DOCX or PDF: which source converts better, and why
Last reviewed
If you have the same document as both a Word file and a PDF, convert the Word file. That is the whole recommendation, and it holds nearly without exception.
The reasoning is worth knowing anyway, because it explains what the PDF path is fighting, and because the handful of exceptions are real.
7 minute read
The difference in one sentence
A Word file states its structure. A PDF implies it.
That is not a difference of degree. When a document says that this paragraph is a level-two heading, converting it is a translation: read the label, emit two hashes. When a document says only that these glyphs are set at fourteen points at these coordinates, converting it is an inference: measure the height, compare it against the document's usual height, decide whether the ratio crosses a threshold, and accept that documents styling headings by weight rather than size will fool you.
Every conversion error specific to PDFs traces back to that sentence.
What a Word file says outright
A DOCX file is a ZIP containing XML, and the XML carries the structure as named properties rather than as visual side-effects.
| Headings | A named paragraph style, with its level |
|---|---|
| Lists | A numbering reference, stating ordered or bulleted and the depth |
| Tables | A table element with real rows and cells |
| Emphasis | Explicit bold and italic properties on the run |
| Quotations | A quote paragraph style |
| Code | Inferred from a monospace font, since Word has no code style |
Where the two paths genuinely differ
The practical consequences are concentrated in the places where PDF inference is weakest.
- Nested lists come out right. Depth is stated rather than derived from left-edge measurements, so a five-level outline survives intact instead of being flattened at four.
- Merged table cells stop being fatal. The table is a real structure, so a cell spanning two columns is a fact rather than a broken alignment pattern.
- Headings styled by weight are found. A bold eleven-point heading over eleven-point body text is invisible to a size ratio and completely visible to a paragraph style.
- Multi-column layout is irrelevant. Reading order is the order of the document, not something recovered from a gutter.
- Running headers and footers never appear, because they live in separate parts of the file and are never part of the text flow to begin with.
Where DOCX is still lossy
It converts better, not perfectly, and the losses are worth stating plainly.
Images are not carried. Markdown has no way to embed the bytes and this converter will not upload them anywhere, so figures are counted and the count is reported rather than being dropped in silence. A Word document that is mostly diagrams converts into a document that is mostly captions.
Colour, fonts, spacing and every other visual property are gone, as they are in any Markdown conversion. Content and structure survive; presentation does not.
Tracked changes and comments are not the document, and are not included.
Why a Word file has no page count
A converted DOCX reports no page numbers, and that is a deliberate refusal rather than a missing feature.
Word stores no pagination. The document is repaginated at render time against the current font metrics, page size and printer driver, which is why the same file is twelve pages on one machine and fourteen on another. There is no page count in the file to report.
So where the PDF path counts pages, the DOCX path counts words, and any limit is expressed at an assumed five hundred words to the page. The limit is therefore an estimate, which is fine. Printing an estimate as if it were a page number would not be.
The .doc trap
The older binary format, with no x on the end, is a completely different thing: a compound binary document from the nineteen-nineties, not a ZIP of XML. It cannot be read by an OOXML reader and renaming it does not help, so it is rejected by name with a message saying to save it as .docx first.
There is one genuinely confusing case. A DOCX encrypted with a password is wrapped in that same old binary container, so its first bytes look exactly like a legacy .doc file. The only thing distinguishing them is the filename. If you are told a .docx file looks like a legacy document, the likeliest explanation is that it is password-protected — remove the password in Word and convert again.
When the PDF is the better source
Three cases, and they are narrower than people expect.
- The PDF is the document of record. A signed contract, a filed submission or a published paper is the artefact that matters, and the Word file is a draft that may differ from it.
- The Word file is a template and the PDF is the filled version. Mail-merged and form-generated documents are common, and the source often contains field codes rather than content.
- The document was assembled from several sources. A PDF combining an exported report, a scanned appendix and a signature page has no single Word original, and the parts are best handled separately anyway.