Uploading a job description extracts it into structured fields rather than storing a blob you cannot search.
What it extracts
The role title, seniority, location and working pattern; the responsibilities; and — the part that matters most — the skills, each marked required or preferred, and each resolved against a shared skills vocabulary so that "JS" and "JavaScript" are one skill rather than two.
Supported files
PDF and DOCX. A PDF that is a scan of a printed page rather than text is handled by optical character recognition, which is slower and less accurate; a text PDF is always better if you have the choice.
What to check afterwards
Extraction is good and is not perfect. The three places to look:
- Skills. Did it find them all, and are any of them wrong? A skill invented from a passing mention will be scored against every candidate.
- Required vs preferred. Prose does not always make this explicit, so the parser has to infer it.
- Experience level. "5+ years" is easy; "significant experience" is not.
If parsing fails
The error says why rather than "failed" — a password-protected file, an unreadable scan, a file that is not really a PDF. See document upload problems.