When a PDF does not parse

Updated July 11, 2026

Occasionally a dropped PDF won't extract cleanly. Here's exactly what happens, why, and what to do about it.

#The exact message

If ProClose can't pull enough real fields out of a dropped file, no insured name, no premium lines, it shows this toast:

"Could not deterministically read filename. The parser reads text-layer PDFs. Showing the sample flow instead."

#Why it happens

The most common cause by far is a text layer that doesn't exist. ProClose's parser, see How ProClose reads a PDF, reads the text embedded in a PDF. That works great for PDFs generated straight out of a carrier's quoting system, where every character on the page is real, selectable text.

A scanned document is different. If a policy or quote was scanned as an image, printed then scanned back in, faxed, photographed, there's no embedded text at all, just a picture of a page. There's nothing for the parser to read, no matter how clean the scan looks to the human eye. This is the single biggest reason a PDF fails to parse: text-layer PDFs work, image-only scans do not, at least not yet.

A few less common causes can also trigger the same message: a corrupted or unusually large file, or a carrier layout so unfamiliar that too few of the expected fields, insured name and at least one premium line, come through to clear the minimum bar.

#What the app does

ProClose doesn't leave you at a blank screen or an error page. On a parse failure, it automatically falls back to the sample proposal flow, so you can see a complete, working deck immediately while you sort out the source document. The event is also logged to the audit trail so there's a record that this particular file didn't parse.

#What to try instead

  • Get the original from the carrier portal, not a scan. If you have access to the carrier's system, download the quote or policy directly rather than working from a copy that was printed and rescanned. The carrier-issued original almost always has a real text layer.
  • Ask whoever sent it for a native PDF. If a quote arrived by email as a scanned attachment, the sender likely has (or can regenerate) the original, unscanned version.
  • Try a smaller or cleaner file if size or corruption might be the issue. An oversized or damaged PDF can fail for reasons unrelated to the scan-vs-text-layer distinction.
  • Use the sample flow to keep moving. While you track down a better source file, the sample deck that loads automatically shows you the full shape of a finished proposal, useful for getting oriented or demonstrating the app while the real document gets sorted.

Once you have a text-layer version of the document, drop it back in the same way, on the New proposal page, and the parser reads it normally.