What an ATS Actually Does With Your Resume

Most advice about applicant tracking systems is written to make you anxious enough to buy something. The underlying technology is much less dramatic than the advice suggests, and knowing what it really does changes which parts of your resume are worth worrying about.

An ATS is a filing cabinet with a search box

An applicant tracking system is the software a company uses to post a job, collect applications, store them, search them, and move candidates through stages. Greenhouse, Lever, Workday, Ashby and the rest are all variations on that. The core job is record keeping.

When you submit a resume, the system parses it into structured fields: contact details, employers, titles, dates, education, skills. That parsed version is what becomes searchable. Your original file is usually still attached, and a recruiter can open it.

Nothing in that description involves the software forming an opinion about whether you are good.

The scary statistic

You have probably seen a claim that some large percentage of resumes are rejected by ATS software before a human sees them. The number varies depending on who is selling what, which is a reasonable signal about how well sourced it is. I am not going to give you a number here, because I do not have one I can stand behind.

What is true is less alarming and more actionable: plenty of applications are never read by a human. That usually happens because a recruiter ran a search, looked at the first two pages of results, and filled the role. Not because an algorithm scored you and hit reject.

That reframing matters. You are not optimizing against a robot gatekeeper. You are trying to appear in a human's search results and survive their skim.

What actually filters you out automatically

Two things genuinely reject applications without human involvement, and neither is about resume formatting.

Knockout questions. The application form questions about work authorization, willingness to relocate, required certifications, or minimum years of experience. You answer these explicitly, and a "no" on a required one can close the application immediately. These are worth reading carefully, because they are the real automated filter.

Failed parsing. If the system cannot extract your work history, you end up in the database as a mostly empty record that matches no search. This is the formatting problem people worry about, and it is real, but it is narrower than the folklore suggests.

What actually breaks parsing

In rough order of how often it causes trouble:

• Text inside images. A resume exported as a picture, or with your name and contact details inside a graphic, is invisible to the parser. Scanned PDFs have the same problem.
• Multi-column layouts. A two-column design can be read straight across, interleaving your skills sidebar into the middle of a job description. The result is technically all there and completely incoherent.
• Critical information in headers and footers. Some parsers skip these regions. Your phone number and email should be in the body of the document.
• Tables for layout. Handled inconsistently. A table used to align dates against titles may come out scrambled.
• Invented section headings. "Where I've Made an Impact" is a worse heading than "Experience" for exactly one reason: the parser is looking for the words it knows.

What is mostly fine: a text-based PDF, a .docx, a sensible font, bold and italic, bullet characters, one column, and a normal amount of colour.

What does not work

Keyword stuffing. A wall of skills you do not have gets you into search results for jobs you cannot do, where a human reads two lines and moves on. You have spent your one impression.

White text on white background. This is old advice that was never good. It is trivially visible when the document is parsed or the text is selected, and it reads as an attempt to cheat.

Matching the posting word for word. Overlap should come from describing what you did in the vocabulary the reader uses. If the overlap requires claiming things you have not done, the ATS is no longer your problem. The interview is.

The short version

Make the document machine readable, then stop thinking about the machine:

1. One column, standard section headings, contact details in the body.
2. Text-based file, not an image or a scan.
3. Consistent date formats.
4. Real vocabulary overlap with the posting, limited to things that are true.
5. Answer the application form questions carefully, because those are the actual automated filter.

That is the whole ATS problem. Everything past this point is the ordinary difficulty of a human deciding you are worth thirty minutes, which is a question about what your resume says rather than how it was parsed. Tailoring to the specific posting is what moves that one.