How to Convert Word DOCX to PDF Free on Mobile and Desktop

What the converter actually produces

Understanding the pipeline saves you a surprise at the other end. The tool unzips the .docx, reads the document XML with Mammoth, and turns it into plain HTML. That HTML goes into a hidden, page-width element on the screen, which is then photographed by html2canvas at double resolution. The resulting image becomes the PDF.

So the output is a picture of your document. The consequence that matters is the text: in the finished PDF it is not selectable, searchable or copyable, because it is pixels. The pages themselves are real A4 pages. The hidden element is sliced into page-height windows, each one photographed separately, and the images are laid onto A4 pages with 0.75 inch margins, so a long document comes back as a proper stack rather than one tall sheet. If you need text a recruiter's software can read, this is still not the route to take.

A picture of a document is still the right answer for some jobs. A menu, a flyer, a signed form going into a file: nobody needs to search those, and nobody should be able to reflow them. An image holds the layout still on every device that opens it, which is the one thing a PDF is always asked to do.

Read the preview before you convert

When you add a file, the page shows you the parsed document above the convert button. That preview is not decoration. It is exactly the HTML that gets photographed, so whatever is missing from the preview will be missing from the PDF. Check it first. Thirty seconds there beats downloading a PDF and discovering your table of contents did not come along.

Tables are the usual surprise. They come across as tables, with the right cells in the right order, but none of the borders, shading or column widths you set in Word survive the trip. A table that carried its meaning in its formatting comes back considerably plainer than you left it.

What survives, and what does not

Mammoth is deliberately conservative. It keeps document structure and drops presentation:

• Headings, paragraphs, bold and italic, bulleted and numbered lists, tables, hyperlinks and embedded images all come through.
• Your fonts do not. The hidden container renders everything in Arial at one size, so a document set in Calibri, Garamond or a corporate typeface comes back in Arial. Point sizes and colours set directly on the text are not carried over either.
• Headers, footers and page numbers are ignored, because Mammoth never reads those parts of the file.
• Manual page breaks do carry over: a break you inserted in Word starts a new page in the PDF. Section breaks, columns, text boxes, comments and tracked changes do not appear.
• It reads .docx only. The picker will let you choose an old binary .doc, and the conversion then fails, so save it as .docx in any word processor first.

The fallback, and why text sometimes comes back with question marks

If the visual render throws an error, the code quietly falls back to a second method: it pulls the raw text out of the document and lays it out with pdf-lib on real A4 pages. That version is genuinely paginated and its text is genuinely selectable, which is better in several ways. The catch is the font. It uses the standard Helvetica built into every PDF reader, which only covers the WinAnsi character set, so the text is scrubbed first. Smart quotes become straight ones, dashes become hyphens, and anything outside that character set, including Greek, Cyrillic, Chinese, Japanese, Devanagari and most emoji, is replaced by a question mark. A document that arrives full of question marks took the fallback path.

Doing it on a phone

The browser route works the same on iOS and Android as it does on a desktop, which is the point of it: no app store, no account, nothing installed. Each page is drawn into its own canvas, so the old problem of a long report coming back cut off is gone, but a phone still runs out of patience before a laptop does: every page is rendered at double resolution, and a fifty page report is fifty of those renders. Short documents, a CV, a cover letter, a two page invoice, are what this handles best on a phone. It takes one file at a time, up to 100 MB, which no ordinary .docx gets anywhere near.

When the built-in export is better

Word on iOS and Android exports to PDF from its own share menu, Google Docs does it from Download, and Pages on a Mac does it from Export. All of them keep the real fonts, the real page breaks and the real text layer, because they lay the document out properly rather than photographing it. Use one of those whenever the PDF has to be searched, printed at exact size, read aloud, or parsed by an applicant tracking system. Use the browser converter when you have a .docx, no software that opens it, and you need a fixed picture of it in the next minute.