Redacting documents correctly — step by step, DSGVO-compliant

Anyone who wants to redact a document usually means: black bar over it and done. That is precisely the mistake. A document is only correctly redacted when the text is actually gone, not just masked. This article shows what "redacting correctly" means, where the most common mistakes lie and how to implement it step by step.

Short answer: Redacting documents correctly means permanently removing selected content from the PDF — not just visually masking it with a black bar. This prevents simple reconstruction of confirmed findings; the legal basis, scope and result must nevertheless be reviewed.

What "redacting correctly" really means

Redaction is not the same as redaction. A black rectangle over a word looks like redaction on screen, but is only a visual overlay. The text underneath remains in the PDF and can be restored with a text editor or by copying. Correctly redacted means: the text is physically removed from the document, so that it is neither legible nor reconstructable. That is exactly what the DSGVO requires when personal data is no longer needed.

The most common mistakes when redacting

• Bars instead of removal: The text remains copyable. A sample of publicly indexed PDFs found that in roughly every sixth document text remained selectable under the redaction.
• Forgotten metadata: Comments, bookmarks and PDF properties often still contain names or IBANs. In the AstraZeneca report, the bookmarks contained the redacted text in plain text.
• Redacting too much: Anyone who makes entire pages illegible risks authorities or opposing parties rejecting the document.
• Not checking: The redaction should be tested before sending. The simplest test: select the spot and try to copy the text.

According to the guidance of the state data protection commissioners — such as the Lower Saxony Data Protection Commissioner (LfD Niedersachsen), the Saxon Data Protection Commissioner (SächsDSB) and the Federal Commissioner for Data Protection and Freedom of Information (BfDI) — visual masking alone is not sufficient. The authorities repeatedly make clear in their guidance on file inspection and IFG access requests that the text must be physically removed from the document, otherwise the redaction is not DSGVO-compliant.

Step by step: redacting documents correctly

1. Upload document (PDF; scanned pages via OCR).
2. Start automatic detection of personal data — names, IBANs, email addresses, telephone numbers, addresses, social security numbers.
3. Review each suggestion and confirm only what is not needed for the respective purpose.
4. Also sanitise metadata, comments and bookmarks.
5. Test the result: select the redacted spot — no text should appear.
6. Save the audit log, in case someone later asks what was redacted and why.

Which documents are redacted particularly often

Bank statements for the Jobcenter, invoices before sharing, personnel files during file inspection, contracts with IBANs, authority mail in IFG proceedings. The same principle applies everywhere: process only the data that is really necessary for the purpose, and irreversibly remove the rest.

Tool instead of manual work

Redacting manually is not only slow, but error-prone. Automatic detection followed by human review can accelerate the process. Datenmaske combines rule-based patterns (regex for IBAN, email, telephone) with name recognition (NER), permanently removes confirmed findings from the PDF and sanitises metadata. For OCR and an optional NER fallback stage, Azure can be used in configured EU regions; details are set out in the privacy policy and the DPA.

In short: redacting correctly means removing, not covering up. Anyone who adheres to this avoids data leaks and DSGVO fines and, with the audit log, also has the evidence in hand. You can redact a PDF free of charge — with the Datenmaske Free-Check.