What Applicant Tracking Systems Actually Do With Your Resume

A walk through what happens to a resume between upload and recruiter screen — which formatting choices survive parsing, and which quietly destroy an application.

By TODO_BEFORE_LAUNCH(PlaceMeFast)August 10, 20266 min read

In short

Most applicant tracking systems do not score or reject your resume. They store it, extract text from it, and make it searchable for a human.
The damage from bad formatting is not rejection. It is a resume that parses into the wrong fields and never surfaces in the searches recruiters actually run.
Multi-column layouts, text inside tables, and content in headers or footers are the three structures most likely to be extracted in the wrong order or dropped.
"Beat the ATS" advice built around keyword density is aimed at a system that mostly does not work the way people think.

There is a number that follows job seekers around: that some large majority of resumes are "rejected by the ATS" before a human sees them. It is repeated in career advice, in LinkedIn posts, and in the marketing copy of products that sell you a fix for it.

It is worth being precise about what these systems do, because the real failure mode is different from the one people prepare for — and the preparation aimed at the imagined problem often makes the real one worse.

What an applicant tracking system is

An applicant tracking system is a database with a hiring workflow attached. Greenhouse, Lever, Ashby, and Workday are the ones a senior candidate in the U.S. meets most often. Their job is to hold candidates, applications, interview stages, scorecards, and offers in one place so a recruiting team can work from a shared record.

Automatic rejection is not the core function. In most configurations, applying puts you in a pool. A recruiter or coordinator then works through that pool — sorting, searching, filtering, and reading.

Some systems can be configured with knockout questions: work authorization, location, a required licence. Those are explicit, and you answered them yourself on the form. That is a rules engine acting on your answers, not a machine reading your resume and forming an opinion about you.

What happens to the file you upload

Between upload and the recruiter's screen, your resume goes through a parser. Roughly:

Text extraction. The parser pulls a stream of text out of the PDF or DOCX.
Structural detection. It looks for the boundaries of sections — experience, education, skills — usually by recognising conventional headings.
Field mapping. It attempts to fill structured fields: name, email, phone, then each role with an employer, title, and date range.
Indexing. The extracted text is stored so it can be searched later.

Every one of those steps can partially fail, and — this is the important part — partial failure is usually silent. Your application still submits. You still get the confirmation email. Nothing tells you that your last two roles landed in the wrong field, or that your headings were never recognised as headings.

Where parsing actually breaks

Multi-column layouts

The single most common cause of scrambled output. A PDF has no concept of columns; it has text positioned on a page. Extraction generally walks the content in reading order, and with a sidebar layout that order may run straight across the visual columns, interleaving your skills list with your job titles.

The result is text that still contains all your words but no longer contains coherent sentences. A human reading the original sees a clean two-column design. The parser produced something else entirely.

Tables

Tables are the same problem with more structure. A skills matrix or a date-and-role table often extracts as a run-on sequence with the relationship between cells lost. Where the table carried the meaning — this date belongs to this employer — the meaning is what gets dropped.

Headers and footers

Contact details placed in the page header are a persistent trap. Some extractors treat header and footer content as page furniture and skip it. If your email address lives only in the header, you can end up with a parsed record that has no email address on it.

Text rendered as an image

A logo containing your name, a skills graphic, a rating chart. None of it is text, so none of it is extracted. It is invisible to search.

Unconventional section headings

Parsers recognise sections by matching against expected headings. "Experience", "Work Experience", "Professional Experience", and "Employment History" are all understood. "Where I've Made an Impact" is not. Rename the section and the parser may not find the boundary at all.

The tell

Open your resume in a plain text editor, or copy everything out of the PDF and paste it into a blank document. What you see is close to what the parser sees. If it reads as scrambled, the file is the problem — not your experience.

Why the real failure is invisibility, not rejection

Here is the part that reframes the whole exercise.

Recruiters working a pool of applicants search it. They search for a title, a technology, a certification, a former employer. That search runs against the extracted text and the mapped fields.

If your resume parsed cleanly, you appear in results where you are relevant. If it parsed badly, you are still in the database — you are simply not findable. No one rejected you. No one made a decision about you at all. You are a row that never matched a query.

This distinction matters because it changes what you optimise for. If the enemy is an automatic rejection engine, the strategy is to game a score. If the reality is a search index, the strategy is to be accurately represented in it. Those two strategies produce very different documents.

What actually helps

Use a single-column layout. This is the highest-value change available, and it costs you nothing but visual flourish.
Use conventional section headings. Experience, Education, Skills. Save the creativity for the content.
Put contact details in the body, not in the page header.
Keep dates in a consistent, conventional format — Jan 2021 – Mar 2024 — and put them on the same line as the role.
Submit the format the employer asks for. When a system offers both, DOCX generally parses more predictably than PDF, because the structure is explicit rather than inferred from position.
Use the language of the job description, because that is the language the recruiter will search with. This is not keyword stuffing. If the posting says "site reliability" and your resume says "infrastructure quality", a search for the former will not find you.

What does not help

Keyword density targets. White text hidden on a white background — which is detectable, is read by humans reviewing the extracted text, and reads as dishonesty when found. Long lists of skills you cannot speak to for twenty minutes. "ATS score" percentages from tools that have no access to the employer's actual system and are scoring against a model of their own invention.

The honest summary

Applicant tracking systems are mostly boring infrastructure. They are not gatekeepers with opinions. They are databases that need your document to be legible in order to represent you accurately, and the cost of illegibility is not a rejection letter — it is silence.

Formatting your resume for machine legibility is not gaming a system. It is making sure the record of your career survives the trip into a database intact.