A document uploaded but will not process

What to do

1. Open the document. If the page says it has no extractable text, the file is an image rather than text: a scan, a photograph, or a print-to-image PDF. It goes to text recognition rather than failing.
2. For a long scan, give it time and check back. Text recognition runs to a per-document page budget and continues in later passes, so a several-hundred-page scan finishes over several passes rather than in one. Partial text is real text; it is the pages read so far, not a corrupted result.
3. Check the file opens on your own machine without a password. A file the platform cannot open produces the same empty result as one with nothing in it.
4. If the document page says the file is held after an upload check, that is quarantine, not deletion. The bytes are intact and an attorney can review and release it from that card. Never delete it to retry: a held production may itself be evidence and may be under a litigation hold.
5. If none of the above fits, report it with the matter and the file name. A document that will not process is worth a ticket, because the alternative is a filing built on a record with a hole in it.

What you should see

You can distinguish a pending or partial text read from an unreadable or held file and report the remaining problem with its status.

Limits and unresolved problems

Text recognition reads a bounded number of pages per document per pass, so a very long scan is not instant and the count you see part way through is genuinely partial. Separately, upload malware scanning is not armed in every deployment; where it is not, the document page says so per document rather than implying a check that did not happen.

Product scope and requirements

• Matter intelligence

Related guides

• Review a document held after an upload check
• Inspect a fact, its sources and its recorded status