ATS · 11 min read

How to pass an ATS in 2026: what the software actually checks

You pass an ATS by doing three unglamorous things well: use a single-column layout, mirror the job's own wording where it's true, and keep the formatting clean. An applicant tracking system (ATS) almost never auto-rejects your CV (called a resume in the US). It is mostly a database that reads, stores, and ranks applications so a human recruiter can search them.

The widely repeated claim that "75% of CVs are filtered out before a person sees them" has no credible source that we could find. What actually buries a CV is a layout the software can't read and content that doesn't obviously match the job. The rest of this guide walks you through what an ATS does, what really gets a CV filtered, and the short list of fixes that matter.

Note on terms: we say CV throughout; in the US the same document is called a resume. The ATS rules below are the same either way.

Q.Does an ATS automatically reject your CV?

No, and the people who actually run these systems say so themselves. In a 2025 study (reported by IT Brief), the CV-tools company Enhancv interviewed 25 US recruiters across tech, healthcare, finance and other sectors: 92% said their systems do not automatically reject applications based on formatting, design, missing keywords, or a low match score. Treat it as a small vendor sample rather than an industry measure. Only about 8% used any automatic rejection at all, and only for hard "knockout" requirements such as a legally required license or work authorization.

So where did "75% of resumes are never seen by a human" come from? According to The Interview Guys and recruiter Jan Tegze, it traces back to a 2012 claim by a resume-optimization company called Preptel, which they say went out of business in 2013 and never published a methodology. It has been repeated for years in posts and templates, and we could not find a study behind it.

Here's the part that is real, and worth taking to heart: if you're genuinely qualified and still getting passed over, you're not imagining it. Rigid screening does filter out capable people by matching exact criteria too literally, without any malice. A 2021 study by Harvard Business School's Project on Managing the Future of Work and Accenture, Hidden Workers: Untapped Talent, found that 88% of employers agreed qualified, high-skilled candidates get screened out simply because they don't match the precise criteria in a job description (rising to 94% for middle-skills roles). The same report estimated more than 27 million "hidden workers" in the US, meaning capable people whose applications quietly get passed over. If that's been your experience, it's often less about your ability than about a system reading too literally, and that's something you can work with.

The takeaway: the danger is a system matching exact criteria, plus sheer volume. A robot deleting you is rarely the issue. The fix is concrete: make your match obvious, and make your CV easy for the software to read.

Q.What does an ATS actually do?

Think of an ATS as four steps: parse, store, search, rank. As the University of Virginia's career center explains, the system scans your CV, compares it against the job description, and helps recruiters sort a large pile of applications; some systems score each CV by how well it matches, and recruiters then either work through the ranked list or search the database by keyword. Systems you may meet include Workday, Greenhouse, Lever, iCIMS and Taleo.

Two things follow from that, and both are good news:

  • The match score is a hint, and a human makes the call. Many recruiters use it only to order the pile, and some ignore it. A high score doesn't guarantee an interview, and a middling one doesn't doom you, because a human still decides.
  • Many applications get no reply, and scale is part of the reason. A single posting can attract hundreds of applicants (CareerPlug's 2025 analysis of millions of applications put the average near 180 applicants per hire), and overloaded readers are likely to start with the CVs that are obviously relevant. Being easy to read and clearly on-target is simply how you rise in that pile. The full checklist of where an application can stop is in why am I not getting interviews.

Q.What actually gets your CV filtered out?

Three real things, none of them mysterious and all of them in your control:

  1. The parser can't read your layout. Columns, tables, text boxes, and graphics get scrambled or dropped, so your experience never makes it into the database cleanly.
  2. Your CV doesn't obviously match the job. If the role asks for specific skills, tools, or a job title and they're nowhere on your CV, neither a keyword search nor a human reader is likely to find you.
  3. Weak content. Even when a person reads it, a CV that lists duties instead of results doesn't stand out. Harvard's Mignone Center for Career Success advises this: lead with action verbs, show results, and quantify them — "responsible for managing a team" is far weaker than "managed a team of 8 and cut onboarding time 30%."

Some things you may have read about are not documented filters. None of the career-center checklists above sets a length rule, so going slightly over one page is mainly a question for the person reading, and we found no source saying an ATS rejects a CV for skipping an "objective" line. Shading is different: the same checklists ask for one color, so keep the page plain.

The full self-check of the page, with before and after examples, is in what is wrong with my resume.

Want to see this on your own resume? Check it against a job posting, free and with no account.

Q.How should you format a CV so an ATS can read it?

The most important rule is to use one column, left-aligned. Multi-column layouts are the most common way to get scrambled, because many parsers read straight across the page and interleave your two columns into nonsense (see two-column resumes and ATS). University career centers are unusually consistent here: Virginia, Pittsburgh and Ohio Northern give much the same checklist:

  • One column, left-aligned, and not centered or justified.
  • No tables, text boxes, columns, images, or graphics. Parsers drop or jumble them; anything inside a text box often isn't read at all.
  • Standard section headings ("Work Experience," "Education," "Skills") rather than creative labels like "What I've Done," which the system may fail to map.
  • Standard fonts (Arial, Calibri, Georgia, Times New Roman), 10–12pt, one color.
  • Put your contact details in the body of the document, not the header or footer. Some ATS can't read header/footer text, so a name or email placed there can vanish. Your top line should be just your name.
  • Margins 0.5–1 inch. Skip page numbers in the body.

A photo deserves its own note: it can break parsing, and in most English-speaking markets it also invites bias, so leave it off for ATS submissions. (In parts of Europe a photo is still expected, and those norms vary country by country; see photos on German and EU resumes.)

Q.PDF or Word: which file should you upload?

Both work for most modern systems, so you can relax about this one. As UVA's career center notes, a PDF, .doc or .docx is fine for most submissions, and a PDF has the advantage of holding your layout together across devices (more in PDF vs Word). Older or stricter systems sometimes parse .docx more reliably, so if you're unsure, Word is the safe fallback.

There are only two real traps:

  • Never upload an image-based PDF. A CV that was scanned or exported as a picture contains no selectable text, so a parser has nothing to read. Pittsburgh's career center spells it out: save or export as PDF, and don't print to PDF.
  • If the posting names a format, use that. Following instructions is itself part of the screen.

A quick check: open your PDF and try to highlight a sentence with your cursor. If the text selects cleanly, that is a good sign the parser can read it, though not a guarantee of a clean parse. If you can't select it, a parser probably can't either.

Q.How do you check your CV parses correctly?

You don't have to guess, and you don't have to hope. Three quick tests, in order of usefulness:

  1. Run the cursor test by trying to select the text in your PDF. (If you can't, it's an image.)
  2. Run the copy-paste test by pasting your whole CV into a plain-text editor. If sections jump out of order or disappear, the parser will see the same mess. This is the fastest way to catch a broken two-column layout.
  3. Compare your CV against a specific job posting in a CV-and-job-description analyzer and look at which of the posting's terms it found and which it says are missing, as well as the score. That shows whether your key content is visible.

This is the honest version of an "ATS score," and it's how CVder's check is built: it shows you what was matched and what's missing against a specific job, rather than handing you a black-box number to worry over. The goal is to see where your CV and the posting line up, and a percentage matters less.

Not sure which of these your resume is missing? Run the free check.

Q.Do keywords matter, and how many?

Yes, but it's about matching, and stuffing is the wrong game, which takes a lot of the pressure off. Read the job description and mirror the exact terms it uses (the named skills, tools and job titles, especially ones that appear more than once) and place them where they're genuinely true: in your skills section and, better, inside your experience bullets. UVA's guidance is to pull the hard and soft skills that recur in the posting and work them in through action verbs.

Two limits keep this honest:

  • Don't list skills you don't have to "trick" the system. It backfires the moment a human reads the CV, and recruiters spot padding fast.
  • There's no magic number. Chasing a keyword count is the wrong game; relevance is the right one.

The full method for matching a CV to a posting lives in our guide on how to tailor your resume to a job description; the narrower question of how many keywords and how to avoid stuffing is covered in how many keywords a resume should have.

TL;DRThe short version
  • Single column, standard headings, standard fonts, no tables or graphics.
  • Contact details in the body, not the header/footer.
  • A text-based PDF (or .docx), never a printed/image PDF.
  • Mirror the job's real wording where it's true; don't stuff.
  • Lead with quantified results, not duties.
Sources

Base material adapted (logic only, fully rewritten in original wording) from a resume guide by Ksu Dranitsina. Her self-reported response-rate figures were excluded as unverifiable. Author credit to Ksu pending her okay.

FAQ

Quick answers

It can fail to read them (graphics, text boxes and images often don't parse), and a photo also invites bias in most English-speaking markets. Keep the document text-based and single-column. (Photo norms differ across Europe, country by country.)

Both are fine for most modern systems. A text-based PDF preserves your layout; .docx is the safe fallback for older systems. Never upload an image-based or "printed-to-PDF" file, and if the posting asks for a specific format, use it.

In the sources above, none of the career-center checklists lists a length rule, and length is mainly a human-reading preference. Employers configure their own systems, so follow the posting if it sets a limit. How long your CV should be is a separate question, covered in How long should a CV be?.

No. Mirror its real terms where they're true and relevant; don't copy it wholesale or stuff in skills you don't have. See the tailoring guide for the method.

That figure comes from a commercial eye-tracking study with a small sample, later updated to 7.4 seconds. It is useful as a reminder that recruiters skim, but it is not a number to build your CV around. We cover how recruiters actually read a CV in What recruiters scan first.

Sometimes. In the Enhancv study above, the few recruiters who used any automatic rejection said it was for hard knockout requirements such as work authorization or a required license, so a very fast rejection can mean an employer-configured knockout question, though we cannot tell from outside what happened in a given case.

We found no source for any universal cutoff. Systems that score applications do so by their own rules, and in the Enhancv study 92% of the recruiters said their systems do not automatically reject on a low match score, though the sample was only 25. Treat any single threshold as unverified.

No. The score is one input that helps recruiters sort applications; a human makes the decision. Clear relevance and quantified results matter more than any number.

You might also be interested in

Get your score for this job.