Ask around about applicant tracking systems and you’ll hear a lot of folklore: hidden white text stuffed with keywords, invisible tables, a magic keyword density that “beats the algorithm.” Most of it is wrong, some of it will get your application binned, and none of it is necessary.
Let me walk through what an ATS actually does — and how to be reliably found by one without pretending to be someone you’re not.
What an ATS is (and isn’t)
An applicant tracking system is, first and foremost, a database with a workflow on top. When you apply, it parses your CV into fields — name, contact, work history, skills — and stores it so recruiters can search, filter, and move candidates through stages.
Two myths worth killing right away:
- It is not an AI gatekeeper that auto-rejects you. For the vast majority of roles, a human decides. The ATS just makes candidates findable and organised.
- There is no secret keyword score to hit. Recruiters search it like you’d search anything else. If your CV contains the terms they search for, in a form the parser understood, you show up.
So the real goal isn’t to trick the parser. It’s to make sure the parser reads your CV correctly and that the true, relevant terms are actually present.
Where CVs actually fail the parser
Most ATS problems are formatting problems, not content problems. The parser chokes and your beautifully designed CV turns into scrambled fields. Common culprits:
- Multi-column layouts and tables used for structure. The parser reads left-to-right, top-to-bottom, and a two-column design can interleave your job titles with unrelated text.
- Text baked into images or icons. If your skills live inside a graphic, the parser sees nothing.
- Headers and footers for critical info. Some parsers skip them entirely.
- Exotic fonts or heavy styling tricks that don’t map to clean text.
The fix isn’t a trick — it’s a clean, single-column, genuinely parseable structure. Boring to a designer, perfect to a parser.
The keywords that matter are the true ones
Here’s the honest version of “keyword optimisation”: read the job description, notice the terms it uses for skills and responsibilities, and — where they’re genuinely true of you — make sure your CV uses those same terms.
If the job says “CI/CD pipelines” and you’ve built them, say “CI/CD pipelines,” not just “automated our deployments.” Same fact, matched vocabulary. That’s not gaming the system; that’s speaking the recruiter’s language about work you actually did.
What you must not do is add terms for skills you don’t have. It gets you found for roles you can’t do, wastes everyone’s time, and collapses the moment someone asks a follow-up question. Being findable for the wrong job is worse than not being found.
This is exactly what SiviGen automates — honestly
Matching your real experience to a job’s vocabulary, in a clean parseable format, is tedious to do by hand for every application. It’s also precisely what SiviGen is built to do.
When you paste a job description, SiviGen reads its requirements and aligns them to the facts your CV already supports — reaching for the job’s own terms where they’re true of you, and leaving them out where they aren’t. Then it exports a clean, single-column, ATS-friendly PDF: no nested tables, no hidden text, no tricks that get applications rejected.
And because every tailored line is audited against your fact graph, you never end up “optimised” into a claim you can’t back up. You get found for the right roles, with the right words, for the right reasons.
The honest checklist
If you take nothing else from this, here’s the short version:
- Use a clean, single-column layout. Let the parser read you correctly.
- Keep real text as text — never bury skills or titles in images.
- Mirror the job’s vocabulary for skills you truly have. Match the words, not by inventing the experience.
- Cut the tricks. White-text keyword stuffing is a fast way to get binned.
- Make sure the terms are true. Being found for a job you can’t do helps no one.
You don’t beat the ATS by outsmarting it. You beat it by being clearly, accurately, findably you — which, conveniently, is also how you do well once a human finally reads the thing.
Want that done for you, per application, without the fabrication? Try SiviGen.