How to Check If a PDF Redaction Worked: A Checklist

Before you start

Write down what was supposed to be removed: the names, numbers, dates, addresses and phrases. You can't test for something you haven't listed. Include fragments people might search for, such as a surname on its own, the last four digits of an account number, or an email domain.

Test a copy of the final file, the exact one you plan to send. Tests on a draft tell you nothing about the file that goes out.

Keep a short note of what you checked and when, with the file name. If a question comes up later about what was sent, a dated checklist is far more useful than a memory of having looked.

Most of these checks work in the PDF reader you already have. Checks 3 and 6 use free command-line tools, which are optional but catch things a reader can hide.

1. Select all, copy and paste

Open each redacted page, press Ctrl+A (Cmd+A on a Mac), copy, and paste into a plain text editor such as Notepad or TextEdit. Also try dragging a selection across the black boxes themselves.

What it catches: the most common failure, a box drawn over text that is still there. This is the check that exposed redactions in the court and government documents covered in our redaction guide.

What a pass looks like: the redacted words are missing from the pasted text. If a page was flattened to an image, nothing on it can be selected at all, which is also a pass.

2. Search for what you removed

Use the reader's Find command for every item on your list, one at a time. Search the whole document, not only the pages you redacted, because details repeat: in headers, footers, tables of contents, signatures and appendices.

What it catches: hidden text you didn't select in check 1, such as white text or text behind an image, and copies of the same detail on pages you didn't redact.

Its limit: search only finds what the reader can decode as text. A scanned page has no text layer, so search can't see what's printed on it. Read scanned pages with your eyes, or run OCR on them first (see how OCR works).

3. Extract the text with a second program

Different programs read PDFs differently, so a second opinion is cheap insurance. The free Poppler utilities include pdftotext, available through most Linux package managers, Homebrew on a Mac, and Windows builds:

pdftotext -layout redacted.pdf out.txt

Then search out.txt for your list. If you don't use a terminal, opening the PDF in a different reader, such as your web browser's built-in viewer, and repeating checks 1 and 2 is a reasonable substitute.

What it catches: text that one reader hides or fails to select, including text outside the visible crop area of a page.

4. Read the metadata

Open the document properties: File, then Properties or Document Properties, in most readers. Read the title, author, subject and keywords. These fields are often filled in automatically from the original document, so a file called "Smith settlement draft" used to keep that title after you redacted the name from every page — in Vault Studio Pro, a redacted save now clears them itself, but still run this check, both to confirm it and because not every tool does this.

If you have ExifTool installed, exiftool redacted.pdf prints every metadata field it finds, including XMP metadata that some readers don't show.

A trap when fixing it: some tools edit PDF metadata by appending the change to the end of the file instead of rewriting it. ExifTool's own documentation says its PDF edits work this way and can be reverted, so the old values are still in the file, and it suggests rewriting the file with qpdf afterwards. Whatever tool you use, run check 6 after changing metadata.

5. Look for attachments, bookmarks and comments

Open the reader's side panels and look through each one:

Attachments: a PDF can carry other files inside it, such as the original spreadsheet or an earlier draft. Vault Studio Pro removes these on a redacted save, but still check, especially for a file that came from somewhere else.
Bookmarks: their labels are text and may repeat a heading you redacted.
Comments and annotations: review notes, sticky notes and form field values are separate from the page content.
Layers: some PDFs have optional content layers that can be switched on and off.

6. Look for earlier versions inside the file

PDF allows incremental updates: instead of rewriting the file, an editor appends the changes to the end. Readers show the latest version, but the earlier one can still be inside, including text that was later deleted or covered. Security researcher Didier Stevens demonstrated recovering an earlier version by cutting the file at the end of its first revision.

A quick way to spot this: every revision ends with the marker %%EOF. On a Mac or Linux, run grep -a -c "%%EOF" redacted.pdf. A count of 1 suggests a single revision. A higher count means there may be earlier revisions, though not always: files optimised for fast web viewing (linearised) normally contain two markers. Some readers also show a signed document's earlier revisions in their signature panel.

If you find more than one revision and aren't sure it's harmless, rewrite the file as a single revision. qpdf, a free command-line tool, does this when you run it on the file and save the output as a new PDF. Then repeat checks 1 to 5 on the new file.

What these checks cannot prove

That you boxed everything. The checks confirm that what you meant to remove is gone. They can't tell you what you forgot to put on your list.
That an image is safe. Scanned pages and photos contain no searchable text. A name visible in a scan must be covered with a solid box and the page flattened, or it stays readable to anyone who looks.
That context doesn't give it away. The length of a black box, a surrounding sentence or an unredacted table total can reveal what was removed. Read the document as an outsider would.
That a file from someone else is complete. These checks look for leaks, not tampering.

What Vault Studio Pro does and does not check

Vault Studio Pro's redaction replaces each redacted page with an image of the page with the boxes burned in. By design, that removes the text, fonts and vector drawings on those pages, and the app saves a complete new file rather than appending to the old one. That covers checks 1 to 3 and 6 for the redacted pages.

Saving with at least one redaction box also strips metadata: title, author, subject and keywords are cleared, producer and creator are reset to a neutral value, and embedded attachments are removed — automatically, with no extra step. Saving with no redaction leaves all of that as it was unless you turn on "Remove metadata before saving" in the toolbar, which does the same thing on demand.

It does not do the rest for you, and we would rather you know that up front:

It doesn't touch bookmarks, so run check 5 for those either way.
It doesn't search the document for sensitive terms or suggest what to redact.
It doesn't verify redactions made in other programs, or produce a redaction log or certificate.

If you need Bates numbering for a production set, our sister product PDF Vault has it. Vault Studio Pro's plans are on the pricing page; redaction itself is free.

Try it

To make a redaction that passes checks 1 to 3 by design, use the redact tool in the free PDF editor: redacted pages are rebuilt as images, so no text is left on them. The redaction page explains the method, and the full redaction guide walks through it.

Open the PDF editor See plans and pricing

Sources

ExifTool PDF Tags (notes on incremental updates and reverting PDF metadata edits). Phil Harvey, ExifTool documentation
Solving a Little PDF Puzzle (incremental updates and earlier versions). Didier Stevens, May 2008
pdftotext(1) manual page. Poppler utilities, Debian manual pages
Can Epstein files be unredacted with a simple copy and paste? What we know. Snopes, December 2025
Redacting with Confidence. US National Security Agency, via the Federation of American Scientists