What is a good ATS score?

There's no universal threshold because there's no universal score; each tool grades its own checklist. On our checker, structure issues are binary enough that a healthy file typically passes all four checks outright; anything a check flags is worth fixing regardless of the total. Read the checks, not just the number, in your free report.

Do real applicant tracking systems assign scores?

Classic ATS platforms parse and store resumes and support recruiter search; they don't grade files. Some platforms have added ranking or matching features that order candidates against a posting, but those are per-job comparisons, not a quality score a third-party tool could read out. Any number you see on a checker was computed by that checker.

Why did my resume get different scores on different sites?

Different checklists, different extraction engines, different incentives. Nothing about your resume changed. Compare the individual findings across tools instead of the totals: if two tools independently flag your headings or contact details, that's signal worth acting on.

How is the Wynhire checker score calculated?

Four deterministic checks with published weights: text extraction 40 points, contact information 20, section headings 20, dates 20. A partial pass (warn) earns half weight. Content suggestions (quantification, summary, cliches, keyword coverage) are listed separately and never affect the score. The report also shows the full extracted text so you can verify the summary yourself.

Can I get a perfect score and still not get interviews?

Yes, and understanding why protects you from score-chasing. A structure score measures whether machines can read your resume, not whether it says anything compelling. Interviews come from relevant, quantified, tailored content on top of clean structure. See how to pass ATS screening for the full picture beyond the number.

Should I pay for a detailed ATS score report?

Never pay to see the evidence behind a verdict; that inversion is a sales pattern, not a service. Extraction checks are computationally trivial, which is why ours are free and unlimited. Paying for human resume feedback can be worthwhile; paying to unhide a gauge's reasoning is not.

What Is an ATS Score? What the Number Means, and What It Can't Mean

Updated August 16, 2026 · 14 min read

An ATS score is a number produced by a resume-checking tool, not by an applicant tracking system. Real ATS software parses resumes into database records; it doesn't grade them. A useful checker score summarizes named, verifiable checks; a misleading one hides its method.

Somewhere in your job search you've probably seen it: a gauge, a percentage, a verdict. 'Your resume scores 62/100 against the ATS.' The number feels authoritative, and the tools that produce it rarely explain where it comes from. So let's do the explaining, including for our own score.

The uncomfortable truth first: applicant tracking systems do not score resumes. Workday, Greenhouse, iCIMS, Taleo and their peers parse your resume into a structured record and let recruiters search it. There is no hidden grade inside the machine that a tool could read out. Every 'ATS score' you've ever seen was computed by the checking tool itself, against that tool's own checklist. The score can still be genuinely useful, but only if you know what it measures, and only if its maker tells you. This guide covers what scores can honestly measure, what they can't, how our checker's score is built (with the exact weights), and how to read any score, from any tool, without being played.

Where ATS scores actually come from

When a tool shows you a resume score, it ran some set of checks on your file and compressed the results into one number. The checks vary wildly between tools, which is why the same resume can score 90 on one site and 55 on another. Nothing contradictory happened; two different checklists produced two different summaries.

Broadly, checker scores are built from three families of signal:

Parseability signals. Can the text be extracted from the file? Are there recognizable section headings, contact details, dates? These are the closest thing to 'what an ATS experiences', because parsing really is what an ATS does.
Content-quality signals. Are bullets quantified? Is there a summary? Are cliches present? Useful writing feedback, but no tracking system evaluates any of it. A resume full of cliches parses flawlessly.
Keyword-match signals. How much does the resume's vocabulary overlap with a specific job posting? This approximates recruiter search behavior, and it only means anything when a real posting was provided for comparison.

The question that exposes a score's honesty

Ask: 'which checks produced this number, and can I see their individual results?' A trustworthy tool names its checks and shows each one's finding on your actual file. An untrustworthy one shows a gauge, a scary verdict, and a paywall between you and the details. The score isn't the problem; the opacity is.

How 'the ATS score' became a sales device

It's worth pausing on why this number is everywhere, because the incentive structure explains most of its distortions. Resume checkers are, overwhelmingly, acquisition funnels: for rewrite services, for subscription builders, for coaching upsells. A funnel needs urgency, and nothing manufactures urgency like an authoritative-looking number attached to a scary verdict. 'Your resume scored 58' converts better than 'we ran nine heuristics and four raised suggestions', even when the second sentence is what actually happened.

This is how the mythology of the grading robot sustains itself. Each tool that presents its checklist as 'the ATS score' reinforces the belief that a single machine judgment exists inside employer systems, waiting to be appeased. Job seekers then search for that score, more tools appear to supply it, and the loop tightens. The systems themselves never participated: parsing software has no opinion of you.

None of this makes checker scores useless. Structural checks catch real, costly problems, and we obviously think so, since we built one. It means the burden of proof sits with the tool: named checks, visible findings, published arithmetic, and the raw material behind it all. When a tool meets that bar, its score is a legitimate summary. When it hides the method and quotes a price to see more, you're not looking at a diagnostic; you're looking at a pitch.

What a score cannot tell you, no matter who computes it

Some limits are structural, and any tool that implies otherwise is overreaching:

It can't predict a specific employer's system. Parsers differ across vendors and versions, and none publish their internals. 'You will pass Workday' is not a claim anyone can verify.
It can't measure how good your career looks. A thin work history in a perfectly structured file scores high on structure, because structure is what's being measured. No number substitutes for content.
It can't know the recruiter's search terms unless you supply the posting. Generic keyword grades computed without a job description are checked against someone's generic list, not against your actual target.
It can't guarantee interviews. Screening is one gate. Referrals, timing, competition, and the substance of your experience dominate outcomes past that gate.

How our checker's score is computed, weight by weight

Since we just argued that scores are only as honest as their published method, here is ours in full. The free resume checker computes a structure score from 0 to 100 out of four deterministic checks on your file:

Check	What it verifies on your file	Weight
Text extraction	The file has a real text layer and its words read back whole (at least 90 percent well-formed tokens), not shattered by scans or decorative letter-spacing	40
Contact information	An email address and a phone number can be read back from the extracted text	20
Section headings	Standard headings (Experience, Education, Skills) are detectable, so a parser can assign content to fields	20
Dates	Enough year markers read back to build a work timeline (two or more)	20

The rules behind the arithmetic

Text extraction carries 40 of the 100 points because without extraction, nothing else exists: a scanned image scores zero there and the rest is academic. Each check can pass, warn, or fail; a warn (for example, an email found but no phone) earns half its weight. The same file always produces the same score, because every check is a deterministic rule, not a model's opinion.

Just as important is what the score excludes. The checker also reports content Opportunities: unquantified bullets, a missing or overlong summary, mixed date formats, cliche phrases, overall length outside the 350-to-900-word working range, and (if you paste a posting) uncovered job terms. None of these move the score by a single point. They're improvable writing, not parsing failures, and folding them into one number is exactly the confusion this article is about. The score measures what the machine can read; the Opportunities list suggests what you might say better.

And the part no score can replace

Above the score, the checker shows the extracted text itself: ATS View, the exact character stream a parser gets from your file. We built the report this way because a score, ours included, is a summary, and summaries are checkable only when the underlying material is visible. If you read the extraction and it's complete, ordered, and intact, you know something no gauge can tell you.

A worked example: one file through the arithmetic

To make the weights concrete, walk a plausible file through them. Say your resume is a text-based PDF exported from a word processor: extraction reads back clean, whole words, so Text extraction passes and banks its 40 points. Your header shows an email address in plain text, but your phone number lives inside a graphic banner: the parser reads one of the two, so Contact information warns and earns half weight, 10 of 20. Your headings are standard, so Section headings passes for 20. Every role carries a start and end date, so Dates passes for 20. Total: 90 out of 100.

Now the useful part: the report doesn't just say 90, it says which 10 points are missing and why, and the ATS View lets you confirm it: scan the extracted text for your phone number and it genuinely isn't there. Move the number into plain text, re-export, rescan, and the file scores 100. Nothing mystical happened at any step; that's what deterministic means.

Contrast that with the same file scoring 64 on a tool that blends structure with writing style. Maybe it disliked your bullet verbs, maybe it wanted a different summary length; without published arithmetic you cannot know, and without the raw extraction you cannot verify the parts that claim to be about parsing. Same resume, two numbers, only one of them auditable.

'Match rate' scores: the other number you'll meet

Alongside structure scores, many tools (and some enterprise ATS add-ons) show a match percentage between your resume and one job description. This number answers a narrower, more legitimate question: how much of the posting's vocabulary appears in your resume?

Used correctly, a match rate is a tailoring speedometer. Paste the posting, see which recurring terms you're missing, add the ones that honestly describe you, watch the coverage rise. Our checker's job-match feature works exactly this way, with plain term-frequency matching and no AI, because literal matching is what recruiter search actually does.

Used incorrectly, match rates become a stuffing incentive: chasing 100 percent by pasting requirements you don't meet. Two problems. First, the human who opens your resume reads the padding immediately. Second, interviews collapse resumes into conversations; vocabulary you can't back up becomes visible in minutes. Match honestly, then stop. A resume that covers the posting's core terms and reads like a person wrote it beats a keyword mosaic every time. The tailoring guide shows the honest version of the workflow, and the keywords guide covers finding terms worth matching.

What the recruiter sees on the other side

Since every score claims to predict what happens inside the employer's system, it helps to know what that side actually looks like. A recruiter working a requisition opens a list of applicants. For each one, the system shows a profile assembled from your parsed resume: name, contact fields, a work history timeline, education, extracted skills, plus the original file a click away. There is no score column produced from your resume's quality, and no red stamp from a robot. What there is: search.

Recruiters narrow large applicant pools with keyword searches and filters over those parsed fields: a certification, a tool, a job title, sometimes a boolean combination. The resumes that surface are the ones whose parsed records contain the searched terms; the rest remain in the database, unseen rather than rejected. Some modern platforms add per-requisition ranking or match features that order candidates against the posting's requirements. Those are comparisons against one specific job, computed inside one specific system, and no external checker can read or reproduce them.

This is the reality a good checker approximates and a bad one obscures. Extraction checks approximate 'will the profile assemble correctly'; keyword matching against a pasted posting approximates 'will the searches find me'. Both approximations are honest as long as they're labeled as such. A universal grade predicting your fate across all employers approximates nothing, because nothing like it exists on the recruiter's screen.

Why the same resume scores differently everywhere

Run one file through five checkers and you may get five numbers. Here's what actually differs under the hood:

Different checklists. One tool weights extraction heavily (ours: 40 points); another mostly grades writing style. Neither number is wrong about its own checklist.
Different extraction engines. Tools use different PDF libraries, which can genuinely extract different text from tricky files. This part mirrors reality: real ATS parsers differ the same way.
Different incentives. A checker attached to a rewrite service benefits from alarming you; low scores convert. A checker attached to a builder, like ours, benefits from being right when you re-test the rebuilt file. Ask what the tool wants you to do next, and weigh the verdict accordingly.
Different definitions of 'pass'. Some tools declare failure below an arbitrary threshold. There is no industry threshold; the tracking systems being simulated don't score at all.

Red flags and green flags in any score report

A quick field guide you can apply to any checker in about thirty seconds, ours included:

Green flag: named checks with per-check findings. 'Section headings: found Experience, Education; Skills not detected' is verifiable. 'Formatting: B minus' is not.
Green flag: the raw extracted text is shown. If you can read what the tool read, every claim it makes is auditable against it.
Green flag: published arithmetic. Weights and pass conditions in the open, so a score change after an edit is explainable.
Green flag: separation of parsing and writing. Structure problems and style suggestions labeled as different things, because they are.
Red flag: a verdict before an upload finishes processing. Instant dramatic scores suggest theater, not analysis.
Red flag: the details cost money. A tool that knows what's wrong with your resume but charges to say so is negotiating, not diagnosing.
Red flag: urgency mechanics. Countdown timers, 'your resume is rejected by 9 out of 10 systems' style claims with no source, scores that conveniently hover just below a 'fixable' threshold.
Red flag: guarantees about named employers or systems. Nobody outside a company can test that company's configuration; certainty here is a tell, and the more specific the promise, the less testable it is.

How to actually use a resume score (a short protocol)

Scores are useful the way a dashboard warning light is useful: as a prompt to look, never as the inspection itself.

Run the scan and ignore the headline number for a moment. Read the individual checks first: what specifically was found or not found in your file?
Read the extracted text if the tool shows it (walk away from tools that won't). Complete, ordered, intact text is the actual pass condition
Fix structural failures first: extraction problems, missing contact details, unrecognizable headings, missing dates. These are cheap to fix and expensive to ignore
Treat content suggestions as an editing list, applied with judgment. Quantify the bullets you can honestly quantify; skip advice that would make you sound like a template
Add the job posting and close the honest keyword gaps. Then stop optimizing the number
Re-verify after your final export, since 'the same resume' in a new file format is a new file. Scans cost nothing, so verify the version you'll actually send

Where templates fit in

If your score keeps failing on structure no matter how you edit, the layout itself is usually the culprit, and swapping it beats fighting it. Every template in our ATS-verified gallery passed an automated extraction gate (at least 95 percent token coverage with section order intact; measured between 98.2 and 98.9 percent across the twenty verified designs in the latest run). Rebuilding on one of those, or starting from a worked example like the registered nurse resume or the software engineer resume, gives you a base where the structural 100 is the starting point rather than the goal.

Key takeaways

No applicant tracking system scores your resume. Every 'ATS score' is computed by a checking tool against its own checklist, which is why the same file scores differently everywhere.
A score is trustworthy in proportion to its transparency: named checks, visible per-check results, published weights, and ideally the raw extracted text behind it all.
Our score weights text extraction 40, contact info 20, section headings 20, dates 20, with half credit for partial passes. Content suggestions never move it.
Match-rate scores are tailoring tools, not targets: cover the posting's honest core terms and stop before you become a keyword mosaic.
The real pass condition is readable in seconds: extract your resume's text and check it's complete, ordered, and intact. The score just points you there.

Frequently asked questions

What is a good ATS score?: There's no universal threshold because there's no universal score; each tool grades its own checklist. On our checker, structure issues are binary enough that a healthy file typically passes all four checks outright; anything a check flags is worth fixing regardless of the total. Read the checks, not just the number, in your free report.
Do real applicant tracking systems assign scores?: Classic ATS platforms parse and store resumes and support recruiter search; they don't grade files. Some platforms have added ranking or matching features that order candidates against a posting, but those are per-job comparisons, not a quality score a third-party tool could read out. Any number you see on a checker was computed by that checker.
Why did my resume get different scores on different sites?: Different checklists, different extraction engines, different incentives. Nothing about your resume changed. Compare the individual findings across tools instead of the totals: if two tools independently flag your headings or contact details, that's signal worth acting on.
How is the Wynhire checker score calculated?: Four deterministic checks with published weights: text extraction 40 points, contact information 20, section headings 20, dates 20. A partial pass (warn) earns half weight. Content suggestions (quantification, summary, cliches, keyword coverage) are listed separately and never affect the score. The report also shows the full extracted text so you can verify the summary yourself.
Can I get a perfect score and still not get interviews?: Yes, and understanding why protects you from score-chasing. A structure score measures whether machines can read your resume, not whether it says anything compelling. Interviews come from relevant, quantified, tailored content on top of clean structure. See how to pass ATS screening for the full picture beyond the number.
Should I pay for a detailed ATS score report?: Never pay to see the evidence behind a verdict; that inversion is a sales pattern, not a service. Extraction checks are computationally trivial, which is why ours are free and unlimited. Paying for human resume feedback can be worthwhile; paying to unhide a gauge's reasoning is not.