ATS resume checker and the ATS score that does not exist

Also in: DeutschNederlandsPolski

Thousands of people a month look for a tool that reports their ATS score. The number those tools return is their own invention, and the useful thing they do is something else.

Try it on your own CV: ATS CV checker. Runs in your browser. Your CV is not uploaded or stored.

Thousands of people a month search for a tool that will tell them their ATS score. The uncomfortable answer is that the number those tools return is their own invention, and the useful thing they do is something else entirely.

There is no universal ATS score

No applicant tracking system publishes a portable score, and no external checker can read the ranking a specific employer sees. What exists inside a system is a match ranking against one posting, computed with that vendor's own logic and visible to the recruiter, not to you.

An analysis of more than 1.7 million applications found that these platforms do not auto-reject on a score at all: they parse, index and rank for a human who then filters. The score you are being sold is a proxy invented by the tool selling it.

What a checker is genuinely good for

A checker earns its place by showing you the text a parser actually extracted, because that is the one thing you can fix. Judge a tool by whether it shows you the extraction, not by whether it gives you a number.

The free check that needs no tool

Open your PDF, select all, copy, and paste it into a plain text editor. That is the same operation a parser performs, and it finds most real failures in under five minutes.

• Is the very first line your name, rather than your profile paragraph?
• Do Experience, Education and Skills appear as whole words?
• Did your email survive as one unbroken token?
• Has anything from a sidebar landed inside a job entry?
• Are all your dates in the same format, and did every one survive?

We built that check into a scored form and ran it on our own document: two columns passed 4 of 9, one column passed 9 of 9. The method and the per-system differences are in what actually parses in 2026, and the template side is in what makes a template parse.

Run the extraction yourself, with the same library parsers use

The plain-text copy-paste check finds most failures, but if you want the same view a parser gets, extract the PDF with a text-extraction library and read the output. That is exactly how we produced our own numbers, and it is free.

• Extract in strict position order, which is the harshest reading and the one that exposes columns.
• Then extract in column-aware mode and compare: a big difference means your layout is ambiguous.
• Check the first non-empty line is your name in BOTH runs.
• Search the output for your email as one unbroken string.

What a good checker report contains

Judge a paid checker on whether it hands you the extraction and the detected fields, not on the size of its number. If the report is a score and a list of missing keywords, you have bought an opinion.

When a low score is telling you nothing

Two checkers will disagree on the same file, because each scores against its own model of the posting. If one says 62% and another says 88%, neither number is wrong; they are measuring different things, and neither is what the employer sees.

Do employers really scan resumes for keywords?

Some may: MIT's career office says employers may use keyword scanning, so you should know which words are relevant to the industry and the position and make sure they appear. Note the wording: may, not always, and nobody outside the employer can see which it is.

Harvard's career center asks for resume language that is written for people who, or systems that, scan quickly. CareerOneStop, the US Department of Labor's site, says a person will typically scan a resume for 15 to 20 seconds. Both a reader and a program reward the same thing: the right words near the top, in plain text.

How do you use a checker's keyword gaps without stuffing?

Treat the gap list as a prompt, then check it against the postings themselves. CareerOneStop's own method for finding the right keywords does not use a score at all.

• Read 10 to 15 postings for your target job title and list the requirements that recur.
• Compare that list with the checker's gaps and ignore anything you cannot truthfully claim.
• Work the matching terms into your experience, skills or summary where they fit.
• Use the exact term from the postings, since CareerOneStop warns that a variation may fail to be recognised.

Can a checker read a resume that is an image?

No. CareerOneStop says that a scanned resume sent as an image will not be recognised. It tells applicants to use the requested format and, if none is given, Word, PDF or plain text.

This is why the plain text paste is a better first check than any score. If nothing appears when you copy from your file, no checker and no employer system has any text to work with either.

Sources

Every source here was fetched and checked before it was linked. Our own parser numbers are reproducible: render the same CV in both layouts and extract it.

• Huntr, how applicant tracking systems actually work (1.7M+ applications analysed)
• ResumeAdapter, where the 75% rejection figure actually comes from
• SelectSoftwareReviews, applicant tracking system statistics
• CareerOneStop (US Department of Labor), resumes
• MIT Career Advising and Professional Development, resumes
• Harvard College Mignone Center for Career Success, guide to creating a strong resume