What does an ATS see when it reads my resume?

An ATS sees extracted text, not your page. A parser pulls a stream of characters out of the file and maps it into fields — name, roles, dates, education. Columns, text boxes, and spacing are layout instructions that do not survive that step, so what gets stored can differ from what you see.

Two steps: extraction, then field-mapping

Extraction turns your file into plain text. A PDF is a set of instructions for placing glyphs at coordinates; there are no paragraphs or columns in it, only positioned characters. The extractor reconstructs reading order from those positions, and reconstruction is where a visually clear document can come apart.

Field-mapping then tries to work out which piece of that text is your most recent job title, which is the employer, and which dates belong to it. This is why standard headings and dates on the same line as their role matter so much: they are the landmarks the mapping step navigates by.

What this looks like measured, on our own pipeline

We ran one resume through five renderings of itself and published the numbers. Two findings are worth carrying around.

First, a two-column PDF extracted to 19 lines where the same resume as plain text produced 25. Nothing was missing from the word count — the columns were interleaved, so the sidebar arrived spliced through the work history. The scan still scored it 98. A good score did not mean a good parse, which is the single most useful thing that study taught us.

Second, before we fixed our own extractor in August 2026, that same single-column PDF came out as one unbroken line with 0 of its 9 bullets recognised, and the bullet sub-score collapsed to its floor. The failure was mechanical, silent, and invisible to the person who wrote the resume.

What we cannot tell you

We measured our extractor, not anybody else's. Workday, Greenhouse, Lever and Taleo publish nothing about how their parsers work, so any page claiming to know what a specific vendor stores — or that one auto-rejects at a score — is inventing it.

What generalises is the mechanism, not the vendor. Text is extracted from a file before anyone reads it, and layout that exists only visually is the part at risk. That is true wherever your file lands.

See your own extracted text in fifteen seconds

Open your PDF, select all, copy, paste into any plain text editor. What you are looking at is close to what gets stored. Read it for three things: are your dates still beside the right roles, are your bullets still separate lines, and has anything from a sidebar been welded into the middle of a sentence.

The free checker below does the same extraction in your browser and shows you the text it recovered, with no upload and no account.

Frequently asked questions

Does an ATS see my resume formatting at all?

Not as design. Bold, colour, and spacing are visual properties that do not survive text extraction, so they neither help nor hurt. Structural formatting is different: multi-column bodies, text boxes and tables change what the extracted text looks like, and that is what gets stored.

Does an ATS reject my resume automatically?

We cannot tell you what any specific system does, and neither can anyone who has not seen its configuration. What is reliably true is the shape of the funnel: your file has to parse into fields, a recruiter's search has to surface you, and knockout questions on the application form can screen you out before any of that.

Is a PDF or a Word document better?

Follow the posting's instruction first; if it names a format, that beats any general rule. In our own testing both extracted acceptably once known defects were fixed, with one caveat: Word's Symbol-font bullets arrive as Private Use Area codepoints rather than bullet characters, which is a real trap in .docx-derived files.

Why did my bullets disappear?

Usually because the bullet glyph came from a decorative or symbol font and was not recognised as a bullet marker. Our own parser once found 0 of 9 bullets in a resume that visibly had nine. A standard bullet character removes the problem.

Can I see exactly what a specific employer's ATS stored?

No, and nobody outside that company can. The closest honest proxy is to extract the text yourself and read it, because the extraction step is the part that generalises across systems.