What does ATS mean on a resume?

ATS stands for applicant tracking system, the software an employer uses to receive, store, search and rank applications. An ATS resume is one written and laid out so that software can read it into the right fields: single column, standard section names, real text. It is not a special format or a certificate; it is an ordinary resume that does not lose sections on the way in.

Do ATS systems reject resumes?

Not the way people imagine. An ATS does not read a resume and decide no; it parses it, files it and ranks it. The failure is silent: in 218 real resumes run through the ResumeAtlas checker rules, 16.5% did not parse into a usable candidate record, and nobody told those applicants. A resume that parses badly is not rejected, it is invisible to the recruiter's search.

Can an ATS read PDF resumes?

Yes, when the PDF has a text layer. We rendered 35 single-column templates as both PDF and Word and parsed both: 35 of 35 pairs reached the same result and the median score was 88 for PDF against 88 for Word. The file type is not the variable; a scanned image saved as PDF is, because it contains no text at all.

Is there one ATS score every company sees?

No. Each employer configures its own system, thresholds and must-have filters, so the same resume scores differently at two companies using the same vendor. A tool that shows you a score is estimating what a parser recovers, which is useful, but it is not a number the employer calculated.

Does every company use the same ATS?

No. Workday, Taleo, Greenhouse, Lever, SAP SuccessFactors and iCIMS are all common, and each uses its own text extractor, which is why the same file can read differently in two systems. On the 28 Canva templates we tested, 3 of 5 open-source extractors split letter-spaced headings into single letters and the others merged them back. The four stages above are the same everywhere; what survives them is the layout you chose.

How it actually works, with real rates

What is an ATS resume, and how does an ATS read one?

An ATS resume is a resume an applicant tracking system can read into the right fields: your name where the name goes, your jobs under Experience, your skills under Skills. The system does that in four stages, and the first two are where resumes quietly go wrong. In 218 real resumes submitted to ResumeAtlas (June 2026 – September 2026), 16.5% did not parse into a usable record, 40.8% were flagged for columns or tables and 39.4% for headings the parser could not recognise. None of those applicants was told. Below is what each stage does, how often it fails, and the fix; you can run your own resume through it free with no signup.

Check my resume ATS compatibility (free)

Free, no signup, see your own resume run through the four stages

Last updated: September 20, 2026

What does “ATS-friendly” actually mean?

It means the text comes out of the file in the right order and lands under the right headings. That is all. It is not a font, a file type or a template brand, and there is no certification behind the label. The three things that decide it, all measured on this site:

Text placed side by side. A sidebar or a date column beside the content interleaves under a plain read. 21 of the 28 Canva templates we tested do this; the Canva parse test has every row.
Letter-spaced headings. Wide tracking on small capitals is exported letter by letter; “E X P E R I E N C E” matches no heading list. 16 of 28 Canva templates use it, and 2 points of spacing was enough to break it in our font test, where the font itself changed nothing.
Headings the parser does not recognise. “My journey” instead of Experience, “Toolbox” instead of Skills. 39.4% of 218 real resumes were flagged for this in Stage 2 below; the section-by-section breakdown lists which headings people actually use and which ones map to nothing.

ATS-compliant and ATS-compatible are the same claim under other names; the format guide covers which layouts meet it.

How an ATS reads a resume, stage by stage

Stage 1: the file is opened and the text is pulled out

The system does not look at your page; it reads the text layer inside the PDF or Word file and gets back a list of fragments with positions. Rebuilding those into reading order is where the quiet failures happen. Two columns read as one interleaved stream, a heading with letter-spacing comes back as separate letters, and text drawn as a picture is not there at all. In 218 real resumes submitted to ResumeAtlas, 40.8% were flagged for columns or tables holding core content; across 28 Canva templates, 21 placed text side by side and 16 used letter-spaced headings.

40.8% of 218 resumes flagged for columns or tables, June 2026 – September 2026.

Fix: Single-column layout, real text (not an image), plain headings with no letter-spacing. Fonts do not matter; one resume rendered in 20 fonts parsed identically.

Stage 2: headings are matched and the text is filed under them

The extracted text is split into fields by recognising heading words: Experience, Education, Skills, Summary. Anything under a heading the system does not recognise is filed as unknown and drops out of scoring even though a person would understand it. In the same 218 resumes, 39.4% were flagged for headings the checker could not map, only 60.6% yielded all three core sections, and 16.5% did not produce a usable record at all. Contact details fare better: 7.8% lost the name, 5% the email, 6.9% the phone number.

39.4% of 218 resumes flagged for unrecognised section headings, June 2026 – September 2026.

Fix: Use the standard section names as their own lines: Experience, Education, Skills, Summary, Certifications. Put your name on the first line and contact details as plain text under it.

Stage 3: the filed text is matched against the posting

With the fields filled, the system looks for the posting's terms inside them: tools, certifications, job titles, methods. This step is more literal than people expect. Many systems treat an abbreviation and its full form as different words unless both appear, and synonyms are not reliably credited. This is the stage keyword tools measure; it is not a parsing failure, and no formatting fix changes it. Before you submit, you can compare your resume to the specific job description to see what terms it is missing.

Fix: Use the posting's exact terms in the bullets that describe real experience. Only where the experience is real.

Stage 4: scoring, ranking and the recruiter's search

Every application gets a relevance score and sits in a ranked list; recruiters also search the database directly for a term. A resume whose skills were filed under an unrecognised heading in Stage 2 does not come up in that search. Nothing rejects it. It is simply never found, which is why parsing failures cost more than weak phrasing: a person can forgive an average sentence and cannot read a section that never arrived.

Fix: Fix Stages 1 and 2 first. A score only moves once the text is in the right fields.

Passing the four stages does not mean a computer hires you. It means your resume reaches the list a recruiter opens and the search they run. Everything before that exists to decide whether a person ever reads what you wrote.

See how your own resume reads

This page explains the mechanism. The checker applies it: upload your resume and see which sections and contact fields a parser recovers, before a recruiter or a filter ever sees it.

Free ATS resume checker

Want the numbers behind every stage, check by check? ResumeAtlas ATS parsing benchmark.
Resume already has the columns or graphics that break Stage 1? Convert it to a clean single-column layout.
Want a pre-submit checklist instead of the explanation? 30-point ATS resume checklist.
Want a starting template that passes Stages 1 and 2? ATS resume template guide.
Got a score from a checker and wondering what it means? What is a good ATS score.
Format parses clean but still no callbacks? The remaining reasons are content, not mechanism: ATS rejection reasons.

FAQ

What does ATS mean on a resume?: ATS stands for applicant tracking system, the software an employer uses to receive, store, search and rank applications. An ATS resume is one written and laid out so that software can read it into the right fields: single column, standard section names, real text. It is not a special format or a certificate; it is an ordinary resume that does not lose sections on the way in.
Do ATS systems reject resumes?: Not the way people imagine. An ATS does not read a resume and decide no; it parses it, files it and ranks it. The failure is silent: in 218 real resumes run through the ResumeAtlas checker rules, 16.5% did not parse into a usable candidate record, and nobody told those applicants. A resume that parses badly is not rejected, it is invisible to the recruiter's search.
Can an ATS read PDF resumes?: Yes, when the PDF has a text layer. We rendered 35 single-column templates as both PDF and Word and parsed both: 35 of 35 pairs reached the same result and the median score was 88 for PDF against 88 for Word. The file type is not the variable; a scanned image saved as PDF is, because it contains no text at all.
Is there one ATS score every company sees?: No. Each employer configures its own system, thresholds and must-have filters, so the same resume scores differently at two companies using the same vendor. A tool that shows you a score is estimating what a parser recovers, which is useful, but it is not a number the employer calculated.
Does every company use the same ATS?: No. Workday, Taleo, Greenhouse, Lever, SAP SuccessFactors and iCIMS are all common, and each uses its own text extractor, which is why the same file can read differently in two systems. On the 28 Canva templates we tested, 3 of 5 open-source extractors split letter-spaced headings into single letters and the others merged them back. The four stages above are the same everywhere; what survives them is the layout you chose.

Rates: ResumeAtlas parsing benchmark, n = 218 distinct submitted resumes, June 2026 – September 2026, data generated 2026-09-20. Template and extractor figures from the Canva and font tests, generated 2026-09-20. One parser and five open-source extractors, not any vendor's product.