Archive the Source Package, Not Just the PDF

Choose a complete package

A final PDF alone cannot explain how a book was made. Archive manuscript source, layout source, final print PDF, EPUB source and delivered file, cover source and exports, images and licenses, fonts or font records, metadata brief, identifiers, proof notes, and rights and contributor agreements.

For a hypothetical revised edition, make a new package rather than overwriting the first. Add a machine-readable manifest with file name, version, date, checksum where practical, and a human note explaining the relationship.

Keep provenance with files

Store source URL, retrieval date, license or permission reference, and the decision that made each third-party item usable. A caption copied into a layout file is not a provenance record. Keep private personal data and confidential agreements under suitable access controls.

Practice recovery

Once, try to rebuild a product record from the archive without opening your inbox. Missing password ownership, unlabelled fonts, or absent source files will become visible while correction is still possible.

Specific production checks

Use a readme at the package root with edition name, publication status, designated source file, export commands or vendor settings, and who can answer access questions. A future editor should not need to inspect 40 folders to discover which EPUB was actually delivered.

Test restore on one small object: retrieve the cover source, license record, final export, and product brief, then recreate a retailer-ready package in a separate folder. Do not delete the original; the test is evidence that the archive can support correction, not a reason to overwrite history.