What an ATS Actually Does to Your Resume File

Does Software Read and Score Your Resume?

An applicant tracking system does three things to your file: parses the text into fields, stores the result as a record, and gives a recruiter a way to search or rank it. None of the three grades your prose. The layer is common: in a survey of 2,275 senior leaders in Germany, the UK and the US in early 2020, 63 percent said they use a recruitment system alongside the core ATS ¹.

• Parse: read the file and extract fields such as name, contact details, employment history and education.
• Store: keep both the original file and the extracted fields attached to your application.
• Rank or filter: order the stored records, or remove some of them, against rules or a search a person set up.

Employers with more than 1,000 workers reported using it more often still, at 69 percent, and more than 90 percent of those running the layer said they use it to filter or rank candidates before a person looks at the pile ¹. That is where the fear comes from, and it is real: being filtered is a genuine outcome. But which company happens to own which system explains less of what happens next than the folklore assumes. A study that sent more than 83,000 fictitious applications to 108 of the largest US employers found that firms differ sharply in how often they contact applicants with distinctively Black names, and that the third-party system powering a firm's hiring site explained only 0.1 percent of that variation across firms; the authors conclude that screening algorithms are unlikely to drive the differences between employers ². What varied was the employers themselves, and the top fifth of firms accounted for nearly half of the lost contacts. The brand on the login page is not the thing separating one employer from another.

What the Parser Actually Extracts

A parser reads the text it can find and tries to map it onto fields: name, contact details, employment history, education, sometimes a skills list. Workday's own documentation says parsing results can vary by resume format and word order, and recommends avoiding images or image-based styles for the best results ³.

Greenhouse publishes a longer, more specific list of what breaks its parser: a file over 2.5 megabytes, letters typed with spaces between them, graphics or word art, a resume uploaded as an image rather than a document, tables, headers and footers, contact details placed in a header or footer, a columned layout, sections with no clear boundary, a company name with no suffix like Inc. or LLC, and an abbreviated title such as "Sr. Account Exec" instead of the full words ⁴. Most of that list describes extraction failing quietly, which is a different event from an application being rejected.

A failed parse does not end your application. When one happens on Greenhouse, the resume stays attached to your candidate record and a person on the hiring team types the missing fields in by hand ⁴. The cost is a slower, more manual entry on the recruiter's side, not disqualification on yours. It is still worth formatting for the parser, because Greenhouse names the specific choices that break extraction and every one of them is free to avoid ⁴.

Structure Your Resume So It Parses Cleanly

Structure the document for extraction first and for a human reader second, because both readers see the same file. Use one column, standard section headings, and plain body text rather than tables or text boxes for anything you need read.

• Put your name and contact details in the body of the page, not in a header or footer.
• Use a single-column layout; Greenhouse names a columned resume on its own list of what breaks its parser ⁴.
• Save as a plain .docx or .pdf rather than an image, a scan, or a design-tool export.
• Write full job titles and full company names, with a legal suffix where the company has one.
• Skip tables, text boxes, icons and graphics for anything you want a machine to read.

Once the file parses, a recruiter's own search still runs on the words you chose. Greenhouse's Talent Filtering tool lets a recruiter search resumes and notes for keywords, and states plainly that a search term has to exactly match the word in the application to surface it ⁵. That is a reason to use the exact phrase the posting uses for a tool or a skill you genuinely have, not a reason to bury a term list where a person will not read it.

Why Fast Rejections Happen, and What Is Fixable

A rejection that lands in minutes is almost never a person reading your resume and deciding no. Workday's documentation describes automatic stage routing: an employer sets condition rules that advance or decline a candidate the moment they apply, using the application, the requisition and the questionnaire just completed ⁶. The vendor's own example is declining someone ineligible to work in the country where the job sits.

That is a rule firing on a stated fact, not a reading of the resume text, and it splits what you can fix from what you cannot. A rule tied to a form question, work authorization, a license, a location, is something you can answer accurately and sometimes negotiate before you submit. A rule tied to years of experience or an exact-match keyword is worth naming for what it is: a crude filter that the employers running it openly describe as losing people who could do the job ¹. If a specific dropdown answer is what actually ends applications for you, knockout questions are the mechanism worth understanding next, because they are the one part of the form built to reject on its own.

Employers on the other side of the same screen are already being told what replaces a keyword filter once every resume matches it: keep only the requirements a third party will confirm, delete the scored terms around them, and send a short work sample earlier. Knowing that is the honest reason to stop treating the resume as the whole contest.

See a sample report >