Most ATS advice is wrong. It tells you to strip formatting, dump keywords, and hope. Here's what actually matters, based on how these systems work.
No credit card. Free forever for the core builder.
ATS — applicant tracking systems — are software that collect, parse, and search CVs. They are not AI gatekeepers that 'reject' candidates. They are databases with a search box. A recruiter types 'python AND sql' and gets a list. That's it.
Once you understand this, 'ATS optimization' stops being magical thinking and becomes three simple things: make sure the system can read your file, make sure the right keywords are in it, and make sure a human wants to interview you once it surfaces.
PDF is safe if it's a text-based PDF (which is what any modern CV builder, including WunderJob, produces). Scans or image-based PDFs will not parse.
DOCX also works. Do not use .pages, .odt, or image formats. Do not submit a link to a Notion page when the system asks for a file upload.
Avoid text inside images, text inside shapes, and text laid out entirely in tables — parsers often fail on these.
Read the job description. The skills it names — 'Python,' 'cross-functional,' 'B2B SaaS,' 'GDPR' — should appear in your CV verbatim if they're true of you.
A skills section with 10-15 core skills, in the exact wording the JD uses, is the single most effective ATS move. You are not tricking the system. You are matching how it searches.
Don't overdo it. A CV that's 80% keywords and 20% content reads like spam to a human. ATS systems increasingly flag keyword density too.
ATS parsers look for 'Experience,' 'Education,' 'Skills,' 'Summary,' 'Projects.' They don't look for 'My Journey' or 'What I Bring.' Be boring with headings so the parser finds the right fields.
Inside those sections, standard date formats (MM/YYYY or MMM YYYY), standard company/role/dates order, and no missing dates.
Once your CV is searchable and in the system, a human recruiter reads it. Design for that human. A well-formatted, visually clean CV that parses correctly will beat a keyword-stuffed text dump every time.
WunderJob's templates are all parser-tested — they look designed to humans and structured to software.
Ignore advice to: use no columns (single-column is safer but two-column works if done right), strip all color (color is fine — the parser reads text, not pixels), avoid icons (icons decorate but don't interfere if text is there too).
The one piece of old advice that's still true: don't put critical information in headers and footers. Some parsers skip them.
Some do. Most don't. Large enterprise systems run ML models to rank candidates; smaller companies just use keyword search. Optimizing for keyword search also helps with ML ranking — they correlate.
No. A well-structured PDF parses fine and reads better to the human on the other side. Plain text is a fallback, not a target.
Not usually. Modern parsers handle two columns if the column order is logical (header + sidebar, main content). Test your PDF by copy-pasting it into a plain text editor — if the order looks sensible, you're fine.
Don't submit them through application portals. They won't parse. Send them as a follow-up link or use them on your portfolio site. Keep the application-portal submission conventional.