How to anonymise a Word document without leaving data behind

Where personal data hides in a Word file

Before changing anything, list what needs checking. These are the elements that most often leak information when documents are disclosed under a subject access request, released under FOI or exchanged in litigation:

Two classic traps: highlighting a name in black does not remove it (the text is still there and can be copied), and a picture cropped in Word keeps the cropped part unless you explicitly delete cropped areas when compressing pictures. The same goes for information you never see on the page: a file's metadata travels with it whether or not anyone opens the properties.

The manual method: Inspect Document in Word

Word's Document Inspector (File > Info > Check for Issues > Inspect Document) finds comments, revisions, versions, document properties and personal information, headers, footers and watermarks, hidden text, custom XML data and invisible content, and lets you remove each category. It is worth running every time as a final pass, but it has clear limits:

• It does not anonymise the body text. Names, NI numbers, addresses and phone numbers in the visible content stay put; you have to find and replace them yourself.
• It removes whole blocks, not data. Remove headers and footers and you also lose the logo or page numbering you wanted to keep.
• Find and Replace misses variants. “Sarah Collins”, “Ms Collins”, “S. Collins” and “the claimant” are four different searches, and an NI number may be written with or without spaces.
• It does not look inside embedded objects. An Excel sheet pasted as an object keeps every row, including those not shown on the page.
• Some removals cannot be undone, so always work on a copy.

For a short document with no history, the manual method may be enough. For long reports, batches of case files or documents that have passed through many hands, the risk of missing a mention grows with every page.

Cover or replace? Anonymisation versus pseudonymisation

If the document is going to be published or leave your organisation for good, you need anonymisation: removing the data irreversibly. If you still need to know that “Person 1” in chapter three is the same person as in the appendix, for internal analysis or for the solicitor handling the case, pseudonymisation with consistent substitutes is the better fit. Under the UK GDPR and the EU GDPR alike, only truly anonymous information falls outside data protection law; pseudonymised data is still personal data and must be protected as such.

In a Word file that must stay editable, black boxes make little sense: the recipient can simply move the text. The consistent approach is to replace each item with a placeholder or a fictitious value and then confirm that no trace remains anywhere in the file.

How to do it in Anonimatum and keep the file editable

Anonimatum accepts .doc, .docx and .odt files. Since version 1.9.0, .docx and .odt files can be downloaded as editable Word: the document keeps its formatting and the data is replaced in headers, footers, notes and tables too; comments are removed, tracked changes are accepted and metadata is stripped. If any detected data remained in the file after processing, the Word file is not delivered and the anonymised PDF is offered instead.

• Detection combines Anonimatum's own AI, run on servers in the EU, with patterns for email addresses, phone numbers or IBANs; for identifiers such as NI numbers or SSNs you can add a custom pattern with a regular expression.
• Before anything is applied, the preview highlights every match so you can drop false positives or mark anything that was missed.
• The whitelist keeps terms that must stay visible, such as your organisation's name or a public office.
• The Template selector applies the usual options for a document type in one go, and you can save your own for the whole group.

Final checks before sharing

If the final version will circulate as a PDF, our guide to anonymising a PDF covers the checks that are specific to that format.