PDF to Word without losing formatting
Formatting survives in proportion to how the PDF was made. A PDF exported from Word converts back almost perfectly; a scanned page has no structure to recover and converts to an image or to OCR text with approximate layout.
"Without losing formatting" is the most searched phrase in this category and the least honest promise in it. No converter can guarantee it, because a PDF does not store paragraphs, tables or styles. It stores instructions for drawing glyphs at coordinates.
First, find out which kind of PDF you have
Click and drag across a sentence in a PDF viewer. If words highlight, the file has a text layer. If the cursor turns into a hand and selects a rectangle, it is an image scan.
| Type | What conversion gives you | Layout quality |
|---|---|---|
| Exported from Word or Google Docs | Real paragraphs, real tables | Usually near-identical |
| Exported from a design tool | Text boxes, one per block | Looks right, hard to edit |
| Scanned page (image) | An image, or OCR text | Requires OCR; layout approximate |
| Mixed (text + scans) | Both, page by page | Varies per page |
What almost always breaks
- Tables. The most fragile element. Cells become text boxes or lose their borders. Expect to rebuild any table with merged cells.
- Multi-column layouts. Text from column one and two gets interleaved into a single stream, which is why academic papers convert badly.
- Headers, footers and page numbers. They become body paragraphs, repeated in the middle of the text.
- Fonts that were not embedded. Substituted with a similar font, which changes line breaks and therefore pagination.
- Bullets and numbering. Often become literal characters rather than a list style, so editing the list does not renumber it.
Getting the best result
- Convert, do not rebuild. If the PDF came from Word, converting gives you something close to the original. Rebuilding by hand takes longer than fixing the conversion.
- Fix structure, not appearance. Apply real Heading styles and real lists. The document will then reflow correctly, which matters more than matching the PDF pixel for pixel.
- Rebuild tables last. They take the longest and depend on the text around them being right first.
If it is a scan
You need OCR, and OCR produces text with no formatting at all. The realistic workflow is: OCR to get the words, then rebuild the layout in Word. Any tool claiming to preserve layout from a scan is either embedding an image of the page or guessing generously.
One thing to check before uploading
A contract, a payslip or a medical letter is exactly what people convert. If the tool uploads the file, a copy now exists on a server with its own retention policy. Converting in the browser avoids that entirely, which for these documents is the more important feature.
Questions
Why does my converted document look wrong?
Almost always because the PDF was a scan, or because it was produced by a design tool where every text block is positioned absolutely. Both give the converter no paragraph structure to work with.
Can I keep the exact fonts?
Only if they were embedded in the PDF and the converter is allowed to extract them. Licensing often prevents that, so a substitute font is used and line breaks shift.
Is a scanned PDF editable after conversion?
Only the OCR text layer is editable. The original page image is not, so the result is text you reformat yourself.
Does this upload my document?
No. The conversion runs in the browser, which matters for contracts and anything confidential.
Updated 2026-11-16 · written for this site, no trackers, no account.