ATS scores · original research
What is a good ATS score? Percentiles from 1,085 real checks
Across 1,085 real resume checks on ResumeAtlas (March 2026 – June 2026), the median ATS score was 80. 37.8% of resumes scored 85 or higher and 12.5% scored below 70 — so an 85 is a good score (above 62.2% of resumes we scored), an 80 is average, and anything under 70 is a formatting problem before it is a keyword problem. The number is a proxy for how cleanly a parser reads your file, not a pass mark an employer uses. You can check where your resume falls free with no signup, or score it against one job description — also free, no signup.
Last updated: September 12, 2026
Where your number sits
Every “good ATS score” article on the web quotes a threshold — 70, 75, 80 or 85 — and none of them says how many resumes actually clear it. This is the distribution behind the blended ATS score ResumeAtlas assigned in 1,085 resume-vs-job-description analyses between March 2026 – June 2026. The right-hand column is the one that answers “is mine good”: the share of resumes at or above each band.
| Score | Share of resumes | At or above this band |
|---|---|---|
| 90–100 | 27.6% (300) | 27.6% |
| 80–89 | 27.7% (301) | 55.4% |
| 70–79 | 32.1% (348) | 87.5% |
| 60–69 | 4.5% (49) | 92% |
| 50–59 | 0% (0) | 92% |
| 40–49 | 3.3% (36) | 95.3% |
| 30–39 | 2.2% (24) | 97.5% |
| 20–29 | 1% (11) | 98.5% |
| 10–19 | 0.9% (10) | 99.4% |
| 0–9 | 0.6% (6) | 100% |
n = 1,085 analyses, 2026-03-26 to 2026-06-19. Median 80; middle half of resumes scored 72–90. Bands with 0% are real: the model's scores cluster around round numbers, so some ten-point ranges are empty.
Is 75 a good ATS score?
Below average. A 75 is above 31.2% of the resumes we scored and below the median of 80. The file reads, but layout or section problems are costing points.
Is 80 a good ATS score?
Average — it is the median. An 80 is above 44.6% of resumes and 55.4% scored 80 or more. Mid-pack, not a pass; the fixes left are usually small.
Is 85 a good ATS score?
Yes. Only 37.8% of resumes reached 85; a 90 puts you in the top 27.6%. Past this point, stop polishing the score and work on fit.
“ATS score” means three different things
The reason one tool gives your resume a 60 and another an 85 is usually not that one of them is broken. They are measuring different things and calling both an “ATS score”. Three numbers travel under that name, and you should know which one you are looking at before you react to it.
| Score | What it measures | In our data | A low number means |
|---|---|---|---|
| Parse score | Whether a parser extracts your name, contact details, section headings, jobs with dates, skills and education from the file. A property of the file alone. | Median 83 across 181 submitted resumes; 17.7% below 70, 47% at 85+. | Formatting is losing content before anyone reads it. Fix the layout, not the words. |
| Match score | How many of one posting's required keywords your resume demonstrates. A property of the pair (resume, job) — the same resume gets a different number for every posting. | Median 79% across 1,779 resume-vs-posting analyses; 36.5% below 70, 40.4% at 85+. | Either the fit is weak or the evidence is in the wrong place (skills list, not bullets). |
| Blended / “resume score” | A model's overall judgement of ATS-friendliness — structure, section coverage, formatting red flags, bullet quality — in one 0–100 number. Some builders call a pure writing grade this too. | Median 80 across 1,085 analyses (the table above). | Several small problems at once. Useful as a first look; act on the two numbers above. |
Parse score: ResumeAtlas's own analyzer over 181 resumes people submitted (June 2026 – September 2026). Match and blended scores: resume-vs-job-description analyses in the same dataset as Resume Insights. Each series has its own sample; they are never pooled.
How an ATS score is calculated
Every tool keeps its exact weights to itself, so here are ours — the two formulas behind the numbers on this page. If a tool will not tell you what its number is made of, that is a reason to weigh it less.
Parse score = weighted checks on the file
The ATS resume checker extracts the text the way an applicant tracking system's first stage does, then scores parsing reliability (30%), field extraction (25%), formatting safety (20%), section detection (15%) and ats readability (10%). Each component is 0–100 from pass / warn / fail checks — is the name inside a header band, are the dates paired with the right job, is there a Skills heading a search can find — and the weighted sum is the score.
Match score = required keywords demonstrated ÷ required keywords
The resume-vs-job-description check reads the posting, lists what it actually requires, and marks each requirement as demonstrated only when an experience bullet proves it — a word sitting in a skills list does not count. Ten requirements with four missing is a 60. Which words count as required is a judgement call; the arithmetic is not.
What an employer's system does instead
It does not show a recruiter a score. It parses your file into fields, stores them, and lets the recruiter search, filter and sort. There is no universal pass line — every platform weighs and searches that data differently. So a checker's number is a proxy for two questions, “will my fields extract?” and “will their search hit my resume?” — never the employer's verdict on you.
Why the same resume gets 60 on one tool and 85 on another
Four reasons, in order of how often they explain the gap:
- Different measurement. One tool is a parse score, the other a match score (see above). A resume can parse at 95 and match at 40 — right file, wrong job — or match at 90 and parse at 55 — right skills, two-column PDF.
- Different keyword extraction. Match scores depend on which words the tool decides the posting requires. One reads “5+ years of Python” as one requirement; another as three. Same resume, different denominator.
- Different weighting. A tool that puts 30% on parsing reliability and one that puts 30% on writing quality will rank the same two resumes in opposite order.
- A writing grade in disguise. Several resume builders label their prose score — action verbs, bullet length, buzzwords — an “ATS score”. It is not measuring anything a parser or a recruiter search does.
The checks that actually break parsing are known and measurable. In our parsing benchmark of 181 submitted resumes, the most-flagged problems were section detection (43.6% of resumes), tables & columns (42% of resumes), skills (31.5% of resumes) and education (27.6% of resumes). 61.3% had no readable LinkedIn URL and 8.8% lost the candidate's name to a header band or image. None of that shows up in a match score. If your high score is not turning into interviews, the reasons are usually in the fit, not the file — see why a high ATS score still gets no interviews.
A good score is not a verdict
Since June 2026 every ResumeAtlas analysis also records an application verdict for the specific posting — Apply, Optimize First, or Skip — and an estimate of shortlist odds. Across 707 analyses with a verdict, even with median scores around 80:
13.6%
rated Apply as-is · median shortlist odds 72%
70.2%
rated Optimize First · median odds 42%
16.3%
rated Skip · median odds 12%
70.2% of resumes — most of which would pass any published “good score” threshold — still came back Optimize First — a requirement the posting insists on was not proven in a bullet. The score tells you the file reads; the verdict tells you whether to send it. That is why this page does not end at “85 is good”. See why resumes with good scores still get no interviews.
Median ATS score 80. Across 1,085 real resume checks (March 2026 – June 2026), 37.8% of resumes scored 85 or higher and 12.5% scored below 70 — an 85 is a top-38% score, not a pass mark. See all 1,793 analyses →
What to do at each score
The bands are ResumeAtlas's own (below 70 / 70–84 / 85 and up) and are stated as such — no employer publishes theirs. They are useful because each band has a different next step.
Below 70 — a formatting problem before it is a keyword problem
Fields are failing to extract. Do not touch the wording yet: rebuild the layout as a single column with standard headings, then re-check. The ATS format converter does that in one pass (free, no signup, plain-text download included), and the checker's fix list tells you which field failed and why.
70–84 — the file reads; the fit leaks
Contact and sections extract; points are going on a skills grid, a creative heading or a keyword that lives only in the skills list. Move the evidence into experience bullets — one bullet per requirement the posting names — and score the resume against the actual posting to see which requirements are still unproven. If the bullets have no numbers in them, start with quantifying your accomplishments.
85 and up — stop polishing the score
Another five points on a parse score changes nothing a recruiter sees. Spend the time on the specific job: run the verdict for that posting, fix the one or two requirements it flags, and send it. Chasing a 95 is how people end up with a keyword-stuffed resume that reads badly to the human who eventually opens it.
Can ChatGPT check my ATS score?
It can give you a number, and the number is not meaningless — but it is a writing grade with a guess at keywords, not a parse. A chat model reads the text you paste; it never sees the file. So it cannot tell you that your two-column PDF interleaves line by line, that your name sits in a header image, or that your dates are inside a table cell — the things that actually zero out a resume at the parsing stage. Use it to rewrite bullets and to sanity-check whether your experience addresses a posting. Use a parser to check the file. Both of ours are free with no signup: the checker for the file, the job-description check for the fit.
Do ATS scores matter? Are checkers legit?
A checker can accurately tell you one thing: whether a parser extracts your name, dates, job titles, skills and education cleanly from your file. That matters, because a resume whose fields come back empty is sorted too far down a recruiter's search to be opened — not rejected by a switch, just never seen. No checker, including ours, can tell you an employer's scoring or ranking logic, and any tool that claims a single universal pass/fail score is selling certainty it does not have.
When to ignore a score: when it is a prose grade from a builder that rewards buzzwords, when it moves ten points after you change a font, or when it is the only feedback the tool gives. When to take it seriously: when the tool shows you which field failed and what it read, so you can see the failure instead of trusting a number about it.
Method and limits
These are resume-vs-job-description analyses run on ResumeAtlas, not a survey of all resumes. The sample skews toward software, data and engineering roles, so the specific tools below reflect those postings more than hiring overall.
- Blended ATS score — the 0–100 judgement the ResumeAtlas analyzer assigned to each resume in resume-vs-job-description runs, n = 1,085, 2026-03-26 to 2026-06-19. Later analyzer versions report the match score in that field instead, so this series is fixed at that window rather than mixed with a different measurement.
- Match score — required-keyword coverage against the pasted posting, n = 1,779, 2026-03-26 to 2026-09-15, updated as the dataset grows.
- Parse score and parse checks — ResumeAtlas's own analyzer (the exact rules the checker runs) over n = 181 distinct resumes real people submitted, 2026-06-23 to 2026-09-12; 18.2% did not produce a usable record.
- Verdict split — n = 707 analyses that recorded a verdict, 2026-06-19 to 2026-09-15.
What this data does not prove
- PDF vs DOCX parse rates — the source file type is not stored, so the analyzer ran on extracted text only.
- Page-count effects — not stored.
- Bullet structure — text extraction normalises or strips bullet glyphs, so a low bullet count here is an extraction artefact, not the resume.
- Any specific commercial ATS (Workday, Greenhouse, Lever, iCIMS, Taleo, SAP). This benchmark tested the ResumeAtlas analyzer, which models the shared text-extraction stage — not any one vendor's parser.
- Hiring outcomes — nothing here records whether a resume led to an interview.
- Whether any score predicts an interview — nothing here records a hiring outcome.
Data generated 2026-09-15 (analyses) and 2026-09-12 (parsing benchmark). Aggregate only — no resume text or personal data. Free to cite with attribution; the full dataset and a CSV are on Resume Insights.
FAQ
What ATS score do you need to pass?
There is no pass mark, because an employer's system does not grade you — it parses your file and lets a recruiter search and sort. A checker's number tells you how cleanly that parse goes (a parse score) or how many of the posting's requirements your resume demonstrates (a match score). Treat below 70 as "fix the file or the fit first" and 85+ as "stop polishing the number"; nothing in between is a rejection.
What ATS score gets an interview?
None on its own. In the 707 analyses that recorded an application verdict, only 13.6% were rated ready to send as-is even though the median score sat around 80. The interview comes from the specific posting's must-haves being proven in your bullets — that is what the verdict measures, not the score.
Is a 100 ATS score possible?
Rarely, and it is not the goal. Only 1% of blended scores reached 95 or higher. A 100 on a match score just means every required keyword the tool extracted appears in your experience — which you can reach for one posting and lose on the next.
Why did my score drop after I edited my resume?
Usually one of three things: you moved a keyword from an experience bullet into a skills list (match scores count evidence in bullets), you introduced a layout element the parser cannot read (a table, a text box, a two-column section), or you are comparing against a different posting. Re-run the same posting through the same tool before concluding anything.
Does the file type change the score?
It can change the parse score, not the match score. A single-column PDF or DOCX with real text extracts the same; a PDF exported from a design tool with text in shapes, or a scanned image, extracts nothing. Our benchmark does not store file type, so we cannot put a number on the PDF-vs-DOCX gap — that is listed under what the data does not prove.
How often should I check my ATS score?
The parse score once per file version — it only changes when the layout does. The match score once per posting you actually intend to apply to, because it is a property of the pair (your resume, that job), not of the resume alone.
See where your resume falls
Upload or paste your resume and get the parse score, the fields that failed and the fixes — free, no signup. Then score it against the posting you are actually applying to.