Supported files
PDF and DOCX. Not pages exported as images, not ZIP archives, not links to a document in cloud storage.
The common causes
It is a scan. A PDF that is a photograph of a printed page has no text in it. Optical character recognition is attempted, but a slightly skewed phone photo of a printout is genuinely hard. Export from the original application rather than scanning if you possibly can.
It is password protected. Remove the password and upload again. Nothing can read an encrypted file.
It is not really a PDF. A file renamed to .pdf is still whatever it was. Re-export it
properly.
It is very large. Files over the limit are refused with the limit named. A résumé over a few megabytes is usually carrying uncompressed images.
Parsed, but wrong
Extraction is good and not perfect. The places it most often needs a correction:
- Multi-column layouts, where reading order is ambiguous.
- Dates in unusual formats.
- Roles with no employer named — freelance and contract work especially.
- Headings the parser has not seen, so a whole section is filed oddly.
Always read the extracted result. It is a starting point that saves you retyping, not a finished record.
The error just says it failed
It should not — the message names the cause. If you genuinely get a bare failure, that is worth a ticket with the file attached, because a message that does not say why is itself the bug.