Translating a document usually creates a new file, so the original document properties — author, title, subject, keywords, company, custom fields, and revision history — are not guaranteed to carry over. Some are dropped, some are regenerated by the translation tool, and some (like tracked-change history or hidden custom fields) can behave unpredictably. For most professional work this is fine or even desirable, but if your workflow depends on specific metadata — matter numbers, document IDs, retention tags, or classification labels — you need a translation approach that lets you control what happens to it.
Bluente is an AI-powered document translation platform used by 30,000+ professionals to translate files in 120+ languages while preserving the original formatting. This article explains what document metadata is, what typically happens to it during translation, and how to keep the fields that matter without leaking the ones that shouldn't travel.
What Counts as Document Metadata?
Document metadata is the information stored inside a file that isn't part of the visible body text. In a DOCX, PPTX, or XLSX file, that includes standard properties like author, title, subject, manager, company, category, and keywords, plus system-generated data like creation date, last-modified date, revision count, total editing time, and the template the file was based on. It also includes custom properties that organizations add — a matter ID, a document classification, a retention code, or a workflow status.
PDFs carry their own metadata set: title, author, subject, keywords, producer, creation and modification timestamps, and XMP data. None of this appears when you read the document, but downstream systems — document management platforms, e-discovery tools, records retention systems — rely on it heavily. That's why "what happens to metadata during translation" is a real question, not a trivial one.
What Happens to Metadata When You Translate a File?
It varies by tool, and that variability is the point. Because translation produces a new output file, the tool decides which properties to copy, which to blank, and which to set fresh. A text-first tool that dumps translated words into a blank document loses essentially all of the original metadata. A document-first tool that rebuilds the file from the original preserves far more of the structure, but may still reset system fields like creation date (the translated file was, after all, just created) and author (it may be attributed to the service or the uploading user).
Custom fields are the most fragile. A matter number stored as a custom document property, or a classification label injected by a data-loss-prevention system, may or may not survive depending on how the tool reconstructs the file. The safe assumption is that unless a translation platform explicitly preserves custom properties, you should verify them on the output rather than assume they carried across.
Why Does Metadata Matter for Legal and Regulated Documents?
Because metadata can be evidence, a control, or a leak — sometimes all three. In e-discovery, document metadata (authorship, dates, revision history) is often discoverable and can be material to a case, so silently altering or stripping it during translation can create problems. In records management, retention and classification tags stored as metadata drive how long a document is kept and who can see it; lose the tag in translation and the file falls out of its retention policy.
There's also a confidentiality angle that cuts the other way. Metadata frequently contains information the author never meant to share — the real author's name, the internal template path, edit times, or comments and tracked changes buried in the file. When you translate a document that's headed outside your organization, you may actively want to strip that metadata, not preserve it. The right behavior depends on the direction the document is traveling.
How Do You Control Metadata During Translation?
Decide, per document, whether metadata should be preserved or cleaned, then use a platform that respects that intent and handles the file securely. For internal documents where matter IDs and classification tags must persist, you want a document-first engine that reconstructs the file faithfully and keeps custom properties. For outbound documents, you want the confidence that hidden metadata, comments, and revision history aren't riding along unnoticed.
Underneath both cases is security. The reason to avoid free online translators for anything sensitive isn't only formatting — it's that you can't verify what they do with your file or its metadata. Bluente operates with zero data retention, automatic deletion within 24 hours, and end-to-end encryption, and never uses your documents to train AI models. It's SOC 2 Type II, GDPR, and ISO 27001 compliant, so the whole document — body and metadata — stays inside a workflow your security team can approve. And because it's document-first, the visible formatting (tables, footnotes, legal numbering, layout) comes back intact regardless of how you handle the metadata.
What Should You Check on a Translated File?
Before you send or file a translated document, verify three things. First, the properties you rely on — matter number, document ID, classification — are present and correct on the output. Second, the properties you don't want to travel — original author, internal comments, tracked changes, template paths — are gone if the document is leaving your organization. Third, the visible content and formatting match the original exactly. Two minutes of verification prevents a retention-tag mismatch or an accidental metadata leak from becoming a bigger problem later.
Frequently Asked Questions
Q: Does translation remove the author and creation date from a document? Often, yes. Because translation creates a new file, system fields like author and creation date are typically reset rather than copied. If you need the original values preserved, use a document-first platform that lets you retain document properties, and verify them on the output.
Q: Will custom document properties like a matter number survive translation? Not automatically. Custom fields are the most fragile metadata during translation. Confirm on the translated file rather than assuming they carried over, and choose a tool that explicitly preserves custom properties if they're critical to your workflow.
Q: Can translating a document accidentally leak hidden metadata? The risk is usually the reverse — hidden metadata, comments, and tracked changes can travel with a file if the tool copies the original faithfully. For outbound documents, check that this information is stripped before the translated file leaves your organization.
Q: Is metadata discoverable in litigation? Frequently. Authorship, dates, and revision history are often discoverable and can be material in e-discovery. Altering or stripping metadata during translation without a clear process can create issues, so handle translation of litigation documents through a controlled, secure workflow.
Q: How does Bluente handle document metadata securely? Bluente is document-first, so it preserves the file's structure and visible formatting, and it operates with zero data retention, 24-hour auto-deletion, and end-to-end encryption. Your documents and their metadata are never used to train AI models, and the platform is SOC 2, GDPR, and ISO 27001 compliant.
Start translating documents for free. Bluente preserves your formatting across 120+ languages in under 2 minutes. Try BluTranslate free — no credit card required.