ATS-friendly resume formatting in 2026: what passes and what fails

Most resume advice is downstream of one fact: a piece of software reads your resume before any human ever does. In 2026 that software is the applicant tracking system, and the major ones — Workday, Greenhouse, Lever, iCIMS, Taleo — all do roughly the same thing. Understanding what they actually do (and where they fail) is the difference between getting a callback and getting auto-rejected by a regex.

This post is opinionated. It is also the longest, most concrete writeup we have of why some resumes parse well and others do not, based on running real resumes against the parse APIs of five widely-used ATSes. If you only have time for one section, jump to the five formatting choices near the bottom — that is where the rubber meets the road.

The pipeline, in order

An ATS does three jobs the moment your PDF lands in its queue. First, it extracts the raw text from the file. Second, it tags spans of that text into structured fields — name, email, phone, work history, education, skills. Third, it runs that structured record against the job's keyword and rule-based filters. Each step has its own failure mode, and your formatting choices decide which ones bite you.

Step 1: text extraction

A PDF is not, fundamentally, text. It is a description of where characters appear on a page. Modern ATSes use one of two strategies to recover the text: they read the embedded text stream the PDF generator wrote (fast, accurate) or they OCR the rendered page (slow, lossy). If your PDF was exported from a real word processor or our builder, the embedded stream is clean and lossless. If your PDF is a scan or was stitched together from images, you are at the mercy of OCR — and OCR confuses ligatures, drops bullet glyphs, mangles tables, and frequently transposes characters from adjacent lines.

A useful sanity check: open your PDF in any reader, select all text with Cmd-A or Ctrl-A, copy it, and paste it into a plain text editor. If the result reads top-to-bottom in the order you intended, an ATS will see roughly the same thing. If the paste is scrambled, missing your name, or breaking your bullets into single characters, that is exactly the picture the ATS is forming.

Step 2: section tagging

After extraction, the ATS scans for section headers it recognizes — Experience, Education, Skills — and slots the lines beneath each header into the right field. If your header reads Career Highlights or What I've Done, you are rolling the dice on whether the system maps it back to experience. The simplest, most ATS-friendly thing you can do is use the boring header names. Reserve creativity for the content under those headers.

Step 3: keyword and rule matching

Finally, your structured record is compared against the job posting's keyword expectations and any human-defined rules — 5+ years Python, based in the EU, security clearance preferred. This is the layer recruiters interact with. When a recruiter searches a candidate pool, they are searching the structured fields the ATS built, not your original PDF. The implication is sharp: anything the ATS failed to extract or tag in step two is invisible to the recruiter, even if it is right there in your document.

The five formatting choices that quietly tank parse rate

These come from running 200 real resumes through five major ATSes and watching which ones produced clean structured records and which ones produced garbage. Ranked roughly by how much they hurt:

• Multi-column layouts. ATSes read left-to-right, top-to-bottom. A two-column resume often gets read as left-column-line-one, right-column-line-one, left-column-line-two — interleaving your skills section into your work history. Single-column resumes parse 30 to 40 percent more reliably in our testing. If you must use two columns, use our ats-two-column template, which arranges the columns into table rows so reading order is unambiguous.
• Text inside graphics, icons, or shapes. If your name or your skills are rendered as a vector logo or sit inside a colored ribbon, the text extractor often skips them. Anything you need the recruiter to find must be selectable text. A useful test: try to highlight the text with your cursor in a PDF reader. If you cannot highlight it, the ATS cannot read it either.
• Custom fonts and ligatures. Stick to safe families — Inter, Helvetica, Arial, Calibri, Times. Exotic display fonts can render perfectly visually but get mangled by extraction. The letters fi become a single ligature glyph that some parsers drop entirely. The letters tt and ff have the same problem in many display fonts.
• Headers in unconventional positions. Your name, phone, and email need to be in the first ten percent of the document. ATSes parse the top of page one as the contact block and ignore later occurrences. A name pushed below a banner image or buried in a sidebar is a name that the system never associates with the record.
• Tables, especially for skills grids. Tables are a layout primitive in PDF, and table extraction in 2026 is still unreliable. A 3x4 skills table will usually parse, but you are betting against the worst case. Use simple comma-separated lists instead. A line that reads JavaScript, TypeScript, Python, Rust will parse perfectly in every ATS we tested.

Things that look bad but actually parse fine

Some choices look risky but turn out not to matter. Color is fine — the ATS ignores it. Reasonable use of bold and italic survives every parser we tested. Bullet points are universally supported, as are unicode dashes and em-dashes. A horizontal divider rule between sections is fine. A monochrome logo or photo in the corner is fine, as long as your contact info appears as selectable text elsewhere.

The general principle: anything that is rendered as text in the PDF will be read. Anything that is rendered as graphics will be ignored. Whitespace, color, and font weight do not change that classification.

What about Word documents?

Modern ATSes parse both Word and PDF well, with one important caveat. If you exported a PDF from a tool that flattened everything to images — some online resume builders do this so you cannot edit the result — parse rate drops sharply because the only way back to text is OCR. A clean PDF exported from Word, Google Docs, or our builder will parse on par with the original .docx.

If a job application gives you the choice, PDF is generally safer because the formatting is locked. If you submit a .docx, the recruiter's word processor might re-render it with different fonts than you used, and that can shift line breaks in unfortunate ways. But the structured record the ATS extracts will be the same either way.

How to test before you submit

Beyond the copy-paste sanity check, the fastest signal you can get is to score your resume against the role you are applying for. Our free ats-score tool runs the same extraction logic an ATS would, scores your parse cleanliness, and flags the structural issues it finds. It is free, no signup, no upload — the file never leaves your browser.

The other useful exercise: pick the most ATS-aggressive job posting you are targeting (usually a big tech company or a Workday-using enterprise) and submit a test application to it. Most companies will at least send you an automated email acknowledgement; if that email gets your name right and references the role correctly, your parse went through. If it addresses you as null or skips fields you filled in by hand on the portal, your PDF parsed badly and you have work to do.

A practical pre-submission checklist

• Open the resume in your PDF reader, select all, copy, paste into a plain text editor. Read the result top to bottom. If you cannot follow the structure, the ATS cannot either.
• Confirm your name, phone, and email appear as selectable text within the first inch of page one.
• Confirm every section uses one of the boring header names: Experience, Education, Skills, Projects, Certifications.
• Confirm no critical information lives inside an image, icon, or sidebar that breaks reading order.
• Confirm the file is under 2MB. Some ATS upload widgets silently truncate or reject larger files.
• Score the resume on our ats-score page to catch anything you missed.

Resume formatting is one of those areas where the boring answer is almost always the correct one. The recruiter's eye and the ATS's regex want the same thing — clear sections, clean text, an unambiguous reading order. Optimize for that and the rest takes care of itself.