The Folio ATS Score
An ATS score is a 0 to 100 rating of whether applicant tracking software can read your resume, not of whether a recruiter will like it. Folio computes it from seven weighted criteria: column structure worth 30 points, standard headings 18, real selectable text 16, contact details 12, sensible length 10, readable contrast 8, and no risky elements 6. The scorer is deterministic, so the same resume always returns the same number, and 90 or above earns the ATS-friendly badge. The full rubric is published on this page and served as JSON at /api/ats-standard.json, so you can reproduce the arithmetic yourself.
Read the rubric. Then build a resume that satisfies it.
The standard is free to read, quote and fork under CC BY 4.0, and no account is needed for any of that. If you do sign up, resume PDF and DOCX export carries no watermark and no paywall on the Free plan. Free also means zero custom domains, a Made with Folio badge, and 10 AI drafts a month, and you should know that before you click, not after.
ATS score meaning
An applicant tracking system is the software an employer uses to receive applications. It opens your file, pulls the words out, and drops them into fields: name here, employer there, dates over there. An ATS score estimates how much of that survives the trip. A high score means a parser will get your job titles, your dates and your email address out intact. A low score means it will get some of them, and you will never find out which ones went missing.
Which brings up the awkward part. Search for a resume checker and every product hands you a number out of 100. Ask any of them how the number was reached and the trail goes cold. The weights are not published. The bands are not published. Whether the same resume scores the same twice is not stated. A score you cannot check is not a measurement, it is a mood, and the reason the arithmetic stays hidden is that arithmetic in the open can be argued with.
So here is ours. Seven criteria, one hundred points, the exact weights below, the bands below that, and a list of the things the score is simply not able to see. The scorer is deterministic, which means there is nothing to hide: no sampling, no randomness, no quiet adjustment that nudges a paying customer toward a flattering result. Publishing it costs us nothing except the ability to bluff.
The rubric
The weights on this page are imported from the scorer, not retyped beside it. If an engineer changes one in the code, this page changes with it.
30 points
The heaviest criterion, because it is the one that actually loses your data. A single column reads top to bottom the way a parser expects. A purpose-built ATS layout scores a factor of 1.0 and any other single-column layout scores 0.96. A two-pane layout that keeps the story in one main column scores 0.8. A true three-pane split scores 0.55, because reading across the page interleaves separate columns and fields land in the wrong bucket.
18 points
Scored on how many of the expected sections carry the names a parser hunts for: Experience, Education, Skills. Renaming Experience to something wittier is not a personality, it is an unindexed section. The parser is not charmed, it is confused.
16 points
Whether the words in the file are text or pixels. Folio always awards this in full because it never flattens type into an image, and we list it anyway, because a gorgeous resume exported as a picture is the most common way a design scores nothing at all. If you cannot drag your cursor across the words in your PDF, no keyword tuning will rescue it.
12 points
Awarded in proportion to how many of name, email, phone and location the parser can find. This is the quietest failure in the whole rubric: the software reads your entire career and comes away with no way to reach you, usually because the contact block was parked inside a header or a floating box.
10 points
Measured against the density the chosen layout can carry, not against a folk rule about pages. The question is whether the content overflows its frame or leaves it half empty, both of which distort what comes out the other end.
8 points
Awarded when the body text is dark enough to be safe on white. Pale grey type looks considered on a retina display and then vanishes in a photocopy, a low-quality render, or a screening tool that thresholds the image.
6 points
Awarded when the layout keeps clear of the things parsers mishandle: tables, text boxes, columns nested inside columns. Folio earns this by construction rather than by detection. The templates are built so those elements cannot get in, which is a duller way to solve it and a more reliable one.
What is a good ATS score
People search for whether 75 is good, or 80. Here is the honest answer, with the thresholds we use rather than a number we made up to sound reassuring.
90+
Excellent
earns the ATS-friendly badge; the structure will not lose fields
75+
Good
parses reliably; usually one structural choice from excellent
55+
Fair
likely to parse, but something is at risk, usually columns
Under 55
Poor
expect dropped or mangled fields, however good the writing is
Published rubric vs asserted number
We are not claiming other scorers are wrong. We cannot claim that, because we cannot see them, and neither can you. That is the entire point of the table.
| Capability | Folio | A typical resume scorer | The real ATS vendors |
|---|---|---|---|
| The scoring rubric is published | In full, all seven weights | Not published | No vendor publishes its parser |
| The same resume always scores the same | Deterministic, guaranteed | Not stated either way | Not observable from outside |
| You can redo the arithmetic by hand | Yes, from this page | You get a number | You get a rejection, or silence |
| A machine-readable version exists | JSON, CORS open, generated from the code | None found | None published |
| The tool says what its score cannot see | Four limits, stated below | Rarely, and never prominently | Not its job; it is the employer tool |
| Claims to score any file you upload | No, and we explain why below | Commonly claimed, method unstated | Yes, parsing files is the whole product |
These cells describe what each category publishes, checked against public documentation on the date below. If a vendor publishes a rubric and we have missed it, write to us and we will correct this table and say we were corrected.
The four limits
A standard that only lists its strengths is an advertisement. These four limits are printed at the top of the machine-readable version too, so nobody can quote the score without also quoting what it fails to measure.
One. It does not score a file you upload. It reads the layout, the theme and the content model of a resume built inside Folio, so it has no way to grade a PDF that came out of Canva, Word or another builder. Honest grading of an arbitrary file means rendering it and re-parsing it, and any tool that reports a confident number on your upload without doing that is estimating and calling it measurement. What Folio does instead is narrower and more useful: it builds the document in templates where the structural rules cannot be broken, and shows you the score before you export.
Two. It is not a keyword match rate. Whether your resume speaks the language of one particular job is a different question with a different answer, and Folio measures that separately when you paste a job description in. Do not read a 94 here as evidence that you are a fit for the role. It is evidence that the machine can read you.
Three. It is not a prediction of an interview. It grades legibility to software. Whether a human wants to meet you turns on things no rubric contains, and a tool that hints otherwise is selling hope by the point.
Four. It is not a measure of any single vendor system. Workday, Greenhouse, Lever, iCIMS and Taleo each parse differently and none of them hand out their parser. The seven criteria are the fundamentals every parser leans on, which is why fixing them helps everywhere and guarantees nothing anywhere.
Use it, cite it, break it
It is released under CC BY 4.0. Quote it, reprint it, put it in your careers-service handout, ship it in your own product. Attribution is appreciated and not required.
01
GET /api/ats-standard.json returns the whole standard: criteria, weights, bands, limits and version. It is CORS open on purpose, because a standard nobody can fetch from their own page is a brochure with ambitions.
02
That endpoint is generated from lib/resume/ats.ts, the scorer that actually runs. The weights are imported, never retyped. A published standard that has quietly diverged from the code it describes is worse than none, because now it is a false claim with a citation attached.
03
The document carries a semantic version. If a weight, band or factor ever changes, that version changes with it, so anything you build on top can pin and detect the change instead of silently inheriting it.
04
Folio ATS Score, version 1.0.0, published by Folio at portfolio.wrxstack.com/ats-score-standard, licensed CC BY 4.0. If you think a weight is wrong, tell us which one and why. We would rather be corrected in public than be vague in private.
FAQ
It is a rating out of 100 for how well applicant tracking software can extract the contents of your resume. It grades the document, not the career inside it. A parser has to turn your file into fields, and the score asks how much of that survives: are the columns readable in order, are the section names ones the software recognises, is the text real text, can it find your email address. In the Folio rubric those checks carry fixed weights that add up to 100, and the number you see is their sum.
In the Folio bands, 90 and above is excellent and earns the ATS-friendly badge, 75 and above is good, 55 and above is fair, and below 55 is poor. So a 75 or an 80 is genuinely fine: the document will parse. It also tells you something specific, which is that you are almost certainly one structural decision away from the top band, and it is almost always the columns. Move to a single-column layout and watch the 30-point criterion swing. Do not chase 100 for its own sake; the difference between 92 and 100 is invisible to a recruiter.
In our case, like this. Each of the seven criteria is scored on its own from 0 to 1, then multiplied by its weight, and the products are added together. Column structure contributes up to 30, headings up to 18, selectable text up to 16, contact details up to 12, length up to 10, contrast up to 8 and risky elements up to 6. Column structure uses a plain factor for the layout family: 1.0 for a dedicated ATS layout, 0.96 for another single column, 0.8 for two panes, 0.55 for three. There is no hidden term, no randomness and no rounding in our favour, which is why we can print it.
It matters in one narrow, unglamorous way, and you should be suspicious of anyone who inflates it beyond that. Nobody at the company sees your score, and it will not win you a job. What it does is remove a stupid reason to lose one. If your file parses badly, a recruiter may never see your best line, because it landed in the wrong field or fell out of the document altogether. Clearing the parse is table stakes. It buys you the chance to be judged on the work, which is the only judgement worth having.
Build the resume in Folio and the score is on screen while you edit it, at no cost, with no scan counter to ration and no charge to download the finished PDF or DOCX. Being straight about the shape of that: what you get graded is the resume Folio renders, because the scorer reads the layout and the content model directly. So this is not a place to drop a mystery file and receive a verdict. It is a place to rebuild the resume where the rules already hold, and see the number before you send anything.
Not as a file, and we would rather say so than take the click. Folio can read the text out of an existing PDF to get your content into the editor, and once it is there it is scored like any other Folio resume. But the score describes the document Folio builds, because structure, headings, contrast and text selectability are properties of a layout, and Folio only knows the layouts it renders. Any product that promises an exact score for an arbitrary upload without rendering and re-parsing it is guessing with a progress bar.
No. There is no shared number, no score that travels with your application, and nothing in an employer system that says you got a 91 from a resume tool. Scores like this one exist purely on your side of the fence, as a proxy for a question you cannot otherwise answer: will the software on the other side be able to read this. Treat it as a preflight check, not as a grade someone else is reading.
Keep going