.

This is a different question entirely from "does my wording match this posting," and it's the one a 0–100 score was never built to answer. ResumeReadout runs your actual resume file, not text you've pasted somewhere, through a parser and shows you plainly what got extracted: whether there's a real text layer, whether your layout holds up outside of columns and tables, whether your name and contact details landed where a system would expect to find them. It doesn't hand you a score, because that's not what this kind of check should produce; it shows you the extraction itself, so you're looking at a fact about your file instead of a formula's best guess.

Before trusting any score a checker gives you, or spending more time tuning wording to chase a higher one, it's worth ruling out the more basic, mechanical stuff first:

If any of those are true, a high match score is sitting on top of a file that might not be read correctly at all, which makes the score close to meaningless regardless of how it was calculated.

The bottom line

A resume checker's score isn't fraudulent, exactly, but it isn't what it looks like either. It's one tool's private formula for keyword overlap, rendered as a number precise enough to feel authoritative and vague enough to mean almost anything. Different tools will give you different scores for the same resume and the same job, because there's no standard they're all measuring against, and no real applicant tracking system is generating or checking that number on the other end. The keyword gaps underneath a score are worth a glance. The score itself is worth almost nothing, and it's never going to tell you whether your actual file is extracting the way you think it is.

FAQ

Keep reading

• Does ATS Software Really Reject 75% of Resumes?
• ATS-Friendly Resume Format: What Actually Matters (and What's a Myth)
• What Happens After You Submit a Resume Online: A Plain-Language ATS Walkthrough

Or skip the reading and run the free parse-check →