How to Anonymize a Document for Double-Blind Peer Review

The Metadata Problem Nobody Talks About

You've done everything right. Your name is off the title page. Acknowledgements are gone. Self-citations read "[Author, 2023]." You export to PDF and hit submit.

And then a reviewer who happens to be curious opens File → Properties.

There you are: full name, your university's IT department in the Company field, and a Last Modified By entry that names your co-author. Sometimes the file path in the PDF Producer string includes the hostname of your lab's shared drive. Sometimes the tracked-change history lists every comment your supervisor left during editing, each one stamped with their name and institution.

This isn't a hypothetical. The Harvard Kennedy School's Misinformation Review published a detailed examination of how blind peer review fails, noting that file metadata is one of the most reliable — and most overlooked — routes through which author identity leaks to reviewers. The Chronicle of Higher Education documented the specific mechanism years earlier: Word embeds the registered user name in document properties automatically, and that name survives PDF export unless you explicitly remove it first.

What Actually Leaks: The Full Metadata Attack Surface

Before the checklist, it helps to understand exactly which fields can expose you. There are more than most people expect.

In a Microsoft Word .docx File

Word populates these fields automatically from your OS user account and Office registration, without asking:

• Author — the name registered in Word's user settings at the time the document was created
• Last Modified By — whoever saved the file most recently; if a co-author or your supervisor edited it, their name appears here
• Company — pulled from the Windows/macOS user account or the Office license, often your institution's name
• Revision Number — how many times the document has been saved; a revision count of 847 suggests a very active working document
• Total Editing Time — the cumulative minutes Word recorded while the document was open and being typed in
• Template path — if you started from a department template, the path (e.g., \\university.edu\dept\templates\lab_paper.dotx) is embedded
• Tracked changes — every insertion, deletion, and comment is stored with the editor's display name and a timestamp, even after you "accept all changes"
• Comments — reviewer comments made by co-authors or collaborators include their display name

In a PDF File

When Word exports to PDF, it carries most of that metadata forward into the PDF's own fields:

• Author — mapped directly from the Word Author field
• Creator — the application that created the source document (e.g., "Microsoft Word for Microsoft 365")
• Producer — the PDF engine that converted it; on a university-managed machine this often includes the hostname or department name (e.g., "Acrobat PDFMaker 23.0 for Word — CHEM-DEPT-PC01")
• Title — may contain the document's internal working title, which could differ from the submission title in a revealing way
• Creation Date and Modification Date — can narrow down the research timeline
• Keywords — if you set keywords in Word's document properties, they transfer to the PDF

The revision count and total editing time deserve special mention. A reviewer who sees a document with 3 minutes of total editing time and 2 revisions is looking at a template quickly filled in. One with 11,000 minutes and 400 revisions is a deeply worked paper. Neither of those numbers identifies you by name, but combined with other signals, they can narrow the field significantly.

The Word Anonymization Checklist

Work through these steps in order before you export to PDF. Doing them after export means starting over — the damage is already in the PDF.

1. Change the Author in Word settings. Go to File → Options → General. Under "Personalize your copy of Microsoft Office," clear the User name and Initials fields (or replace with "Author"). This affects what gets embedded in new documents and what Document Inspector can clean.
2. Run Document Inspector. File → Info → Check for Issues → Inspect Document. Run all checks. The inspector will surface: Comments, Revisions, Versions, and Annotations; Document Properties and Personal Information; Hidden Text; and Embedded Documents. Click "Remove All" next to each category that returns results.
3. Accept or reject all tracked changes before inspecting. Document Inspector removes the stored revision data, but it works more reliably if you've first resolved every tracked change. Review → Accept → Accept All Changes. Then re-run Document Inspector.
4. Delete all comments manually. Review → Delete → Delete All Comments in Document. Even after Document Inspector, double-check: a single overlooked comment thread can name a collaborator.
5. Check the document template reference. File → Info → Properties (right column) → Advanced Properties → Custom tab. Remove any custom properties. In the Summary tab, clear Author, Company, Manager, and Category fields explicitly.
6. Rename the file. Your filename is part of the submission package. "Smith_NatureComms_v12_FINAL_after_advisor.docx" is not anonymous. Use something like "manuscript.docx" or the journal's preferred naming convention.
7. Check embedded images and figures. If figures were created in another application and embedded as objects, they may carry their own metadata. Export figures as plain image files (TIFF or PNG) and re-insert rather than embedding OLE objects.
8. Save as a new file after cleaning. Use File → Save As to create a fresh copy after all cleaning steps. This resets the file creation timestamp and clears some residual metadata that Save (overwrite) doesn't touch.

The PDF Anonymization Checklist

After cleaning the Word file, export to PDF — and then clean the PDF too. Word-to-PDF conversion is imperfect; some fields transfer that Document Inspector missed, and the PDF adds its own Producer field that can name your machine.

1. Check the PDF immediately after export. Open the PDF in Adobe Acrobat Reader (free), press Ctrl+D (Cmd+D on Mac), and review the Description tab. Author, Title, Subject, Keywords, and the Producer and Creator strings should all be blank or non-identifying.
2. Strip PDF metadata with a dedicated tool. If any fields contain identifying information, the fastest approach is our MetaClean PDF tool — drag your file in, click Remove Metadata, download the cleaned version. The whole thing happens in your browser; your manuscript never leaves your device. Alternatively, Adobe Acrobat Pro users can use Tools → Redact → Sanitize Document, which is the most thorough Adobe option.
3. Check the Producer string specifically. The Producer field often escapes notice because it names software, not people. But "Acrobat PDFMaker 23 for Word — BIOCHEM-WORKSTATION-14" names your department. Strip it.
4. Re-examine embedded fonts and images. XMP metadata (a more modern PDF metadata standard) can survive basic cleaning. After stripping, open the cleaned PDF and use File → Properties → Custom tab to verify nothing persists.
5. Verify with a second tool. Drop the cleaned PDF back into MetaClean and review the metadata viewer. An empty or near-empty report (only technical rendering fields remain) confirms the job is done. Alternatively, exiftool document.pdf from the command line gives the most complete field-by-field readout.
6. Rename the PDF file. Same rule as the Word file. "manuscript_blind.pdf" or the journal's required naming convention — not your name, your institution, or the project's internal codename.

For a deeper look at what PDF metadata contains and how to handle it systematically, our complete PDF metadata removal guide covers every method from browser tools to ExifTool command-line scripts.

The Visible Text Checklist

Metadata is the hidden layer — but the visible text has its own anonymization requirements that journals take equally seriously. A submission that passes metadata inspection but contains "as we showed in our previous work at [University]" is still compromised.

1. Remove all author names and affiliations from the title page, headers, footers, and any in-text mentions.
2. Replace self-citations with "[Author, Year]" or "[Anonymous, Year]" — don't just delete them, or the narrative won't make sense to reviewers.
3. Remove acknowledgements. Funders, colleagues thanked by name, and grant numbers can all identify you. Move acknowledgements to a separate title-page document submitted through a different field in the submission system.
4. Check funding disclosures. Grant numbers are often traceable to specific labs or PIs in public funding databases.
5. Neutralize institution-specific references. "Data was collected at the [Distinctive Name] Laboratory" needs to become "Data was collected at our research facility" or similar.
6. Review figure captions and supplementary materials. Logos, institutional watermarks, and instrument serial numbers in figures have all been used to identify authors.

Final Verification Before Submission

Don't trust the process — verify the output. This takes two minutes and catches anything the cleaning steps missed.

1. Open the final PDF in Acrobat Reader. Ctrl+D → Description tab. All identifying fields should be blank.
2. Search the PDF text (Ctrl+F) for your last name, your institution's name, your city, and your co-authors' names. Zero results is what you want.
3. Run the final file through MetaClean's metadata viewer at metaclean.app/pdf-metadata — the viewer shows all fields present in the document. An empty or near-empty report confirms clean submission. Your file is processed entirely in your browser, so there's no privacy concern about uploading a confidential manuscript.
4. Check the file properties in your operating system (right-click → Properties on Windows, Get Info on macOS). The OS-level metadata — created date, modified date, and sometimes author — is separate from the PDF internal metadata and worth a quick scan.

If anything comes up during verification, go back to the appropriate cleaning step rather than trying to edit around it. A rushed fix often misses something.

Why Journals Take This Seriously

Double-blind review exists because research consistently shows that reviewer bias — toward authors from prestigious institutions, from certain countries, or who are well-known in the field — affects review outcomes even when reviewers try to be objective. A 2017 study in PLOS ONE found that papers from authors at top-ranked institutions received more favorable peer review scores, independent of paper quality. The double-blind format is the structural fix for that bias.

But the fix only works if it actually holds. Metadata leaks are one of the documented ways it doesn't. When a reviewer can identify the submitting lab from a PDF Producer string or a tracked-change author name, the double-blind protection collapses — and neither the author nor the journal editorial office may ever know it happened.

The good news is that once you've built metadata cleaning into your submission workflow, it takes about five minutes per submission. The hard part is remembering to do it. The easiest approach: make it the last step before you generate the final PDF, not an afterthought after the file is already in the submission system.

If you're working with sensitive documents beyond academic submissions — financial reports, legal filings, or business proposals — the same metadata risks apply. Our article on hidden data in PDFs as a business risk covers the professional context in detail.

Frequently Asked Questions

Does converting a Word file to PDF automatically remove the author metadata?

No. Word-to-PDF conversion maps most Word metadata fields directly into the PDF's own Author, Title, Creator, and Producer fields. If your Word file contains your name in the Author property, that name will appear in the PDF's metadata unless you clean the Word file first and then verify the resulting PDF with a metadata viewer or tool like MetaClean.

Do I need to anonymize metadata if I'm submitting to a single-blind journal?

Single-blind review means reviewers know who you are but you don't know who they are, so there's no anonymization requirement on your side. That said, removing personal metadata is still good practice for any externally shared document — it limits unintended information disclosure regardless of the review format.

Will accepting all tracked changes in Word fully remove reviewer names?

Accepting all tracked changes removes the visible markup — the colored insertions and strikethroughs — but the underlying revision data, including editor names and timestamps, may persist in the document's XML structure. You need to run Document Inspector (File → Info → Check for Issues → Inspect Document) after accepting changes to remove the stored revision information.

Can a reviewer actually access Word metadata in a PDF submission?

Yes, easily. In Adobe Acrobat Reader, pressing Ctrl+D (Cmd+D on Mac) opens Document Properties and shows the Author, Creator, and Producer fields. Free command-line tools like ExifTool display all embedded fields in seconds. No special forensic software is required — any technically curious reviewer can check metadata in under a minute.

What's the safest way to anonymize a manuscript without expensive software?

The reliable free approach: run Document Inspector in Word, export to PDF, then strip the PDF metadata using a browser-based tool. MetaClean's PDF tool and Office tool handle both steps in the browser with no upload required — your manuscript stays on your device throughout. Verify the output by dropping the cleaned file back into the metadata viewer before submitting.

Are there journals that strip metadata automatically on submission?

Some editorial management systems apply automatic metadata stripping during the upload process, but this is not universal and is rarely documented in author guidelines. Relying on the journal's system to clean your file is not recommended. Author-side anonymization before submission is the only approach that guarantees clean metadata regardless of what the submission system does or doesn't do.