## TL;DR
ocrmypdf's 'PDF/A conversion failed' on a signed contract is the converter refusing to silently break your signatures: PDF/A conversion rewrites the document, which invalidates embedded digital signatures and drops signature annotations. Use --output-type pdf instead of the default PDF/A output to keep it a regular PDF with the OCR layer and intact signatures. If you genuinely need PDF/A, OCR an unsigned copy first and apply signatures afterward - never convert a signed original.

## The query
```text
ocrmypdf 'PDF/A conversion failed' on a signed contract with embedded signatures
```

## Use this when
- ocrmypdf fails with 'PDF/A conversion failed' on a signed contract
- The PDF has visible signatures or embedded signature fields
- Unsigned PDFs convert fine with the same command

## Not for
- ocrmypdf failures on unsigned PDFs - those are different errors (OCR engine, permissions, corrupt input)
- Signature validation failures unrelated to OCR - check the certificate chain instead
- PDF/A validation failures after a successful conversion - a compliance tuning problem

## Steps
### 1. Confirm the document carries embedded signatures
Check for signature fields or embedded certs before changing anything. If the contract is signed, the failure is protective, not a bug - the tool is refusing to destroy evidentiary value.
```bash
pdfinfo signed-contract.pdf | grep -i -E "sign|encrypt" || echo "check fields manually"
```
Expected output: evidence of signatures (or inspect the signature panel in a reader to be sure).

### 2. Switch the output type to plain PDF
Add --output-type pdf so ocrmypdf adds the OCR text layer without attempting the PDF/A rewrite that breaks signatures. Everything else in your command stays the same.
```bash
ocrmypdf --output-type pdf signed-contract.pdf signed-ocr.pdf
```
Expected output: the command completes and signed-ocr.pdf has a searchable text layer.

### 3. Verify the signatures still validate
Open the OCR'd file in a reader and check the signature panel: signatures should still show as valid. Also verify the OCR layer is actually there by selecting and copying text.
Expected output: signatures valid, text selectable - the document gained searchability and lost nothing.

### 4. For PDF/A needs, reorder the pipeline: OCR first, sign last
If the business requires PDF/A, run ocrmypdf with PDF/A output on the unsigned original, then apply the digital signatures to the converted file. The signed artifact is then both PDF/A and validly signed.
Expected output: a PDF/A file whose signatures validate because they were applied after conversion.

## Tool compatibility
ocrmypdf 13+ (--output-type option); any PDF reader with a signature panel for verification. The PDF/A-vs-signature conflict is inherent to the standards, not version-specific.

## Variant phrasings
- ocrmypdf PDF/A conversion failed signed contract
- ocrmypdf embedded signatures PDF/A error
- ocrmypdf --output-type pdf signed PDF
- ocrmypdf breaks digital signatures PDF/A

## Why it happens
PDF/A is an archival standard that forbids or restricts several PDF features, and its conversion rewrites objects throughout the file. A digital signature is a cryptographic hash over the document's bytes as signed - any rewrite invalidates it by design. ocrmypdf fails the conversion rather than producing a file that looks signed but no longer verifies, because a silently broken signature is worse than no OCR.

## Edge cases / pitfalls
- Some 'signatures' are just scanned ink images with no cryptographic value - those survive conversion fine, but verify which kind you have first.
- Certified PDFs (with usage restrictions) may block even the OCR rewrite; those need the certifier's permission, not a flag.
- Keep the signed original archived untouched; the OCR'd copy is a working derivative.
- Approval stamps and annotations can also be dropped by PDF/A conversion - check those too, not just signatures.

## Provenance

Resolved from the public thread: https://vectle.com/posts/pst_-_wDBPH-BhWsedA0EPvUUQ