How to Redact a Word Document or Spreadsheet: What You Gain and Lose by Converting to PDF

The instruction to "just black it out in Word" usually arrives with a deadline attached, from someone who has a records request or a production going out that afternoon. Black highlighting, a black font color and shapes drawn over the text all change how the words look. They leave the words themselves in the file, where copying the text or changing its color brings them back. A Word file also keeps content that the page never shows, so the safe way to redact a Word document runs through a PDF copy. Our PDF redaction guide explains what that copy must not carry once it is redacted.

How the PDF copy should be prepared depends on where it is going and who will read it. The steps below are for a records release under a freedom of information law, where the file as the organization holds it is the record, tracked changes and comments included. A production in litigation follows its own rules, covered straight after them, and the later sections weigh what the PDF route gains and costs for teams handling Word files, spreadsheets and slide decks in volume.

Redacting a Word document for a records release, step by step

The procedure works with the tools most offices already have, plus a redaction tool that removes content from a PDF instead of drawing over it:

1. Work on a copy, leaving the original Word file untouched.
2. Run Microsoft's Document Inspector on the copy to list what the file holds beyond its visible text, such as tracked changes, comments and hidden text, and leave that markup in place. It is part of the record, so deleting it before conversion withholds responsive content with no exemption and no marking. Draft text and reviewer comments that are withheld get a code like any other redaction. In US federal practice that code is often Exemption 5's deliberative process privilege, which the Justice Department's guide to Exemption 5 covers in detail.
3. In a workbook, unhide every row, column and worksheet before converting, so everything the file holds reaches the PDF, where it can be reviewed and redacted wherever an exemption applies.
4. Save the copy as a PDF that shows the markup, so each comment and tracked change reaches the page where it can be read and redacted.
5. Redact the PDF with software that removes the text under each redaction, since highlighting or shapes added in Word would carry through as covered text.
6. Compare the PDF with the original page by page, including every worksheet of a workbook and the speaker notes of a presentation.
7. Check the redacted PDF for hidden content, using the checks in our PDF sanitizing article, then release the PDF.

A litigation production keeps the document as it was

Discovery keeps the markup too, because the aim there is to produce each document as it existed, and the drafting history is often exactly what the other side asked for. A typical production images the document with its markup showing and redacts only privileged or protected content. Rule 34 of the Federal Rules of Civil Procedure requires a party to "produce documents as they are kept in the usual course of business or must organize and label them to correspond to the categories in the request."

Many discovery protocols settle on a split, with untouched files produced natively and redacted files produced as images or PDFs with their text. Our article on privilege review and eDiscovery redaction covers that production side in more depth.

Privilege redactions need their basis recorded as much as FOIA withholdings do, and in VIDIZMO Redactor the code lists go beyond the FOIA and Privacy Act defaults. Each organization can create its own list, so a firm can label every redaction with the privilege or protection it rests on, such as attorney-client privilege or work product. Every reviewer on the matter then picks from the same list. Production numbering and document families stay in the review platform. What Redactor hands back is the redacted PDF of each document, plus the redaction report as a CSV export.

What Word, Excel and PowerPoint keep out of sight

Microsoft's own inspection tool gives the clearest list of what an Office file can hide. Its support page names "Revision marks from tracked changes," comments, "Hidden rows," "Hidden worksheets," text formatted as hidden, and "Document properties and Personal Information." It also warns that "The Notes section of a PowerPoint presentation can contain text that you might not want to share publicly." Deleting a sentence while revision tracking is switched on keeps that sentence in the file as a tracked deletion, so a document can look clean while still holding its earlier wording.

Spreadsheets hide things in ways documents don't, since a cell can show one thing while depending on another. A formula can display a harmless total while pointing at a hidden worksheet that holds the names and account numbers behind it. A chart can keep its own copy of the values it plots, including values that no longer appear in any visible cell. Hiding a row or a sheet in Excel is a display setting, which is why step three unhides everything before conversion.

The UK Information Commissioner's Office fined the Police Service of Northern Ireland £750,000 in October 2024 over exactly this kind of hidden content in a released spreadsheet. A worksheet holding the surnames, initials, ranks and roles of all 9,483 officers and staff went out in a freedom of information release, missed even by quality assurance.

What redacting a PDF copy gains

Rendering a document to a PDF copy fixes what every page shows, and that fixed view is what makes a single removal and clean-up path possible for very different kinds of file. A Word report, a workbook, a slide deck and a saved email all become pages that can be searched, detected, redacted and checked the same way, by the same reviewers, with the same exemption codes. A signature image pasted into a Word letter can then be found the same way as a signature on a scanned contract, a subject our article on signature and form redaction takes further.

The released copy also resists the recoveries that native files invite, because a PDF holds only what was rendered onto its pages, with no live tracked change to accept and no worksheet to unhide. Keeping the native file as the source record matters just as much. Where a withholding is challenged, the original shows what was removed and why, and a later request for the same document can be answered afresh from it.

Saved email goes through the same route, and a message in .eml or .msg format converts to a PDF of its headers and body. The headers carry the names and addresses of everyone on the message, so they need the same review as the text. Attachments are separate parts under the mail standards, which RFC 2183 describes as "separate from the main body of the mail message," so each attachment is its own item to redact.

What it costs, and how to check what the conversion left out

Converting to a PDF gives up editability, live formulas and the native format, and a records statute can make that matter as much as a discovery rule does. The federal Freedom of Information Act says that an agency "shall provide the record in any form or format requested by the person if the record is readily reproducible by the agency in that form or format."

A PDF copy shows only what the conversion rendered, so anything the conversion leaves out is neither released nor reviewed. That protects against a leak, but it can also withhold material the requester was entitled to, since a reviewer who never sees a hidden worksheet cannot decide whether it belonged in the release.

What happens to tracked changes, comments, hidden rows, hidden sheets and speaker notes depends on the converter and its settings. That uncertainty is why step six compares the PDF with the original instead of assuming the conversion carried everything across. Wide spreadsheets that split across several pages, merged cells and embedded charts need the same side-by-side look, since their rendering can differ from what people see on screen in Excel.

Very large files add a different kind of cost, one that falls on the reviewers more than on the format. A slide deck of about 600 pages, or a report of about 750, can produce thousands of detections to review. Reviewing every one of them in a single sitting invites fatigue at exactly the pages where attention matters most. Splitting the review by section, working in batches with a fresh reviewer on each, and sampling a set of pages for a second check all keep a long file from being approved on momentum.

What Redactor converts, and what comes back

Redactor converts every uploaded document other than a DICOM medical image to a PDF rendition and redacts that rendition. The conversion covers 63 document file extensions, from Word, Excel and PowerPoint to the OpenDocument formats, OneNote, Visio, Project, CAD drawings, XPS, HTML and saved email. The redacted copy is always a PDF, so a spreadsheet comes back as pages rather than as a workbook. Document properties carried over from the source, such as author, company and last modified by, are checked against the organization's metadata list, with an optional AI pass, and flagged values are masked in the copy. Where native files are expected, agree the form of redacted documents with the requester, or in the discovery protocol, before redaction starts.

The formats Redactor accepts, and the detection methods that apply to each, are summarized on the Redactor document redaction page.