The metadata most tools leave behind

A PDF carries more about its own history than most people expect: who created it, what software was used, the title of the document it was before you renamed it, and sometimes the full path of the folder it was saved in.

Stripping that before sending a file out is sensible practice. The problem is that "stripping it" usually means stripping one of the two places it lives.

Two stores, one document

The document information dictionary is the old mechanism. Title, author, subject, keywords, creator, producer, creation and modification dates. It is what a reader shows in its properties panel, and it is what almost every "remove metadata" feature clears.

The XMP packet is the newer one: a block of structured data attached to the document, capable of holding far more. Design and publishing tools write to it extensively. It frequently duplicates the basic fields — and often keeps older values that the basic fields have since had overwritten.

Clear the first and leave the second, and a document that reports no author in its properties panel may still name one to anything that reads the packet directly. That includes plenty of document management systems and search indexers.

What neither store touches

Both are separate from the page content, and clearing them does nothing to what is printed on the page.

• A name in a letterhead.
• A signature block.
• A footer with a file path in it, added by whoever printed the original.
• Track-changes remnants and comments, which live in annotations, not metadata.
• Attachments embedded in the document.

This matters because "I removed the metadata" can feel like a completed privacy step when it has addressed only part of the problem. The most identifying thing in most documents is visible on the page.

Dates are metadata too

Creation and modification timestamps are often the most revealing fields present. A document dated three weeks before you claimed to have written it, or modified at two in the morning, tells a story regardless of what the author field says.

Some workflows need the dates preserved for legitimate reasons — archival, chain of custody — so removing them should be a choice rather than a side effect.

A workable habit

1. Inspect first. Look at what is actually there before deciding what to remove. Many documents carry nothing interesting and need no thought. 2. Clear both stores, and confirm the output rather than trusting the operation. 3. Read the pages for anything identifying that metadata removal will never reach. 4. Do it last. Almost every other operation rewrites the document, and some tools will happily stamp their own name into the producer field on the way through — so a document cleaned early can arrive dirty.

That last point is worth dwelling on. A tool that rewrites metadata on every save, quietly, is easy to build by accident and hard to notice. If clearing the author field does not stick after a later operation, that is what is happening.