Convert DOCX to RTF

This conversion is almost never a preference. It is a requirement arriving from somewhere else: a court filing portal that lists rich text and nothing newer, a transcription or dictation platform, a case management system whose import routine was written in the 1990s, a mail-merge engine that assembles letters by splicing text into a template, an e-discovery process that normalises everything it receives.

The reason those systems chose RTF is worth understanding, because it explains why they will not change. RTF can be produced and consumed by a program that knows nothing about ZIP archives, XML schemas or office suites — it is characters and control words. That made it the safe common denominator for a generation of software, and the code that depends on it long outlived the fashion. Microsoft stopped developing the specification in 2008, which in this context is a feature rather than neglect: a format that never changes again is a format an intake system can be written against once and left alone.

Open one in a text editor and you can read the document. The text is there in plain sight, wrapped in backslash-prefixed control words that describe the font table, the colour table, the style sheet and the formatting of each run. It is verbose, entirely printable, and repairable by hand in a way a binary format never is.

That transparency is the format’s whole personality. It is why an RTF survives being pasted into a system that mangles anything with a null byte, why a broken one can often be salvaged, and also why it carries none of the structural guarantees a modern format has. Nothing validates it. A file that is slightly wrong usually still opens, slightly wrong. That is worth knowing before you assume a clean render means a clean file: a DOCX with a malformed part is refused outright, and an RTF with the same problem quietly shows you a document missing its last table.

A DOCX is a ZIP archive, so its XML is compressed before it ever reaches disk. RTF has no compression whatsoever; every control word is stored literally. For a document that is mostly prose this typically means a file several times the size, which is unremarkable until it meets an attachment limit.

Pictures are the dramatic case. An RTF stores an embedded image as hexadecimal text, so each byte of the picture becomes two characters in the file. A five-megabyte photograph does not arrive as five megabytes of RTF. If the document is illustrated and the destination has a size cap, reduce the images in the Word file before converting rather than after. The other lever is to remove pictures that are decorative rather than substantive — a letterhead logo repeated on every page, a stock photograph on the cover — since a filing system is interested in the text and the cost of the graphics is paid in full on every submission.

More than its age suggests. RTF has a style sheet, so headings and named paragraph styles come across as styles rather than as manual formatting. Tables keep their cells, borders and merged regions. Lists keep their numbering. Headers, footers, page numbering, footnotes, hyperlinks, bookmarks and page setup all have direct representations, as do character-level attributes down to kerning and colour.

The conversion runs through LibreOffice, which reads DOCX and writes RTF as first-class formats rather than as an afterthought, and the machine it runs on carries metric-compatible substitutes for Arial, Times New Roman, Courier New, Calibri and Cambria. That keeps line breaks in the same place, which matters when the receiving system paginates the document itself. What RTF cannot do is carry the fonts with the file. It names them and nothing more, so the machine that eventually opens the submission substitutes anything it does not hold — one more reason to draft in the ordinary faces rather than the ones on the brand guidelines.

Everything Word added after RTF stopped being developed. Content controls and structured document tags become ordinary text. SmartArt arrives as a picture, if it arrives at all. Document themes collapse into the concrete fonts and colours they were resolving to. Embedded fonts do not travel, because RTF has no mechanism to carry them.

This is worth checking rather than assuming, because the failure is silent: a content control that turns into plain text still looks correct on screen and no longer does anything. If the DOCX was generated from a template with fillable regions, the RTF is a snapshot of that template’s current contents and not a form. The same applies to any field that was still calculating — a date that updated itself, a page count, a cross-reference. Each of those resolves to the value it held at the moment of conversion and stops being live, which is usually what a filing system wants and never what a template author intended.

The RTF syntax has no place to put a VBA project, so a document’s automation is left behind by definition rather than by policy. The words, the tables and the layout arrive; the code does not.

Several of the systems that mandate RTF do so for exactly this reason. A filing system that accepts inert documents and rejects containers that can execute anything has removed a whole category of problem from its intake, and it will not be persuaded that your particular Word file is harmless. Converting is the answer that respects the requirement instead of arguing with it.

RTF does have representations for revision marks and annotations, and simple cases come across. It is markedly less reliable here than a modern format, though, because the two models are different ages and the mapping is being made across a wide gap.

For anything with an audit trail attached to it, resolve the outstanding changes first and submit a clean document. If the tracked changes themselves are the deliverable — a redline being filed as evidence of what was amended — verify the result in the program that will read it, not in the one that produced it. That is the only check that means anything.

Word, LibreOffice Writer, OpenOffice, WordPad on Windows and TextEdit on macOS all open RTF as a formatted document without any add-in. It is one of very few formats where that sentence is still true across every desktop platform, which is the other reason it hangs on.

It also means you can verify the output yourself before submitting it. Open the RTF, look at the first page, the tables and anything that used to be a form field, and compare against the DOCX side by side. Thirty seconds spent there is cheaper than a rejected filing and a resubmission window. Open it in whichever program most resembles the destination rather than the one you have to hand. A rich text file rendered by Word and the same file rendered by a system’s own parser can differ, and the parser is the one that decides whether the submission is accepted.

This runs on a server rather than in your browser. Converting between two full word-processing formats needs an office engine, and the page says so plainly instead of implying that everything here happens locally. The file goes over an encrypted connection to a container running LibreOffice headless.

That container cannot make outbound connections. Each job is given its own scratch directory and its own LibreOffice profile, and the directory is deleted when the job finishes regardless of the outcome; jobs still running after sixty seconds are terminated rather than left hanging. The free limit is 25 MB on the file you send.

If the receiving system is going to strip the formatting anyway — an ingestion pipeline that indexes words, a system that reflows everything into its own template — send plain text and stop paying for the markup. The result is smaller, unambiguous, and there is nothing left to render incorrectly.

If instead the document must look exactly as you laid it out and nobody should edit it, PDF is the correct target and RTF is a poor one. RTF is the right choice in a narrow band: the recipient needs an editable document, and their software will only take rich text. Outside that band, one of the other two is usually better.