PDF to Word

Turning a PDF back into an editable document

A Word document knows it has paragraphs, headings and lists. A PDF doesn’t — it knows a particular character sits at a particular coordinate. The structure was discarded when the file was made, so converting back means working it out from the layout.

How the structure is rebuilt

filesmith.io/convert/pdf-to-docx

Characters sharing a baseline become a line, lines with consistent spacing become a paragraph, a line stopping short of the margin ends one, larger text becomes a heading, a leading bullet becomes a list item, and text lining up in consistent columns across several rows becomes a real table. A table split across pages is stitched back together when its header row repeats.

What works well, and what needs tidying

Prose, headings, lists, tables and images come across cleanly. Forms, heavily designed pages and complex multi-column layouts may need tidying afterwards — the text is all there, but the arrangement is inferred rather than known.

Scanned PDFs won’t work

If your PDF came from a scanner or a phone camera, the page is an image and the words are pixels. There are no characters to extract, so converting it would need OCR, which Filesmith doesn’t do. If you have the original document, converting that gives a far better result.

Why this is the hard direction

It helps to know what a PDF actually contains, because it explains every quirk in the result. A PDF does not store paragraphs, headings, or tables. It stores glyphs at coordinates — this character, this font, this size, at this position on the page. The structure you can see is something your eye assembles, not something recorded in the file.

So converting to Word isn't a format change, it's a reconstruction. The converter groups nearby characters into words, words into lines, lines into paragraphs, and infers which of them are headings from size and weight. It's inference, and inference is occasionally wrong.

Where it usually needs a hand

Feature	What to expect
Body text	Reliable. Paragraphs and basic styling come through cleanly.
Tables with ruled lines	Usually detected, since the rules give the structure away.
Tables without rules	Often arrive as tab-separated text or separate paragraphs.
Multi-column layouts	The riskiest case — check the reading order first.
Footnotes	Typically land as ordinary text at the foot of the page, not as real footnotes.
Hyphenated line breaks	May leave words split in half where the original wrapped.

A sensible order for tidying up: fix the heading styles first so the document outline is right, then rebuild any tables, then deal with images. Doing it the other way round means redoing work.