Privacy Basics

Deep Clean vs. Standard Metadata Removal: What's Actually Different?

Standard metadata stripping only clears the visible fields.

2026-08-16 · 7 min read

Quick answer: Standard "strip" mode clears the visible metadata fields — title, author, subject, keywords — from a file's Info dictionary or EXIF markers. Deep scrub goes further: for PDFs specifically, it walks every object in the file, deletes embedded XMP metadata streams (which standard stripping leaves behind), clears metadata keys scattered across individual PDF objects rather than just the top-level Info dictionary, and rebuilds the file's internal object structure from scratch. If you're handling a file where "mostly clean" isn't good enough — legal filings, leak-sensitive documents, anything you'd hate to be wrong about — deep scrub is the mode to use.

Most "remove PDF metadata" tools, including earlier versions of this one, only clear what's called the Info dictionary — a small, standard set of fields (Title, Author, Subject, Keywords, Creator, Producer) that every PDF viewer knows how to read. That's the metadata you see when you check "Document Properties" in Adobe Reader. It's also only part of the story.

Where metadata actually hides in a PDF

A modern PDF can carry metadata in at least three separate places:

• The Info dictionary — the classic fields, easy to find and easy to clear. Every basic "strip metadata" tool handles this.
• Embedded XMP streams — a separate, XML-based metadata format that Adobe introduced to carry richer information (including, sometimes, full edit history, GPS data for scanned documents, or software fingerprints). XMP data can duplicate — or contradict — what's in the Info dictionary, and many free tools never touch it.
• Object-level metadata — some PDFs, especially ones exported from enterprise document management systems, embed metadata keys like PieceInfo or custom producer tags on individual objects deep inside the file, not just at the document level.

Clearing only the Info dictionary and calling a PDF "cleaned" is a bit like wiping a whiteboard but leaving sticky notes stuck to the back of it. Technically less is visible. It's not actually clean.

What deep scrub does differently

When you select the deep-scrub option in MetaClean Pro (available for PDFs), the cleaning engine:

1. Clears the standard Info dictionary fields, same as strip mode.
2. Walks every indirect object in the PDF looking for embedded /Metadata streams (the XMP data) and deletes them outright, rather than leaving them in place.
3. Clears object-level metadata keys — PieceInfo, ModDate, ProducerTrails, and similar — from individual objects throughout the document, not just the top-level document dictionary.
4. Rebuilds the object streams and cross-reference table when saving, rather than patching the original file structure in place.
5. Re-scans the output and reports exactly what it found — you'll see a "residual tags" count if anything survived, and a confirmation if nothing did.

That last point matters: a cleaning tool that doesn't verify its own output is asking you to trust it blindly. Deep scrub tells you what it actually removed, not just what it attempted to remove.

Do you need deep scrub every time?

Not necessarily. For everyday use — cleaning a resume before emailing it, stripping author info from an internal draft — standard strip mode is fast and sufficient; most casual recipients never look past the Info dictionary. Reach for deep scrub when:

• The document is going into legal discovery or a court filing, where opposing counsel may specifically dig for metadata.
• You're a journalist or whistleblower submitting a document where the source's identity absolutely cannot leak through embedded XMP data.
• The PDF came from an enterprise system (many law firms, government agencies, and large companies export PDFs with heavier embedded metadata than a simple "Print to PDF" would produce).
• You're simply not sure what's in the file, and the cost of being wrong is high.

The XMP specification is public, and so is the PDF standard that defines the object model deep scrub walks — this isn't obscure data, it's just data most tools skip.

FAQ

Does deep scrub change how the PDF looks or behaves?

No — deep scrub only touches metadata, not page content. Text, images, form fields, and layout are untouched; only Info dictionary fields, XMP streams, and object-level metadata keys are removed.

Is deep scrub slower than standard stripping?

It takes marginally longer since the engine walks every object in the file instead of just the top-level dictionary, but for typical documents the difference is a fraction of a second, not something you'll notice.

Does standard strip mode leave anything dangerous behind?

It leaves the Info dictionary clean, which is what most people check and what most casual sharing scenarios expose. It does not touch embedded XMP metadata streams, which is the specific gap deep scrub closes.

Is deep scrub available for file types other than PDF?

Deep scrub in its current form is PDF-specific, since XMP-stream and object-level metadata are largely a PDF phenomenon. For JPEGs, the equivalent extra step is the optional lossy re-encode, which recompresses the image to catch embedded thumbnail EXIF data that marker-stripping alone can miss.

Try deep scrub on a PDF.

Try MetaClean Pro free — remove metadata from your files in seconds.