Convert HTML to DOCX

A page pinned in place becoming a document that flows

HTML positions things; a Word document flows them. That is the central difference and the source of every surprise here. A web page can pin a sidebar to the left and a panel to the right and let them sit there regardless of length; a .docx is a stream of paragraphs that push each other down the page. So layouts built with CSS positioning, flexbox columns or grids are written as a single flowing column in reading order — not because the layout could not be measured, but because pinning boxes into a Word document produces something nobody can edit, which defeats the purpose of converting at all.

What maps cleanly is everything that was already document-shaped. Headings become Word heading styles — so the navigation pane works and a table of contents can be generated — and paragraphs, bold and italic, lists, block quotes, tables with their borders, and hyperlinks all have direct equivalents. Images are embedded so the document is self-contained. A blog post, an article or a report converts almost perfectly; a landing page does not.

Images have to be reachable at conversion time. A path pointing at your own computer cannot be read — a page in a browser has no access to your filesystem — and a web address only works if that server allows other sites to fetch the file, which many deliberately do not. The fix is to embed pictures as data URIs, or to use addresses you know are publicly fetchable. Any image that could not be included is reported rather than silently missing.

Two smaller omissions worth expecting. Web fonts are not embedded, so text is set in whatever Word has and the look shifts; and CSS background images are not carried, since Word has no equivalent concept for a paragraph background picture. Scripts never run, so a page that builds itself with JavaScript converts as whatever its markup says beforehand.

When you want something else

Pandoc is better at this than any converter that works from rendered markup: pandoc page.html -o page.docx --extract-media=. maps HTML structure onto Word styles properly, pulls the images out, and accepts a reference document so the output matches your own template — headings, fonts and spacing all in house style rather than Word's defaults. That last part is what makes it worth installing.

For a live web page rather than a file, save it first — "Save Page As → Web Page, Complete" — or fetch it with curl and convert the result, since nothing here reaches out to the network. And if what you want is a faithful copy of how the page looks rather than something editable, converting to PDF is the right tool, because a PDF can pin boxes where a Word document cannot.