PDF converterWord converterNo document open
Drop a Word document here
.docx only, up to 25 MB. A legacy .doc has to be re-saved as .docx first.The Markdown appears here block by block. Select one to see where it sits in the document.
- KeptConverted with a direct Markdown equivalent.
- DowngradedConverted, but with less structure than the original.
- DroppedNo Markdown equivalent exists — these need a hand pass.
Structures this audit counts
- Merged table cells
- Equations
- Tracked changes
- Comments
- Text boxes
- Headers and footers
- Images kept as references
Convert a .docx file and the Word to Markdown audit appears here, listing every structure that was kept, downgraded, or dropped.
Want proof? Disconnect from the network and convert again. It still works, because there is nothing to send.
Why the conversion is lossyWord to Markdown converter
Convert a .docx file to clean Markdown in your browser, then read an audit of what the conversion kept, downgraded, and dropped. The file never leaves your device.
Why every Word to Markdown conversion loses something
A .docx file is a zip of XML that describes a page: styles, revision history, floating shapes, page furniture, and the numbering scheme Word will render. Markdown describes a document: headings, paragraphs, lists, tables, links, code. The two overlap in the middle and diverge at the edges, and everything at the edges has to go somewhere.
Most converters resolve that by flattening the .docx into HTML first and then into Markdown. It is a reasonable pipeline, but by the time the HTML exists the evidence is gone: a merged cell is now three ordinary cells, a tracked deletion is now plain text, an equation is now nothing at all. Nothing downstream can warn you, so the output looks complete.
This converter reads the OOXML parts directly and counts those structures as it walks them, which is why the audit can name them. The conversion is still lossy — that part is unavoidable — but the loss is itemised instead of silent.
| In Word | In Markdown | What happens |
|---|---|---|
| Heading 1–6 styles | # to ###### | Converted directly, including Chinese 标题 styles |
| Bulleted and numbered lists | - and 1. | Nesting kept; letter and roman numbering becomes plain numbers |
| Tables | GFM pipe tables | Reliable until a cell is merged or a table is nested |
| Merged cells | no equivalent | Content stays in the first cell of the span, the rest are padded |
| Equations (OMML) | no equivalent | Counted and reported for you to rewrite as LaTeX |
| Comments and tracked changes | no equivalent | Insertions accepted, deletions and comments removed |
| Headers, footers, text boxes | no equivalent | They belong to the page, not to the document flow |
Five things to check before you trust the Markdown
Two minutes of checking catches almost every Word to Markdown conversion problem that people otherwise discover a week later, after the content is already in a wiki.
Open the tables first
Merged cells are the single most common failure. Compare the row count and the header row against Word. If the audit lists merged cells, those rows need either a manual pass or an HTML table block inside the Markdown.
Search for three numbers
Pick three figures from the middle of the document and search for them in the output. A missing figure usually means the value lived in a text box or a header rather than in the body text.
Look for the equations
If the audit counted equations, they are not in the Markdown. Rewrite them as LaTeX now, while you still remember which paragraph each one belonged to.
Confirm the file is final
Tracked changes are accepted silently during conversion. If the .docx still carries revisions, resolve them in Word first so the Markdown matches the version you meant to publish.
Decide what images should be
Pictures become files in a media/ folder and the Markdown links to them, so use the .md + images download to keep those links working. Switch to base64 only when you need one portable file, and to Remove when the pictures are decoration.
Where the Markdown goes next
Converting is half the job. The export profile decides whether the file is actually usable in the place you are taking it.
- Clean Markdown
- Body text only, no extra markers. Best for pasting into a CMS, an issue, or a pull request.
- Obsidian note
- Frontmatter, a table of contents, and block ids, so a converted requirements document becomes a linkable note.
- GitHub / MkDocs
- Frontmatter, stable heading anchors, and source comments for a docs site other people will edit.
- RAG chunks
- JSONL chunks that carry their heading context, for feeding a Word document into a retrieval pipeline without losing its structure.
All four run on the same local conversion. Coordinate-level source maps stay a PDF feature — a .docx has no page coordinates to map back to. See every output profile
Word or PDF — which one should you convert?
If you have both, convert the .docx. A Word to Markdown conversion starts from the document's real structure: heading styles, list numbering, table cells, footnote links. A PDF exported from that same file has only glyphs at coordinates, and every piece of structure has to be inferred back from position and font size.
The exception is a document that only exists as a PDF, or one where the layout is the content: a scanned contract, a two-column paper, a financial statement with tables that continue across pages.
Word to Markdown FAQ
Is my Word document uploaded anywhere?
No. The Word to Markdown conversion runs in your browser: the .docx is unzipped, parsed, and turned into Markdown locally. You can disconnect from the network and it still works. There is no account, no queue, and no server-side copy of the file.
How is this different from converting Word to Markdown with Pandoc?
Pandoc is excellent and produces comparable Markdown, but it runs on the command line and says nothing about what it discarded. This page covers the step before that: convert, read the audit, and find out which parts need a hand pass.
Can it convert .doc files?
No. The legacy binary .doc format is not a zip of XML and cannot be read this way. Open it in Word, LibreOffice, or Pages and save it as .docx first.
What happens to images in the Word file?
By default each picture is written out as a file under media/ and the Markdown links to it, which keeps the text readable — a single base64 photo is tens of thousands of characters that bury the writing and break diffs. Use the .md + images download to take both, or switch Pictures to base64 for one self-contained file, or to Remove for text only.
Does it keep tables?
Yes, as GitHub-flavored Markdown tables. Markdown tables have no merged cells, so a merged cell is padded out and listed as a downgrade in the audit rather than quietly reshaping your data.
Why does the audit say equations were dropped?
Word stores equations as OMML, a Microsoft XML dialect with no Markdown equivalent. Rather than emit a placeholder that reads like content, the converter reports how many there were so you can rewrite them as LaTeX.
Convert a Word document now
Free, local, and no account. This Word to Markdown converter treats the audit as part of the free conversion, not an upgrade.