A manuscript can look perfectly anonymous while quietly carrying your name in comments, file properties, hyperlinks, or PDF metadata. That is the unnerving part: the identity leak is often invisible until an editor or reviewer notices it. In about 15 minutes, this workflow will help you create a clean submission copy, remove obvious and hidden identifiers, and verify both Word and PDF files without damaging your master manuscript. The goal is not theatrical secrecy. It is a calm, repeatable process that protects review integrity and prevents a tiny metadata crumb from undoing months of careful research.
Who This Workflow Is For and Not For
This guide is for researchers, graduate students, independent scholars, conference presenters, and editorial assistants preparing files for double-blind or anonymous peer review. It also helps corresponding authors who receive a last-minute message saying, “Please remove all identifying information and resubmit.” That sentence has a special talent for arriving five minutes before dinner.
The workflow applies to manuscripts created in Microsoft Word and submitted as DOCX, PDF, or both. It is especially useful when several coauthors have edited the file, because each collaborator may leave comments, initials, tracked revisions, linked folders, or personal file properties behind.
Eligibility Checklist: Use This Workflow When
Your manuscript is a good fit for this workflow if:
- The journal uses double-blind, anonymous, or masked peer review.
- You must upload a blinded manuscript separately from a title page.
- The submission includes Word files, PDFs, appendices, tables, figures, or supplementary files.
- Coauthors, supervisors, editors, or statisticians have commented on the document.
- The paper contains self-citations, acknowledgments, funding statements, or institution-specific details.
Who Should Not Follow It Blindly
Do not remove information merely because it looks identifying if the journal explicitly requires it in the review file. Some publications ask authors to retain clinical trial registration numbers, ethics approval details, data repository links, conflicts of interest, or funding disclosures during review.
Single-blind and open-review journals may not require author anonymization at all. Registered reports, preprints, conference proceedings, legal filings, and grant applications may follow different rules. The journal’s current author instructions outrank any generic checklist, including this one.
Why Apparently Anonymous Manuscripts Still Reveal Authors
Anonymization is not one editing task. It is a three-layer inspection: visible text, hidden file data, and contextual clues. Cleaning only the title page is rather like removing your name from a suitcase while leaving the luggage tag attached.
I once reviewed a document whose first page said “Author details removed for review.” The comments pane, however, displayed the author’s full institutional email beside a note reading, “Ask my department chair whether this claim is too strong.” The prose was anonymous. The sidebar was practically a business card.
The Three Identity-Leak Layers
| Layer | Typical Leaks | How to Check | Risk |
|---|---|---|---|
| Visible content | Names, affiliations, acknowledgments, bios, self-identifying citations | Read the manuscript, headers, footers, tables, figures, and references | High |
| Hidden file data | Author properties, comments, revisions, custom fields, embedded objects | Use Word inspection tools and review all collaboration features | High |
| Contextual clues | Named laboratories, distinctive datasets, prior-paper wording, repository accounts | Read as an informed outsider and follow every link | Medium to high |
- Inspect visible wording.
- Inspect hidden document data.
- Inspect clues created by links, filenames, and research context.
Apply in 60 seconds: Search the document for your surname, institution, email domain, laboratory name, and ORCID.
The Double-Blind Workflow at a Glance
The safest process moves in one direction: duplicate, anonymize, inspect, export, sanitize, and verify. Avoid bouncing between your master file and submission file. That creates version confusion, and version confusion is where clean manuscripts go to acquire mysterious old comments.
Visual Guide: The Six-Gate Anonymization Workflow
Create a separate submission copy and preserve the master.
Remove names, affiliations, acknowledgments, and direct self-identification.
Resolve comments, revisions, properties, hidden text, and embedded data.
Create the required PDF from the cleaned Word file.
Check metadata, annotations, attachments, links, and hidden content.
Open both files outside your normal editing environment and search again.
A Practical File-Format Comparison
| Format | Main Strength | Main Anonymity Risk | Best Use |
|---|---|---|---|
| DOCX | Editable and convenient for editorial production | Comments, tracked revisions, author properties, custom XML, hidden text | When the journal specifically requests an editable manuscript |
| Stable page appearance | Metadata, annotations, attachments, active links, document layers | When reviewers need a fixed reading copy | |
| Flattened image PDF | Fewer editable text objects | Poor accessibility, weak search, large file size, remaining metadata | Rarely appropriate unless the journal requests it |
Flattening everything into an image may sound wonderfully final, but it can make the paper difficult for screen-reader users, editors, and reviewers who search within the document. A file should be anonymous without becoming hostile to ordinary reading.
Step 1: Prepare a Safe Submission Copy
Begin with the final master manuscript, but do not anonymize that file directly. Save a duplicate in a separate folder and give it a neutral filename such as Manuscript_Blinded.docx. Keep the author title page, cover letter, disclosure forms, and unblinded source material elsewhere.
One research team I worked with used filenames such as “Smith_Lab_Final_ReallyFinal3.docx.” They removed every name inside the manuscript and then uploaded the filename unchanged. The lesson was inexpensive, fortunately, but memorable.
Build a Simple Submission Folder
- 01_Master_Unblinded: The authoritative working manuscript with complete authorship information.
- 02_Blinded_Working: The copy being cleaned for review.
- 03_Final_Upload: Only files that have passed the final verification.
- 04_Forms_and_Title_Page: Author details, declarations, cover letters, and separate forms.
Use local or institution-approved storage when the manuscript contains unpublished, confidential, commercial, clinical, or identifiable information. Avoid uploading it to random “free metadata cleaner” websites. Free can become rather expensive when the currency is control over an unpublished paper.
Read the Journal’s Instructions Before Editing
Look for terms such as “blinded manuscript,” “anonymous review,” “separate title page,” “self-citation,” “funding disclosure,” and “data availability.” Confirm whether figures, appendices, supplementary files, response letters, and reporting checklists must also be anonymized.
Some systems automatically hide author information entered into submission fields. Others send uploaded files to reviewers almost exactly as received. Never assume the platform will clean the document for you unless the journal states that it does.
- Separate blinded and unblinded materials.
- Use neutral filenames.
- Follow the journal’s current file instructions.
Apply in 60 seconds: Create a folder named “Final Upload” and keep it empty until each file passes verification.
Step 2: Remove Identifiers From the Visible Text
Start with the content a reviewer can see. Work from the first page to the last, including headers, footers, captions, tables, appendices, references, and supplementary notes. Search helps, but it does not replace reading. A search for “University of Michigan” will not find “our Ann Arbor laboratory.”
Remove or Relocate Direct Identifiers
- Author names, degrees, affiliations, departments, and correspondence details
- Email addresses, phone numbers, postal addresses, and ORCID identifiers
- Author contribution statements when they name contributors
- Acknowledgments identifying colleagues, laboratories, institutions, or funders
- Conflict-of-interest text containing names or company relationships
- Running headers containing surnames or shortened author lists
- Institutional logos, letterheads, watermarks, and branded figure templates
Move required author information to the separate title page if the journal requests one. Do not simply delete legally, ethically, or editorially required disclosures from the submission package. Separation is usually safer than erasure.
Handle Self-Citations Without Performing Linguistic Acrobatics
Self-citation does not automatically reveal authorship. A neutral sentence such as “Previous research found a similar effect” followed by a standard third-person citation may be acceptable. The risky version is “In our earlier study, we introduced this instrument,” especially when the reference names the same distinctive dataset or research site.
Follow the journal’s stated rule. Possible treatments include retaining ordinary third-person citations, replacing identifying details with “Author, year,” or temporarily removing a reference entry. Do not invent a method because it feels anonymous. Editors dislike missing references almost as much as reviewers dislike accidental autobiography.
Inspect Contextual Clues
Review descriptions of laboratories, clinics, archives, companies, communities, and specialized equipment. A sentence can identify a team without naming it. For example, “the only pediatric robotics center in northern Alaska” offers a reviewer a very short detective story.
Check data repository URLs, project websites, preregistrations, code repositories, video links, survey landing pages, and supplementary spreadsheets. Opening a link may expose a username, profile photo, laboratory name, commit history, or account owner.
Do Not Break Methodological Transparency
Anonymization should not make the research impossible to evaluate. Preserve information needed to assess sampling, ethics, methods, conflicts, and reproducibility unless the journal permits a temporary blinded substitute.
When a site name is scientifically necessary, use a neutral temporary label only if the journal allows it, such as “University Hospital A.” Keep a private replacement list so the original wording can be restored accurately after review.
Step 3: Clean Hidden Data in Microsoft Word
After visible editing, treat the DOCX file as a container rather than a page. It may carry document properties, collaborator names, revision histories, comments, hidden text, headers, embedded files, and links that do not appear during a normal read-through.
Menu names differ slightly across Word versions, operating systems, and Microsoft 365 plans. When possible, perform the final inspection in a desktop version of Word rather than relying only on a browser editor.
Resolve Comments and Tracked Changes Correctly
Turn on the full markup display before cleaning. Review comments, insertions, deletions, formatting changes, and moves. Accepting all changes does not always delete comments, and selecting “No Markup” merely hides revisions from view. Hiding is not removing. Curtains are not walls.
- Open the Review tab.
- Display all markup and all reviewers.
- Read unresolved comments for content that must be retained.
- Accept or reject every tracked change intentionally.
- Delete all comments after incorporating necessary edits.
- Turn Track Changes off before the final save.
- Close and reopen the file to confirm no markup returns.
A coauthor once deleted a paragraph but left the deletion tracked. In the clean view, the paragraph vanished. In the reviewer’s markup view, it returned with the coauthor’s initials and a sentence naming the principal investigator. Ghost prose has excellent timing.
Inspect Document Properties and Personal Information
In Word desktop, look for the document inspection feature under the file information or issue-checking area. Inspect a copy, not your only master. Removal can affect comments, custom data, headers, hidden content, and embedded elements that may be difficult to reconstruct.
Review every reported category rather than pressing “Remove All” reflexively. For a blinded submission, author properties and collaboration data are usually unwanted. Custom XML or embedded objects may require more care if they support citations, equations, accessibility, or journal production.
Check Headers, Footers, Fields, and Hidden Text
- Open every header and footer, including first-page and odd/even variants.
- Inspect page-number fields, document-title fields, and automatically generated properties.
- Search for hidden text and verify that formatting marks are visible during review.
- Check text boxes, shapes, callouts, endnotes, footnotes, and floating captions.
- Update the table of contents only after heading edits are complete.
- Inspect document templates and linked style sources when institutional templates were used.
Check Hyperlinks and Embedded Objects
Right-click hyperlinks and inspect their full destination. Display text may say “project repository” while the underlying URL contains an author surname. Do the same for linked charts, embedded spreadsheets, citation-manager fields, audio, video, and attached objects.
If a chart was copied from an Excel workbook, the embedded workbook may contain sheet names, authorship details, comments, raw data, or file paths. When editable source data is not required, consider replacing the embedded chart with an accessible static figure while preserving the source privately.
Show me the nerdy details
A DOCX file is a package of XML files and related assets. Renaming a copy from .docx to .zip can reveal folders containing core properties, comments, relationships, media, settings, footnotes, endnotes, and custom XML. This is useful for advanced verification, but manual editing inside the package can corrupt the file. A safer approach is to use Word’s inspection tools, remove collaboration artifacts deliberately, save a new copy, and then inspect the package only when a high-sensitivity submission justifies deeper checking.
- Display all markup.
- Resolve revisions and delete comments.
- Run document inspection on a duplicate.
Apply in 60 seconds: Change the review display to show all markup and check whether any reviewer names appear.
Step 4: Export and Sanitize the PDF
Create the PDF only after the Word file has passed its visible and hidden-data checks. Exporting an unclean DOCX does not purify it. It may transfer the title, author, subject, keywords, comments, links, bookmarks, accessibility labels, or other properties into the PDF.
Use a Controlled Export
- Save and close the cleaned Word document.
- Reopen it and confirm that comments and revisions are absent.
- Export or save as PDF using your approved desktop software.
- Choose options that preserve accessibility and searchable text.
- Do not include document properties unless the journal requires them.
- Save the PDF with a neutral filename in the blinded working folder.
A virtual “Print to PDF” process may remove some interactive features, but it is not a universal privacy tool. It can also weaken bookmarks, tagged structure, hyperlinks, mathematical text, and accessibility. Use it only when you understand what the output retains and what it discards.
Inspect PDF Metadata
Open the PDF’s document properties and review the title, author, subject, keywords, creator, producer, and custom fields. The creator and producer fields usually name software rather than a person, but the author field should not identify the research team in a blinded file.
Check the initial view, bookmarks, layers, attachments, comments, stamps, form fields, and embedded media. A PDF with a blank author property can still contain an annotation signed with someone’s full name.
Search the PDF Independently
Search for author surnames, initials, institution names, city names, email domains, ORCID, laboratory labels, grant numbers, and distinctive project terms. Repeat searches with shortened forms and common abbreviations.
Copy a paragraph from the PDF into a plain-text editor. This can reveal hidden or duplicated text behind figures, redaction boxes, headers, or OCR layers. If covered text can still be copied, the cover is decorative, not protective.
Use True Redaction When Content Must Be Removed
Drawing a black rectangle over text does not necessarily delete the underlying text. Proper redaction marks the content, applies the redaction, and removes associated hidden information. Afterward, save a new file and test whether the removed wording can be searched, selected, or copied.
For routine manuscript anonymization, editing the source document and exporting again is usually cleaner than redacting many PDF passages. Redaction is more useful when the original source cannot be changed or a narrow element must be permanently removed.
- Review document properties.
- Inspect annotations, bookmarks, attachments, and links.
- Test removed text with search, selection, and copy.
Apply in 60 seconds: Open PDF properties now and look specifically at the Author, Title, Subject, and Keywords fields.
Step 5: Verify the Files Like a Curious Reviewer
The final check should imitate the reviewer’s experience, not the author’s editing routine. Move the candidate files into the Final Upload folder, close every related application, and reopen them from that folder. This catches stale exports, incorrect attachments, and the classic error of cleaning one copy while uploading another.
Run a Two-Pass Verification
Pass one is mechanical. Search names, inspect properties, review comments, open links, and compare filenames. Pass two is contextual. Read the manuscript as someone familiar with the field. Ask whether a distinctive sentence, dataset, institution, citation pattern, or project URL effectively names the authors.
For important submissions, ask a colleague who is not an author to perform the second pass. Fresh eyes notice clues that have become invisible through repetition. Authors can stare at their own institution’s acronym for six months and stop seeing it as an institution at all.
Risk Scorecard for the Final Upload
Assign the listed points for each unresolved issue:
| Issue | Points | Required Action |
|---|---|---|
| Name, email, ORCID, or affiliation visible | 5 | Stop and correct before upload |
| Comments, revisions, or reviewer names remain | 5 | Resolve, delete, inspect, and reopen |
| Personal author metadata remains | 4 | Remove and regenerate dependent files |
| Repository or project link identifies the team | 3 | Use permitted anonymous access or ask the editor |
| Distinctive contextual clue remains | 2 | Assess scientific necessity and journal rules |
| Neutral filename or file version is uncertain | 2 | Rebuild the Final Upload folder |
Score 0: Ready for a final visual check. Score 1–3: Review the journal policy and correct what is practical. Score 4 or more: Do not upload yet.
Short Story: The PDF That Passed Every Search
A doctoral researcher cleaned a manuscript carefully. She removed the title page, replaced acknowledgments, neutralized self-references, deleted comments, and searched the PDF for every author surname. Nothing appeared. The file seemed ready.
Before submission, a colleague opened the PDF and clicked a figure. The figure contained a hyperlink to a cloud folder named after the principal investigator. The visible caption was harmless, the metadata was clean, and the manuscript search had passed. The identity leak lived in the object action attached to the image.
They removed the link, exported a new PDF, and added “click every linked object” to their laboratory checklist. The practical lesson is small but durable: verification should include behavior, not only words. Open links, inspect attachments, test bookmarks, and interact with figures. A reviewer does not experience a manuscript as a static pile of searchable sentences.
Final Upload Checklist
- The title page is separate if required.
- All filenames are neutral.
- The DOCX shows no comments or tracked revisions.
- Word document properties contain no personal author information.
- The PDF properties contain no author identifiers.
- Headers, footers, figures, captions, tables, and supplements are clean.
- Every hyperlink and repository page has been checked.
- The manuscript remains readable, accessible, and methodologically complete.
- The files in Final Upload are the exact files selected in the submission portal.
- Search mechanically.
- Read contextually.
- Interact with links, figures, fields, and attachments.
Apply in 60 seconds: Ask one nonauthor colleague to open the final PDF and identify anything that hints at authorship.
Common Anonymization Mistakes
Removing the Title Page and Stopping
This addresses only the most obvious identifier. Names may remain in headers, footers, acknowledgments, author-contribution statements, file properties, comments, figure credits, appendices, and supplementary files.
Using “No Markup” Instead of Removing Markup
No Markup changes the display. It does not necessarily remove revisions. The next person can switch views and restore the edit history, including deleted names and reviewer initials.
Accepting Changes but Forgetting Comments
Tracked changes and comments are separate collaboration features. Clean both. Also check resolved comments, because software versions may display or retain them differently.
Replacing Names With Obvious Placeholders
Text such as “BLINDED UNIVERSITY, the only institution in Vermont offering this program” is technically masked but functionally transparent. Replace only what the journal permits, and avoid surrounding clues that defeat the substitution.
Over-Anonymizing the Reference List
Removing every self-citation can distort the scholarly record, weaken the literature review, and create conspicuous gaps. Use neutral third-person language unless the journal instructs you to substitute or remove identifying references.
Forgetting Supplementary Files
Spreadsheets can include author names, comments, hidden sheets, formulas with local file paths, named ranges, and document properties. Presentations, images, reporting checklists, code files, and response letters also require inspection.
Teams handling larger privacy workflows may find the same principles useful in privacy-preserving data preparation, where removing a label does not always remove the underlying identification risk.
Uploading the Wrong Version
This is painfully ordinary. The author cleans Manuscript_Blinded_Final.docx and uploads Manuscript_Final.docx. Neutral folders and a nearly empty Final Upload directory reduce this risk more effectively than heroic memory.
Using an Online Converter Without Reviewing Its Terms
Unpublished research may include confidential findings, proprietary data, participant information, patent-sensitive material, or contractual restrictions. Use institution-approved tools and secure transfer practices. The same discipline appears in secure file-intake workflows: know where a file goes, who can access it, and how long it is retained.
Scope and Confidentiality Note
This workflow reduces accidental identity disclosure, but it cannot guarantee that knowledgeable reviewers will not infer authorship. A distinctive topic, famous dataset, preprint, conference presentation, writing style, or prior study may make the research team recognizable even when the files are technically clean.
Do not alter scientific facts, fabricate neutral institutions, misstate ethics approvals, or conceal required conflicts merely to make identification harder. Double-blind review depends on reasonable masking, not fictional biography.
When the manuscript includes protected health information, personal data, trade secrets, export-controlled material, security-sensitive information, or unpublished patent claims, follow institutional policy and applicable agreements. Anonymizing authors is not the same as de-identifying research participants.
A laboratory or publisher creating a repeatable policy should document approved tools, access controls, retention periods, and review responsibilities. Broader governance and audit practices can help teams manage manuscript-processing tools without turning every submission into a fresh improvisation.
When to Ask the Journal or an Editor for Help
Ask before guessing when anonymization conflicts with transparency, ethics, reporting standards, or data-access requirements. Editorial offices routinely answer these questions. A concise inquiry is preferable to submitting a scientifically incomplete manuscript or exposing an identity that the journal intended to mask.
Contact the Journal When
- The author instructions contradict the submission portal.
- A required ethics approval statement names the institution.
- A trial registration, preregistration, or repository link identifies the authors.
- The journal requires a data-availability statement but anonymous repository access is unavailable.
- Self-citations are central to the method and cannot be disguised without harming comprehension.
- A proprietary instrument, patented method, or named facility is essential to evaluation.
- The submission system merges the title page and manuscript into one reviewer file.
- You discover an identity leak after submission.
Decision Card: Fix It Yourself or Ask?
Fix it yourself when: the issue is clearly accidental, such as a filename, comment, author property, running header, or visible email address.
Ask the editor when: removing the information could affect scientific interpretation, ethics reporting, conflicts, data access, registration, or compliance with the journal’s stated policy.
Withdraw or replace the file promptly when: a submitted manuscript contains confidential personal data, a serious identity leak, or the wrong unblinded version.
When requesting help, state the manuscript number, file involved, exact issue, and proposed correction. Avoid sending a long account of every click that led to the problem. Editors need the location of the smoke, not the autobiography of the toaster.
FAQ
How do I anonymize a manuscript for double-blind review?
Create a separate copy, remove visible author identifiers, neutralize self-identifying wording according to journal policy, resolve tracked changes, delete comments, inspect Word properties, export a new PDF, inspect PDF metadata, and perform a final search across every submission file.
Does deleting the author’s name from the first page remove Word metadata?
No. The document may still contain author properties, collaborator names, comments, revisions, custom fields, hidden text, templates, or embedded objects. Use Word’s document inspection features and manually review collaboration elements.
Does accepting all tracked changes remove reviewer names?
Not necessarily. Accepting changes handles revisions, but comments may remain. Document properties and collaboration information may also remain. Delete comments separately, inspect the file, save it, close it, and reopen it before submission.
Can reviewers see the author name in PDF metadata?
They may be able to view the Author field and other document properties using a PDF reader. Review the title, author, subject, keywords, annotations, attachments, bookmarks, and links before upload.
Should I remove my own publications from the reference list?
Usually not unless the journal specifically instructs you to do so. Cite your work in neutral third-person language and avoid phrases such as “our previous study.” Follow the publication’s rules for placeholders such as “Author, year.”
Should acknowledgments be removed for peer review?
Often they are moved to the separate title page because they can identify colleagues, institutions, facilities, or funders. However, journals differ. Preserve required disclosures in the appropriate submission file rather than deleting them from the entire package.
Is printing a Word document to PDF enough to anonymize it?
No. Printing may remove some editable features, but it does not guarantee clean metadata or eliminate identifying visible content, links, OCR text, or annotations. It may also reduce accessibility. Inspect the resulting PDF independently.
Can I use an online metadata remover for an unpublished manuscript?
Use caution. Review the service’s privacy terms, retention policy, data location, security controls, and institutional approval before uploading unpublished or confidential work. Local or institution-approved software is generally easier to assess and control.
What should I do if I uploaded the unblinded manuscript by mistake?
Contact the editorial office promptly. Provide the manuscript number, identify the incorrect file, and request replacement or withdrawal. Do not assume the file has not been distributed merely because the submission status has not changed.
Can double-blind review guarantee that reviewers will not recognize the authors?
No. Technical cleaning reduces direct disclosure, but specialists may infer authorship from subject matter, prior publications, datasets, preprints, or writing patterns. The practical standard is reasonable, policy-compliant masking rather than absolute unrecognizability.
Conclusion: Submit the Clean Copy, Keep the Rich Copy
The unsettling clue from the beginning was never just the name on the title page. It was the collection of small traces around it: a comment author, an active figure link, a repository account, a PDF property, or an old filename waiting in the wrong folder.
A reliable workflow solves that problem without mutilating the manuscript. Preserve the complete master. Create a blinded duplicate. Clean visible content, inspect Word’s hidden data, export a fresh PDF, sanitize it, and verify both files from the reviewer’s perspective.
Your next step takes less than 15 minutes: create the three-folder structure, duplicate the manuscript, search for five identifiers, and turn on full markup display. That modest start removes the highest-risk surprises before submission day becomes a small opera of tabs, passwords, and regret.
Last reviewed: 2026-07
