Automated CV Screening — How It Actually Works and How to Handle It
What happens to your CV before a human ever sees it, which designs break the parser, and how to write a resume both a machine and a recruiter read cleanly.
Almost every CV you send passes through a machine before a human eye meets it. It’s not a conspiracy and it’s nothing to be afraid of — it’s just how the industry works now. The question isn’t how to "trick" the system, it’s how to write a document the machine understands and a person wants to read. Let’s break it down plainly.
What's actually happening behind the scenes
An ATS (Applicant Tracking System) or resume parser takes your file and tries to split it into fields: name, contact details, work experience, education, skills. It reads text, recognizes familiar headings ("Experience", "Education"), and ties dates and companies to roles. More recently, AI-based models have joined the party — they don’t just pull keywords, they try to understand context: that react and React.js are the same thing, and that someone who wrote "led a team of five" has management experience.
The key point: parsers work best on clean, well-structured text. The more graphically "clever" your document is, the higher the chance something gets lost in translation.
What actually breaks the machine
These are the things that choke parsers, from experience: two-column layouts (the parser reads across and scrambles the content), text baked into an image or logo (as far as the machine is concerned, it doesn’t exist), a CV saved as a scan or screenshot instead of a text-based PDF, nested tables, and exotic fonts that don’t render properly. Contact details tucked into a header or footer tend to get swallowed too. The fancier the template, the more there is to go wrong.
Myths worth throwing out
Myth one: "write keywords in white on a white background and you’ll sail through." It’s an old trick, modern systems flag it, and any recruiter who spots it will simply reject you. Myth two: "the robot auto-rejects me." In most cases the system ranks and organizes — it doesn’t hit "decline." Myth three: "the more keywords, the better." A word salad with no context looks fake to both the smart machine and the human. You lose credibility, you don’t gain points.
How to write a CV both a machine and a human love
Keep a simple, reverse-chronological structure with standard headings the parser recognizes. Use one column, not two. Always save as a text-based PDF (one you can select and copy text from), never as an image. Spell out technologies and tools by their full name — if you know Python, write Python; don’t rely on the system to infer it from "backend development."
And most important: tailor the content to the role. Not by faking experience, but by choosing which projects and skills to highlight from what you actually did, based on what the job asks for. Honest, natural language is exactly what both the machine and the recruiter connect with.
How it works at esy
At esy we run our own system that helps us surface and rank candidate-to-role fit, so good people don’t get lost in the pile. But the system doesn’t make the call — it moves candidates toward a human recruiter who looks at the whole picture. If you want to see how it works from your side, you can run our match tool or just send us your CV.
Bottom line: don’t try to outsmart the system — just try to be clear. A clean, structured, honest CV passes the machine and convinces the human.
Got a question about your CV? Talk to us — we’re happy to help you put your best foot forward.