People expect document conversion to be lossless — put a DOCX in, get an identical PDF out. When the layout shifts or a table collapses, it feels like a bug. It usually isn't. The formats simply store a document at different levels of abstraction, and conversion can only preserve what the destination format is able to express.
Three formats, three mental models
PDF fixes the final picture. A PDF describes exactly where every glyph sits on the page. It's designed so a document looks identical everywhere, which is perfect for printing and archiving and terrible for editing. Once text is placed, it isn't meant to reflow.
DOCX stores editable structure. A Word document keeps text as reflowable content plus a layer of styles: headings, paragraph spacing, fonts, tables. Nothing is pinned to a fixed position — the layout is computed when you open it. That's why the same DOCX can look slightly different across Word versions or on different machines.
Markdown stores structure only. Markdown records that something is a heading, a list, or a link — and nothing about how it looks. No fonts, no colors, no page geometry. That minimalism is its strength: it's readable as plain text, diffs cleanly in version control, and renders anywhere.
What conversion actually does
Converting is translating between these models, and translation is lossy whenever the target is less expressive than the source.
- DOCX to PDF goes from "editable structure" to "fixed picture." The converter must make final decisions — which font, where the page breaks, how much spacing. If a font isn't embedded, it substitutes one, and the layout moves.
- DOCX to Markdown goes from "styled structure" to "plain structure." Headings and lists survive; fonts, colors, and complex tables don't. That's the right trade when you want content you can publish or version, not a faithful visual copy.
- Markdown to PDF goes the other way — plain structure gets a visual treatment applied by a stylesheet. The result looks polished, but the styling comes from the converter's template, not from your source.
You can see this firsthand with the document converter: convert a styled document to Markdown and back, and watch which attributes survive the round trip.
How to get predictable results
Once you see conversion as translation, the practical rules follow:
- Match the format to the goal. Final, shareable, must-look-identical? PDF. Still editing? DOCX. Publishing to the web or storing in Git? Markdown.
- Embed fonts before making a PDF so the renderer isn't forced to substitute.
- Expect styling loss when moving toward Markdown — and treat it as a feature, since it strips away everything that isn't content.
- Convert once, at the end. Every round trip through a lossy step compounds the drift.
The goal isn't a perfect copy in every direction — that's impossible when the formats disagree about what a document even is. The goal is to know, before you convert, exactly what you're keeping and what you're letting go.
