Convert DOC to DOCX

DOCX arrived in 2007 and the binary .doc it replaced is the Word 97 format. Nineteen years on, the old format still arrives daily, and always for the same reasons: an archive that was never migrated, a shared drive with a decade of minutes and policies on it, a template a small firm has been reusing since before the change, a correspondent on an ancient copy of Word who has no intention of buying a new one.

None of those files are broken. They are written in a container whose support is being withdrawn a little further with every Office release, and the sensible time to move them is while there is still a program that reads both ends comfortably. That program is LibreOffice, and it is what runs this conversion. The window is not closing tomorrow, but it is narrowing: each release of Office is a little less willing to open the old container, and the institutional knowledge of what these files contain leaves with the people who wrote them.

A .doc is a compound binary file: a small filesystem inside a single file, with streams for the text, the formatting tables, the embedded objects and the document summary. It was designed to be written to a floppy disk quickly, not to be read by anything except Word, and its structure is why a single damaged region can make the whole document unopenable rather than losing one page.

DOCX is the opposite arrangement. It is an ordinary ZIP archive holding XML parts and the media files, which is why renaming one to .zip and opening it shows you the document’s text as readable markup. Practically, that means a corrupted DOCX often loses one part rather than everything, the text can be searched by tools that know nothing about Word, and version control can produce a meaningful diff. It also means a document management system can index the contents without licensing anything, which is usually the real reason an organisation wants the archive converted rather than merely kept.

Word used to offer a setting that made saving faster by appending the changes to the end of the file instead of rewriting it. The consequence was that text you had deleted could remain in the file, invisible in Word and perfectly readable to anything that looked at the raw bytes. Several well-known disclosure incidents were exactly this, and the setting was removed from Word in the same release that introduced DOCX.

Converting rewrites the document from what it currently is, so the residue does not come across. If the .doc in front of you is an old contract, a redacted report or a proposal that was edited down before it went out, that is a reason to convert it before forwarding it to anybody — and a reason not to forward the original. Converting is not a substitute for a proper redaction process, because anything still visible in the document is still visible afterwards. What it removes is the material Word itself no longer shows you, which is the part nobody thinks to check.

Modern Office treats the binary formats as a security surface, because they were for many years the most reliable way to deliver a malicious document. Files that arrive from outside open in a restricted mode, and the very oldest Word variants are blocked by default in the Trust Center rather than merely warned about — which is why a document from the mid-1990s sometimes will not open at all.

Teaching a team to click past those warnings is the wrong fix. Converting the archive removes the reason the warning exists. It also removes the compatibility banner across the top of the window, which sounds cosmetic until you count how many people have been trained by it to ignore the message bar entirely.

For anything document-shaped, effectively all of it. Body text and headings keep their styles, tables keep their rows, merged cells and borders, images stay where they were anchored, headers, footers and page numbering carry over, footnotes and endnotes stay attached, and section breaks with their different page setups survive as sections.

LibreOffice reads the old format directly rather than approximating it, which is what makes this a repackaging rather than an interpretation. The areas that do shift are the ones Word itself reimplemented between the two generations: the older drawing layer, floating text frames, WordArt of the era and a few field types that no longer exist. A plain report will look identical; a 1998 newsletter with layered graphics may need a look. Fonts are handled by substitution, and the machine doing the work carries metric clones of the common Microsoft faces — Liberation for Arial, Times New Roman and Courier New, Carlito for Calibri, Caladea for Cambria — so line breaks and page counts normally stay where they were.

This is not a limitation of the converter. The 2007 format split macro-bearing documents into their own extension on purpose, so that a file ending in .docx is guaranteed to contain no executable code. Handing somebody a .docx is therefore a stronger statement than handing them a .doc: it cannot run anything.

If the original document’s value was its automation — a form that calculates, a template that assembles a letter — the conversion gives you the content and not the behaviour, and rebuilding that behaviour is a person’s job. For the archive of ordinary documents this pair mostly serves, there was never a macro in the first place.

Because DOCX is a ZIP, the XML inside it compresses well and text-heavy documents commonly come out several times smaller than the .doc they were made from. Images do not compress further — a JPEG is already compressed — so a document that is mostly photographs will be roughly the same size either way.

On a folder of old files the difference is worth measuring before and after, particularly if the archive lives somewhere that charges for storage or has to be backed up over a slow line. It is also worth remembering that part of any dramatic saving may be the old file’s accumulated residue rather than clever compression.

This pair is almost never a single document. It is a directory: fifteen years of committee minutes, a firm’s letter templates, a researcher’s papers. Drop the folder and each file is converted in turn, keeps its filename with the extension replaced, and the whole set comes back as one ZIP that you can unpack over a copy of the original tree.

Do it over a copy, not the original. Convert, spot-check the largest and the most heavily formatted files rather than the first three alphabetically, and only then retire the sources. The files that break are never the ones you sampled; they are the one with the organisation chart drawn in Word 97 objects. A count of files in equals files out proves only that the engine did not crash. It says nothing at all about whether the twelfth document still has its table borders, and the count is what people mistake for verification.

This conversion runs on a server rather than in your browser, because reading a 1990s binary Word document takes a full office suite and there is no honest way to do it locally. The file is sent over an encrypted connection to a container that runs LibreOffice headless, converted there, and returned.

That container is not permitted to make outbound connections. Every job is given a scratch directory of its own, and the directory is deleted as soon as the job ends, successfully or not. Anything still running after sixty seconds is terminated, because a damaged binary document is more likely to send LibreOffice into a loop than to produce an error. The free limit is 25 MB per file, which is generous for text and easily reached by a scanned document.

Choose DOCX if the destination is still Microsoft Word and the point is that the file keeps working — the same document, the same program, a container with another twenty years in it. That is the ordinary case and it is what this page is for.

Choose ODT instead if you are leaving Word rather than staying with it, or if a policy requires an independently maintained standard for records. Choose PDF if the document is finished and nobody should be editing it again. All three are one conversion away from the same .doc, and the right one is decided by where the file is going, not by what it is. A records office frequently wants two of them: an editable master in whichever format the policy names, and a fixed rendition of the version that was approved. That costs one more pass over the folder and settles every later argument about which copy is the real one.