Someone sends you a PDF and asks for "the Word version". You run it through a converter, open the result, and the first thing you see is a table that has become forty separate text boxes. Nothing has gone wrong. You asked for something the file could not give you.
The two formats describe different things
A DOCX is a description of structure. Underneath the zip wrapper it is XML saying: this is a Heading 1, this is a numbered list at level two, this table has four columns, this paragraph uses the style called Body Text. Where any of it physically lands on a page is worked out later, by Word, when it lays the document out.
A PDF is a description of appearance. It says: draw the glyph "T" at x=72, y=698 in 11pt Calibri, then draw "h" at x=79.2. There is no concept of a paragraph, a heading, a list or a table. A table in a PDF is some lines and some text that happen to be positioned so a human reads it as a grid.
Converting DOCX to PDF is therefore a rendering step, the same category of operation as printing. Converting PDF back to DOCX is an inference step, where software looks at scattered glyph positions and tries to guess the structure that produced them.
One of those is reliable. The other is a reconstruction.
Going DOCX to PDF
This direction is the well-behaved one, but there are three things that change.
Fonts get embedded or substituted. If the document uses a font the converter has, it embeds a subset and the output looks identical. If it does not, it substitutes something metrically similar and your line breaks move. This is why a CV that fitted on one page in Word arrives as a page and a half.
Everything interactive stops being interactive. Comments, tracked changes, hidden text and field codes are resolved and flattened. A { PAGE } field becomes the literal number. A cross-reference becomes the text it happened to say at conversion time. Tracked changes are either burned in or dropped depending on the view being rendered, so it is worth accepting or rejecting them yourself first rather than finding out afterwards.
Reflow ends. A DOCX re-lays out for the page size it is opened at. Once it is a PDF, the pagination is fixed. That is usually the point of converting, but it also means an A4 layout stays A4 on US Letter, with the margins slightly wrong.
Going PDF to DOCX
Here the converter is reverse-engineering. How well it does depends entirely on what kind of PDF you gave it.
| What you start with | What a converter can recover |
|---|---|
| PDF exported from Word, text intact | Paragraphs and most runs of formatting, usually good |
| PDF with a tagged structure tree | Headings, lists and tables, reliable |
| Designed PDF from InDesign or LaTeX | Text, but the structure is a guess |
| Multi-column layout | Text, often in the wrong reading order |
| Scanned pages | Nothing at all without OCR first |
That third row is worth dwelling on. A PDF has no reading order unless someone added one. In a two-column newsletter, the glyphs may be stored in the order they were drawn, which can interleave the columns. The converter reads what it is given and you get alternating half-sentences.
The fourth row is a hard stop. If you cannot select a sentence in the PDF with your cursor, there is no text in it, only an image of text. No converter can pull words out of that. It needs OCR to generate the text first, and then you are working from a transcription with its own error rate rather than from the original characters.
What "tagged" means and why it is the whole game
A tagged PDF carries an invisible structure tree alongside the visual content, marking which glyphs belong to which heading, paragraph, list item or table cell. It exists for screen readers, and it is also exactly what a converter needs.
Word writes tags when you export properly. It does not when you use the print-to-PDF driver, because a printer driver only knows about ink. Those two routes produce files that look identical and convert completely differently.
So if you control the original, the useful habit is: export, do not print. In Word that means File → Save as PDF, or Export, rather than Print → Microsoft Print to PDF. The file is a little larger and it stays convertible, searchable and accessible.
When you are stuck with the PDF
If you did not make the file and it converts badly, the question is what you actually need from it.
- You need the words, not the layout. Extract the text and paste it into a clean document. You will reformat, but you will not be deleting forty rogue text boxes first.
- You need the words in a document you will restyle anyway. Convert to DOCX, then run remove formatting immediately. Stripping the inherited mess is faster than editing around it.
- You need it as portable structured text. DOCX to Markdown keeps headings and lists as plain, diffable text with no layout to fight.
- You need to change three words in a contract. Do not convert at all. Edit the PDF directly or ask the sender for the source. A round trip through DOCX will reflow the whole thing and change what the document looks like to the other party.
The habit that saves the most time
Keep the DOCX. The PDF is the output, not the file.
Almost every painful conversion starts with the original being lost and the PDF being the only copy left. When the source exists, you edit it and re-export in thirty seconds. When it does not, you are paying a reconstruction tax every time the document needs a change, and paying it again at every future revision.
Convert DOCX to PDF for anything final, and keep the original Word file for edits. Converting PDF to Word is a recovery step for when that original is lost, not a way to edit a document, so expect to tidy the result.