What to check before sending a sensitive PDF

Five checks, in order, that catch the overwhelming majority of accidental disclosures — each takes under a minute.

5 min read · Updated

This is deliberately short, because a checklist nobody runs is worth nothing. Five checks, in this order, each under a minute.

1. Read the document as the recipient will

Open the final PDF — not the source file it was exported from — and go through every page. Exports fail in ways the source does not show: a comment rendered into the page, a placeholder never replaced, a chart cut off at a page break, tracked changes that came through as visible mark-up.

Pay particular attention to the last page and to appendices. They are where content nobody re-read tends to sit.

2. Extract the text and read that too

Run the file through PDF to Text and skim the output.

This is the single highest-value check in the list, because it reveals everything that is present in the file but not visible on the page: text underneath graphics, white-on-white text, content hidden in an optional layer. If a name you thought you had removed appears here, so it will for the recipient.

Search the extracted text for the specific things that must not be there — a client name, a salary figure, an address.

3. Clear the metadata

Open PDF Info and look at the title, author, subject and keywords.

The title very often contains the original filename, sometimes with a full folder path. The author is usually a real person's account name. On a document being published or sent outside your organisation, set the title to something intentional and clear the rest. Details of what these fields carry are in what is hidden inside a PDF.

4. Check annotations and any redactions

If the document went through review, open the comments panel in a full viewer. Review notes are stored separately from page content and may not display by default — which is exactly how an internal comment reaches an external reader.

If anything was redacted, verify it properly: select across the black area, and confirm the removed words did not appear in step 2. A covered-but-not-removed region is fully recoverable, as redaction versus covering explains.

5. Check the size and the destination

A document that bounces is a document that gets re-sent, often more carelessly the second time. Most mail systems cap attachments around 20–25 MB, and portals are frequently stricter.

If it is too large, compress it and check the result at 100% zoom before sending. Do not solve a size problem by uploading to a file-sharing service without thinking about who can reach that link — a public link to a confidential document is a worse outcome than a bounced email.

Two things that are not on the list

Password protection is not a substitute for any of the above. It controls opening, not content, and the password is often transmitted alongside the file.

Converting to images to prevent copying does not prevent copying — it prevents convenient copying. Screenshots and OCR exist. Rasterising is a legitimate tool for enforcing a redaction, not a general protection measure.

Doing this without uploading

Every check above runs in your browser on this site. That matters more than usual here, because the documents that need checking are precisely the ones you would rather not hand to a third party in order to find out what they reveal. The mechanics are described in why browser-based tools work this way.

Common questions

Do I need to do this for every document?
No. Apply it in proportion to consequence: a document going to a client, a regulator, a court or the public deserves the full pass; a routine internal file does not.
Should I password-protect the PDF as well?
Password protection helps against casual access but is only as strong as the password and the way you transmit it. Sending the password in the same email as the file achieves nothing.

Tools mentioned in this guide