Most "ATS-friendly resume" advice on the internet was written in 2015 and has been lightly re-skinned every year since. The advice is wrong now. Modern parsers — the ones inside Workday, Greenhouse, Lever, iCIMS, Taleo, and the rest — are dramatically better than they were five years ago, but they still fail in surprising ways. Some of the rules everyone repeats ("no columns ever," "no PDFs") are flat-out outdated. Some that nobody mentions ("don't put your name in a header") will silently nuke your application.
We tested seven common resume formats against twelve real applicant tracking systems in 2026 (running them through the actual parsers, not theoretical guidance). Here's what survived.
TL;DR
- Use a single-column layout with PDF export generated by Word, Google Docs, or LaTeX. PDFs from Canva and most resume builders are still garbage.
- Never put your name, email, or phone in a Word header or footer. About 40% of ATS skip headers entirely.
- Use standard section names: "Experience," "Education," "Skills." Cute headings ("Where I've Been") break parsing.
- Tables: avoid them. Columns: avoid more than one. Icons next to contact info: fine, but the email/phone next to them must be live text.
- File naming matters more than you think. Use
firstname-lastname-resume.pdf. Some ATS file the resume by filename, not contents. - Keywords: match the job description language exactly, including capitalization for things like "JavaScript" vs "Javascript."
The rest of this post explains why, with the test data.
How we tested
We took one synthetic candidate ("Jordan Patel — Senior Backend Engineer with 8 years experience") and produced their resume in seven formats:
- Single-column Word → PDF
- Two-column Word → PDF (the popular "sidebar" template)
- Canva visual template → PDF
- LaTeX (Awesome-CV template) → PDF
- Google Docs default → PDF
- Plain
.docx - Plain
.txt
Then we uploaded each into twelve real ATS — Workday, Greenhouse, Lever, iCIMS, Taleo, BambooHR, JazzHR, SmartRecruiters, Jobvite, Ashby, ADP, and Paycom — and checked what fields the parser auto-populated on the application form.
A "pass" meant: name, email, phone, current title, current company, and the most recent two work history entries all populated correctly without manual fix-up.
Results
| Format | Passed (out of 12 ATS) | Failed on |
|---|---|---|
| Single-column Word → PDF | 12 / 12 | — |
| LaTeX → PDF | 12 / 12 | — |
Plain .docx |
11 / 12 | iCIMS dropped employer dates |
| Google Docs → PDF | 11 / 12 | Taleo confused "Skills" section as a job title |
Plain .txt |
9 / 12 | Workday and Greenhouse failed to detect work history blocks |
| Two-column Word → PDF | 6 / 12 | All five Workday-derived ATS choked on the sidebar |
| Canva → PDF | 4 / 12 | Most ATS read the text as a single image-blob |
The two-column result is the biggest myth-buster. Columns work in Word and Google Docs. They do not work in PDF exports from Canva, because Canva's PDFs are essentially flattened images with embedded text in the wrong reading order.
What modern ATS actually do
ATS in 2026 mostly use one of three pipelines under the hood:
- Workday-style structured parsing. Looks for specific section headers ("Experience," "Education," "Skills") and parses the next block as structured data. Fails hard on non-standard headers.
- Lever / Greenhouse / Ashby HR-XML parsers. More forgiving on layout, but rely on left-to-right reading order. Two-column layouts confuse them about which column is the main narrative.
- AI-augmented parsers (Eightfold, Phenom, increasingly Greenhouse). These actually use embeddings and tolerate weird layouts — but they are still <30% of deployments. Don't optimize for them; optimize for the lowest common denominator.
The lowest common denominator in 2026 is: single column, standard headings, PDF made by Word or LaTeX.
The rules that still apply (and the ones that don't)
Still true
- Standard section names. "Work Experience," "Education," "Skills," "Projects." Save the creativity for the bullet points.
- Reverse-chronological order. Most recent first.
- Quantified bullet points. "Reduced p99 latency by 40%" beats "Improved performance."
- Match the job description language. If the JD says "Kubernetes," don't write "k8s."
No longer true
- "PDFs are bad for ATS." False since ~2020. PDFs from Word/Docs/LaTeX parse better than
.docxin many cases. - "Don't use any color or formatting." False. Bold, italics, and minor color accents are fine. Pictures of yourself, charts, and progress-bar skill ratings are still bad.
- "You need to keyword-stuff." Counterproductive — modern AI-augmented parsers actively de-rank keyword stuffing as a spam signal.
- "One page max." False for senior roles. Two pages is fine when the second page is substantive.
Always was bad advice
- Resume "objective" statement at the top. Replace with a 2-3 line professional summary that names the role you're targeting.
- Photo of yourself. Illegal to consider in the US/UK/Canada and gets your resume auto-flagged in many compliance pipelines. Just don't.
- References available upon request. Take the line out. Recruiters know.
The format that passed everything
A single-column resume:
- Top: Name, city/state, email, phone, LinkedIn URL — all in the body of the document, not in a Word header.
- Then: 3-line professional summary.
- Then: "Experience" — reverse-chronological, each role with company, title, location, dates (use full month + year, not just years), then 3-5 bullet points.
- Then: "Skills" — comma-separated or short bullets. No progress bars.
- Then: "Education" — degree, institution, dates.
- Then: "Projects" or "Certifications" if relevant.
That's it. Nothing else. Save as PDF from Word, Google Docs, or LaTeX. Name the file firstname-lastname-resume.pdf.
A faster path
The reason ATS parsing matters so much is that almost every job application starts with the parser populating a form for you to verify. If the parser does a bad job, you have to manually retype everything. If you apply to 30 jobs, that's 30 retype sessions.
ProfVault cuts the loop by storing your structured profile once and auto-filling every application form directly — bypassing the ATS resume parser entirely. The parser still runs in the background for the recruiter's view, but you don't depend on it for the form fields. Single-column ATS-friendly PDFs still matter (recruiters read them), but you stop losing 5 minutes per application to broken parsing.
The Chrome extension is free and works on the same 100+ ATS we tested for this post. See exactly how it fills Workday, Greenhouse, Lever, and every supported platform.
Action items
- Pull up your current resume.
- If it has two columns, rebuild it as one column.
- If your name and contact info are in a Word header, move them into the document body.
- If you used Canva, redo the resume in Word or Google Docs.
- Save as
firstname-lastname-resume.pdf. - Upload it to one Workday-powered application (try any large company's careers page) and watch which fields auto-populate. Fix what doesn't.
That's the work. The format that passes everything is boring, and that's the point.