How to get past an ATS (and what that phrase gets wrong)

What an ATS actually does with your application

An applicant tracking system is database software for hiring. It receives applications, parses them into structured records, stores them, lets recruiters search and filter the pool, and tracks each candidate through stages from applied to offer. Most of its work is administrative.

What it is generally not doing is reading your resume and deciding your fate. In common configurations there is no automatic reject: there is a queue, ordered somehow, that a human works through. Knockout questions — the ones asking whether you hold a required licence or have the right to work in the country — do filter automatically, but those are answered in the form, not inferred from your resume.

This reframes the problem usefully. You are not trying to defeat a gatekeeper. You are trying to be parsed correctly, so your record is complete, and to be findable, so you surface when a recruiter searches the pool for the terms that define the role.

The 75% statistic, and why it has no source

You will see a version of this everywhere: "75% of resumes are rejected by an ATS before a human ever sees them." The figure appears in blog posts, in sales copy for resume tools, and in advice repeated by people who have no reason to doubt it.

It does not trace back to anything. Follow the citations and they lead to other articles repeating the same number, to a vendor's marketing page, or nowhere at all. There is no published study, no dataset and no methodology behind it — and the claim it makes is not one an outsider could measure anyway, since it would require access to the internal disposition data of a representative sample of employers.

It also does not describe how the software is usually configured. A system that silently discarded three of every four applications would be discarding qualified candidates at a rate no hiring team would tolerate, and in jurisdictions with employment-discrimination law it would be an unusual amount of legal exposure to take on voluntarily.

What is true instead

Most applications do go unread by a human — not because software rejected them, but because a posting can attract hundreds of applicants and a recruiter works down a ranked, filtered list until the role is filled. Being badly parsed or badly matched means being low in that ordering. That is a real problem with a real fix, and it is a different problem from being auto-rejected.

What actually works

1. Make the file parse cleanly. Single column, real text, standard headings, contact details in the body. This is the foundation and it is covered in full in ATS-friendly resume format.
2. Use the posting's words for the work you did. Recruiters search the pool using the terms in their own posting. If your resume calls it something else, you are not in the results. See resume keywords.
3. Put the relevant experience where it is seen first. The top third of page one carries most of the weight, in both passes.
4. Answer the knockout questions accurately. These are the only filters that genuinely reject automatically, and they are usually about eligibility rather than skill.
5. Apply while the posting is fresh. A ranked queue is worked from the top and from the beginning. Late applications compete against a list a recruiter has already started shortlisting from.

Tricks that backfire

• White text and hidden keywords. Parsers extract text regardless of colour or size, so the terms appear in the record — and then the recruiter opens the file, selects the text, and sees them. It reads as deception because it is.
• Pasting the whole job description in a tiny font. Same mechanism, same outcome, plus a document that describes the job rather than the candidate.
• Keyword lists with no evidence. Terms without accompanying work raise a match figure and lower a human reader's confidence in everything else on the page.
• Inflated titles. Titles are checked in references and background checks more often than any other field. If your title was unusual, add the conventional equivalent in brackets rather than replacing it.
• Applying to everything. Some systems show recruiters every role you have applied for at that company. Thirty unrelated applications is itself a signal.

The part the resume cannot solve

A well-parsed, well-matched resume gets you into the shortlist a human reads. It does not get you past the reasons that have nothing to do with your file: the role being filled internally before it was posted, a referral already in the pipeline, a hiring freeze that has not reached the careers page yet.

Which is worth saying because the counterfactual advice — spend another two hours on keyword density — has a worse return than one message to someone who works there. The resume is the part you control, so make it clean and matched, then spend the remaining effort on the parts of the process that outrank it.

For what the score behind all of this actually measures, see what an ATS score is.