Word to Markdown Converter

The one thing that decides how well your document converts

A .docx is a zip file full of XML, and inside it every paragraph carries two separate ideas: how it looks, and what it is. A line can be 18pt, bold and dark blue and still be, as far as the file is concerned, an ordinary body paragraph that someone made large. A different line can look identical while being marked Heading 1 — a named style that says what the line is for.

This converter reads the second thing. Styles, not formatting. That single fact explains the two experiences people have with Word-to-Markdown tools. A document written with the style gallery — Heading 1, Heading 2, Quote, List Paragraph — converts almost perfectly, because every one of those names has an obvious Markdown counterpart. A document formatted by hand, where headings were made by selecting text and clicking bold and bumping the size, comes out as an undifferentiated wall of **text**. Nothing is broken; there was simply no structure in the file to find.

In Word                          As Markdown
─────────────────────────────    ─────────────────────────
Quarterly Review  (Heading 1)    # Quarterly Review
Revenue           (Heading 2)    ## Revenue
Sales grew 12% …  (Normal)       Sales grew **12%** …
Growth was broad  (18pt bold)    **Growth was broad**

The last line is the trap. It looked like a heading in Word and it does not become one here.

Thirty seconds in Word that fix it

If the output is flat, go back to the document, select each heading, and click the matching style under Home → Styles — Heading 1 for the title, Heading 2 for sections, Heading 3 for subsections. Use the list buttons for lists rather than typing dashes and numbers yourself, and the Quote style for pull quotes. Then convert again. The improvement is usually dramatic, and it costs less time than tidying the Markdown by hand afterwards.

What comes through, and what cannot

Everything Markdown has a syntax for is kept: headings one through six, bold, italic and strikethrough, ordered and unordered lists including nested ones, hyperlinks with their text, blockquotes, inline code and code blocks where a code style was applied, horizontal rules, and tables — as long as the table is a plain grid. Merged cells and tables inside tables have no Markdown equivalent, so flatten those first.

The rest is dropped, and it is worth knowing which rest. Fonts, sizes, colours and highlighting go, because Markdown deliberately has no opinion about typography. So do columns, text boxes, headers and footers, page numbers and page breaks, which are all properties of a printed page rather than of a document. Comments and tracked changes are resolved away, so accept or reject them in Word before you convert. Footnotes and endnotes are left out too — Markdown footnotes exist, but only as an extension that many renderers do not support, so quietly emitting them would break as often as it helped.

Images, and why they are off by default

Markdown cannot contain a picture. It can only point at one, with ![alt](path). A .docx, meanwhile, has the images packed inside the zip, so there is no path to point at once the file is open in your browser. The only way to keep them without a server is to inline each image as a base64 data URI, which works everywhere but is expensive: base64 adds about a third to the size, so a document with a few photographs turns into a Markdown file of several megabytes that some editors will refuse to syntax-highlight. That is why the option is off unless you tick it. For a document with a handful of important diagrams, embedding is fine. For a report full of screenshots, it is usually better to convert the text, then save the images separately and link them.

.doc, .docx, and everything else

Only .docx works here. The older .doc is a binary format from an era before anyone expected other software to read Word files, and unpicking it in a browser is not practical. Open it in Word and use File → Save As to save a .docx copy — the conversion is lossless and takes a moment. Pages, LibreOffice Writer and Google Docs all export .docx, and all of them apply real heading styles, so their output converts cleanly. If your source is a PDF rather than a Word file, use the PDF to Markdown converter instead; a PDF has no styles at all, only positioned glyphs, which is a different and harder problem.

After the conversion

Most people are converting because the Markdown is going somewhere that wants plain text: a repository README, a static site, a wiki, a documentation build, or the context window of a model that reads structure far better than it reads Word XML. Skim the result once before you commit it — check the heading levels descend sensibly and no table lost a column — and it is done. If you need to go back the other way later, the Markdown to Word converter rebuilds a styled .docx, and Markdown to Google Docs puts formatted text straight onto your clipboard.

The other common destination is a notes app, and that is where converting starts to feel like a chore you repeat. Bulguritto stores every note as a plain Markdown file — not a database that exports Markdown on request — and syncs those files across iPhone, iPad and Mac. Paste the output of this converter in and it stays a real document you can read with anything else, which means the next time you need Markdown, there is nothing to convert. It is free.