CAREER RESOURCE

How applicant tracking systems read your resume (from vendor docs)

What an ATS does with your resume: parsing, keyword search and optional AI matching, taken from vendor help pages. Plus the formatting that commonly breaks parsing.

By ROLIVA · reviewed by Raghavendra N (owner) · published 2026-10-05 · facts last checked 2026-10-05

What an applicant tracking system is

An applicant tracking system (ATS) is the software an employer uses to collect applications, move candidates through hiring stages and find people again later. Think of it as a database with a workflow on top. Recruiters and hiring managers log in to it; you usually meet it only as the "Apply" form on a careers page.

Two things are worth knowing before anything else:

  • Each employer configures its own system. Two companies on the same ATS can use different stages, questions, filters and add-ons.
  • The vendors publish help pages. Instead of guessing, you can read what the software is documented to do. This article uses one vendor's documentation, Greenhouse, because its help centre describes parsing, searching and its optional AI features in plain language. Everything below about Greenhouse comes from those pages, re-read on 5 October 2026. Other systems may behave differently, and we do not describe ones whose documentation we have not checked.

Step 1: parsing turns your file into profile fields

When a resume is added, Greenhouse Recruiting "scans an imported resume and auto-fills appropriate fields with information it detects", according to its help page on unsuccessful resume parses. That is parsing: the software reads your document and tries to copy your name, contact details, employers, titles and education into structured fields.

The same page explains what happens when parsing fails. The recruiter sees a message that the resume "has only been attached to the candidate", and the details have to be typed in by hand. Your file is not deleted and you are not automatically rejected for it. But a profile that has not been filled in is harder to search, and someone has to do extra work to read it.

What makes parsing fail, from Greenhouse's own list

Greenhouse lists the causes directly. Paraphrased from the same help page:

  • File size. Greenhouse Recruiting "can't parse resumes larger than 2.5MB". Large embedded images are the usual reason a resume gets that big.
  • Letters with spaces between them. A heading typed as spaced-out letters looks like a word to you, but the parser cannot join the letters back together.
  • Graphics, photos or word art.
  • A resume uploaded as an image rather than a document such as a .docx or .pdf.
  • Tables, headers and footers.
  • Name and contact details in a header, footer or text box.
  • A columned layout.
  • No clear sections, or a different format in each section.
  • Company names without identifying words such as Inc., Co., Ltd or LLC, and shortened job titles (its example is "Sr. Account Exec" instead of "Senior Account Executive"). These can lead to a partial parse that someone has to correct.

A second help page, why a resume did not import correctly, adds three more reasons: the file "doesn't have enough information, is a non-standard file type, or it includes characters that aren't recognized" by the text parser.

Notice what is on this list: layout and file problems, not your choice of font colour or the number of pages. The practical fix is a plain, single-column document with your contact details in the body of the page, standard headings and real text. Our guide to an ATS-friendly resume structure walks through that layout.

Step 2: recruiters search and filter

Parsing is about getting your information in. The next step is people finding it.

Greenhouse's help page on searching resumes for keywords describes a Full Text Search on the All Candidates page: a recruiter types search terms and sees matching candidates, with a snippet showing where the words appear.

Its Talent Filtering page goes further. During application review, a recruiter can search "for job titles, skills, locations, and other keywords in the full text of resumes and internal notes", then filter by things like location, referrals or education. Two details matter for you:

  • Keywords can be marked Preferred (any one term must appear) or Required (all terms must appear).
  • "The keyword from your search must exactly match the keyword in the application."

This is the honest reason the job description's own words matter. A person reviewing applications may type the posting's terms into a search box. If you have done the work but described it in different words, a literal search may not find it. The fix is to use the employer's term where it is true for you, which is the subject of our guide to tailoring a resume to a job description.

Step 3: optional AI features vary by employer and plan

Some ATS vendors now sell AI features. Greenhouse lists its own on a Greenhouse AI features page (last updated 3 September 2026). Two are relevant to applicants:

  • Keyword suggestions, listed for all tiers, which suggest search keywords from the job post. The Talent Filtering page also mentions "suggested keywords generated from your public job post".
  • Talent Matching, which "compares candidate resumes against your defined and weighted calibration". The page lists it as available on Core, Plus and Pro with the Real Talent add-on.

So whether AI compares your resume against anything depends on whether that employer bought and switched on an add-on, and on the calibration the employer defined. Nobody outside the employer can see that configuration. Treat any claim about "how the ATS scores you" with that in mind.

Myths worth dropping

"75% of resumes are rejected by a robot." This figure circulates widely. We could not find a primary source for it, and the vendor documentation above describes parsing, search and optional matching rather than automatic rejection. We do not repeat it.

"Every ATS gives your resume a score." There is no universal score. Some optional features rank or match candidates against criteria the employer sets. Tools that show you a number are making their own estimate; they cannot see the employer's set-up. ROLIVA's own checker deliberately shows a plain count of terms rather than a score.

"Hide keywords in white text." Parsers extract text regardless of its colour, and the text then sits in front of a person reviewing your profile. Hidden keyword lists tend to look like what they are.

"PDFs get rejected." Greenhouse's supported formats page lists doc, docx, pdf, rtf and txt as accepted uploads. The problem case is an image-only PDF, not PDFs in general. See PDF or DOCX for the detail.

Which ATS is the employer using?

You can often tell from the address of the application page. Greenhouse-hosted job boards typically sit on a greenhouse.io address, Lever's on jobs.lever.co and Workday's on a myworkdayjobs.com address. Many companies embed the form in their own careers site, so the address is not always a reliable signal.

Every job listed on ROLIVA links to the employer's own application page, so you can see where you will actually apply before you start.

Questions applicants often ask

Does the recruiter see my original file, or only the parsed fields? Both can exist. When parsing fails, Greenhouse says the resume is still "attached to the candidate". Its supported-formats page also mentions that Greenhouse converts a resume to PDF for preview, and that externally linked images are not loaded in that preview while images embedded in the file are. So a person can open your document; the parsed fields are what make it findable.

If I upload a resume, should I still fill in the form fields? Yes, where the form asks. Greenhouse describes partial parses, where some information is filled in and some skipped. Checking the fields an application form pre-fills, and correcting them, is the simplest way to make sure the profile says what your resume says.

Will a designed template get me rejected? Nothing in the documentation above describes automatic rejection for layout. What it describes is a parse that fails or is incomplete, which leaves your information harder to search and more work to read. That is reason enough to keep the layout plain.

Is the job description really the best source of keywords? For one specific application, yes. Greenhouse's Talent Filtering page says it can suggest keywords "generated from your public job post", so the posting is a reasonable guide to the words a reviewer may search for. Use them only where they describe what you have done.

What you can control

  1. Send a text-based file in the format the employer asks for.
  2. Keep one column, standard headings and contact details in the body of the page.
  3. Keep the file small; images are the usual cause of very large files.
  4. Use the posting's terms where they truthfully describe your work.
  5. Read the result as a person would: if the order of your text is confusing when copied into a plain text editor, fix the layout.

Check your own file

ROLIVA's free ATS resume checker compares your resume with a job description. It shows which of the job's recognised skills and terms your resume mentions, as a plain count rather than a score, and flags structure that commonly causes misreads, such as tables, columns, icons and contact details missing from the top. It uses fixed rules and no AI model, and nothing you paste is stored.

If you want to start from a clean layout, the free resume templates are single-column DOCX files with placeholders in square brackets.

Sources

Each source was re-read on 2026-10-05. Help pages change; check the linked page for the current wording.

Continue in ROLIVA

Check your resume against a job description, free ↗

Related resources