How ATS Systems Actually Work (And Reject Resumes)

Most Applicant Tracking Systems — Workday, Taleo, Greenhouse, Lever, Naukri FastForward — are not evaluating your career story. They are running a four-step mechanical pipeline: extract text, map it into fields, match keywords against the job description, then apply recruiter-set filters. None of those four steps involve judgment about whether you are a strong candidate.

Before any keyword matching happens, the system extracts raw text from your PDF or DOCX. A single-column, standard-font resume converts cleanly. A resume built in Canva with a two-column layout, icons, or a skills graphic often converts into scrambled or missing text — sometimes your phone number ends up in the middle of your job title.

This step has nothing to do with your qualifications. It is purely a file-format problem, and it is fixable in minutes by switching to a single-column, text-based layout.

The parser tries to identify which chunk of text is your name, which is your most recent job title, which is your skills list — based on recognizable section headers. "Work Experience," "Education," and "Technical Skills" map cleanly. Creative headers like "My Journey" or "What I Bring" map to nothing, and that entire section can effectively disappear from the system's view of your resume.

This is the step people usually mean when they say "ATS score." The system compares the token set from your resume — hard skills, tools, certifications, job titles — against the token set from the job description, and produces an overlap percentage.

This step is mostly literal. Writing "Postgres" when the JD says "PostgreSQL," or "team management" when the JD says "stakeholder management," can reduce your match even though the meaning is close to identical to a human reader.

This is the step almost nobody talks about, and it explains most of the "I am clearly qualified but got rejected in 2 minutes" experiences. Recruiters configure hard filters on the requisition — years of experience range, location, notice period, education cutoff, visa status — and any resume that fails one of these filters gets auto-rejected before a human ever sees the keyword match score.

A perfectly tailored resume can still get filtered out instantly if your listed notice period or location does not match what the recruiter configured. This is not a resume problem, and no amount of keyword tailoring fixes it — it is a fit problem on that specific requisition.

Workday, Taleo, Greenhouse, and Lever are all built on the same four-step concept, but they differ in strictness and transparency. Workday, for example, often shows candidates a parsed preview of their resume fields — name, experience, education — right at the point of application, which is one of the few moments you can actually see how a real parser interpreted your file before you hit submit.

Greenhouse and Lever tend to be more forgiving of moderate formatting variance but still penalize multi-column layouts the same way. The practical takeaway is the same regardless of which system a company uses: single-column, standard headers, and exact keyword matches are safe defaults across all of them.

You do not need a tool to catch the most damaging failure mode.

It is almost always one of two things: a parsing failure from Step 1 that made a strong resume unreadable, or a hard filter from Step 4 that had nothing to do with resume quality at all. Very rarely is it a case of the system "not understanding" a candidate's value — the pipeline simply is not built to evaluate that.

Run your resume through Career Copilot's ATS score checker against a real job description to see exactly where in this four-step pipeline your resume is losing points — parse structure, keyword gap, or both.