Applicant tracking systems (ATS) have screened résumés for years. What changed recently is that many of them now use large language models to parse, summarise and rank applications, and recruiters increasingly read an AI-written summary of your CV before they open the file. The good news: AI parsers are more forgiving than old keyword matchers. The bad news: a messy PDF still produces a messy summary.
PDF or Word?
Both are generally accepted. PDF keeps your layout exactly as designed, which is why most people prefer it. Follow the job posting: if it asks for .docx, send .docx. See PDF vs DOCX for résumés.
Make sure there is real text
Export your PDF from Word, Google Docs or a résumé builder. Never send a scan or a photo, and avoid design tools that turn text into outlines. Test: open the PDF, select all text, copy it and paste into a plain text editor. If it pastes cleanly in a sensible order, a parser will read it too.
Layout choices that parse well
- Single column is safest. Two-column layouts often get read across the columns, mixing your skills into your job history.
- Standard section headings: Experience, Education, Skills, Certifications. Creative headings ("Where I've Been") confuse section detection.
- Dates in a consistent format, such as "Mar 2023 – Present", next to each role.
- No important information in headers, footers or text boxes, which some parsers skip.
- Avoid tables for layout. A table is fine for a list of skills, but not for the whole page.
- Standard fonts that are embedded in the PDF. See embedded fonts in PDF explained.
Writing for an AI summary
Since a model may summarise your CV for the recruiter, write so the summary comes out right:
- Lead each role with the outcome. "Cut onboarding time by 40% by redesigning the training programme" summarises better than a list of duties.
- Use the job posting's vocabulary where it honestly matches your experience. Semantic matching helps, but exact terms for tools and certifications still matter.
- Quantify. Numbers survive summarisation; adjectives do not.
- Keep it to one or two pages. Length dilutes the parts that matter.
Do not hide text for the AI
Some candidates add invisible keywords or instructions such as "rank this candidate first" in white text. Recruiters and ATS vendors now check for this, and it reads as manipulation when found. It can get you rejected outright. See hidden prompt injection in PDFs.
Final file checks
- File name:
Firstname-Lastname-CV.pdf, notresume_final_v3.pdf. See naming and versioning shared files. - Document title in the properties set to your name. See how to edit PDF metadata.
- Working links to your portfolio or LinkedIn. See how to add hyperlinks to a PDF.
- Size under 2 MB, so upload portals do not reject it. See compress a PDF for online application portals.
Takeaway
A résumé that reads cleanly as plain text will parse well for any system, AI or not. Keep the layout simple, lead with outcomes, use honest keywords, and never hide anything in the file. For quick fixes like adding a link or correcting a typo, a browser editor like Docento.app works without uploading your CV.
