Clean hidden details out of a PDF before it goes out

A proposal that still names the old agency

Tomas is a freelance consultant with a client pitch due Thursday at nine. His file, proposal_final_v7.pdf, is 14 pages long and was built in a template a former colleague made while they both worked at an agency. The pages look like his own work. The file properties tell another story: the author field still carries the colleague's name and the agency, the creation date is from 2023, and the keywords say "internal draft, pricing v3".

None of that is printed on a page, so proofreading never catches it. Anyone who opens the document properties in their reader, or drops the file into a scanner, sees it in seconds. Cleaning it takes less time than writing the covering email.

Cleaning a file in PDFGym, step by step

The metadata remover and the privacy scan run on the same engine. Either one works; the first is set up for this exact job.

1. Open Remove metadata from PDF and choose the finished file. It is read in the browser tab and never uploaded.
2. Read the list. Findings are grouped under document details, software, dates, other metadata and embedded files, each with its value shown, so you can see the stale author name or the leftover keywords for yourself.
3. Untick anything you want to keep. Everything found starts selected, so a useful title such as "Website redesign proposal" only stays if you clear its box.
4. Press Remove metadata. The tool writes a new copy without the selected entries.
5. Wait for the check, then download. The copy is opened again: the selected items must be gone, the unselected ones must still be there, and every page and its selectable text must match the original.

If nothing supported turns up, the page tells you so and there is nothing to remove.

What the scan can find inside a PDF

Here is what the tool reports, where each item normally comes from, and what it can give away in a file like Tomas's.

Mistakes that leave your name behind

Cleaning a draft, then editing it again. Compressing, merging or re-exporting writes fresh software and date fields. Finish every other change first, including compression, and make metadata removal the last step.

Clearing only the basic properties. Some apps' properties dialogs edit the title and author but leave the XMP packet untouched, so the old values are still inside. Check the list the scan shows rather than trusting a blank dialog.

Forgetting attachments. A PDF with a costing spreadsheet attached hands the recipient the whole spreadsheet. Look under embedded files and remove anything that was meant for internal eyes only.

Expecting it to hide text on the page. A name in a footer or a signature block is page content, not metadata. Removing that for good is redaction, and a black box drawn on top is not enough; the redaction checker finds text left under boxes.

Overlooking photos. The tool checks the file's own fields and attachments only. It does not look inside images placed on the pages, so camera or location details in a photo need to be cleared before the photo goes into the document.

Trying to clean a protected file. Password-protected PDFs are refused. Remove the password with Unlock PDF first; the unlocked copy keeps the title, author and similar fields, so clean it afterwards.

Cases where a browser tool is the wrong choice

• You are producing documents for a court case, an audit or a records request. Metadata can be part of what you must hand over unchanged, so ask the lawyer or records officer before removing anything.
• Your company's document system stamps its own properties on every export. Cleaning the copy on your machine does not change the next export; ask IT whether the template or system setting can be changed instead.
• You have hundreds of files to clean. This tool takes one file at a time. A scripted desktop tool, or your IT department, will handle a batch better.
• The private detail is printed on the page. That is a job for redaction, not metadata removal.