How ATS parsing actually works in 2024–26

What's actually running

Greenhouse is one of the few ATS vendors that documents what breaks its parser. Its help center lists the specific formats that fail extraction²:

• graphics and photos
• resumes uploaded as an image rather than a document
• tables
• headers and footers holding the candidate's name or contact info
• columned layouts
• files larger than 2.5MB

The parser accepts .doc, .docx, .pdf, .rtf, and .txt, with no stated preference between PDF and DOCX³.

Workday and SAP SuccessFactors are the two most common ATS platforms for Fortune 500 employers after Greenhouse⁴. Neither publishes a comparable candidate-facing formatting guide, and neither does iCIMS. Blogs claim specifics like "two-column layouts drop 4 of 10 work-history entries in Workday," but none trace to a vendor-published source. Treat such numbers with the same skepticism you would give the 75% figure.

What the parser produces is a structured record: Name, Email, Phone, Company, Title, Dates, Skills. That record is what the recruiter searches later. If your name sits in a page header the parser skipped, your ATS record's Name field is blank, and a search on your first name won't find you, even though you applied.

The 75% myth, and what a 2025 study actually found

The 75% figure has no verifiable origin. No credible, methodology-disclosed source has ever produced this number¹. No peer-reviewed study, no ATS vendor disclosure, no recruiter survey has ever reproduced it. When you next see it quoted with confidence, you are looking at folklore, not evidence.

What is disclosed and reproducible: Enhancv's 2025 research team ran 25 standardized-question interviews with U.S. recruiters across eight industries — tech, healthcare, finance, manufacturing, CPG, energy, education, publishing. Companies ranged from under 100 to over 50,000 employees. 92% of the recruiters (23 of 25) said their ATS is not configured to auto-reject resumes on formatting, design, or content¹. The two exceptions used strict, disclosed thresholds — "match less than 75%" or "fewer than 7 of 10 required skills."

The study points to volume, not automation, as the reason your resume goes unseen.

What recruiters actually do with ATS

A 384-recruiter survey fielded in early 2025 found that 99.7% of recruiters use filters in their ATS or a similar system to narrow candidates before opening a resume⁵. 76.4% filter by skills first; 55.3% filter by job title. Filtering isn't auto-rejection: the resume stays in the system, tagged and searchable, but filtering decides which resumes a human ever opens.

Once a resume surfaces, the read is short. The most-cited industry estimate, a 2018 update to a Ladders eye-tracking study, puts initial recruiter attention at 7.4 seconds per resume⁶. The eye-tracking sample size isn't disclosed, and the vendor sells resume-anxiety-adjacent products, so treat 7.4 as the most commonly cited number rather than a rigorously validated one. Even at twice that figure, nothing in that window rewards typographic cleverness.

The pattern holds across both channels. A one-column, standard-heading resume the parser can extract cleanly is also the one a person can skim fast. There's no separate machine-readable and human-readable version to build: a layout that works for one works for the other.

Two things to act on today

Two places to start:

• Rebuild the resume in a one-column layout. No tables, no icons, no photo, no headers or footers holding your contact info. Chapter 4 walks through the five steps and the exact section-heading conventions the parser looks for.
  Read Chapter 4 arrow_forward
• Check whether your Verifiable layer is the one costing you. The Scorecard's routing table flags the layer scoring lowest across the five: Findable, Followable, Verifiable, Credentialed, Introducible. Four minutes, no signup.
  Take the Scorecard arrow_forward