Why PDF to Word conversions come out wrong
· 6 minute read
A PDF has no paragraphs in it. Understanding that explains almost every bad conversion — text in floating boxes, headings cut off mid-word, grey bars over the text.
If you have ever converted a PDF to Word and got back something that looked almost right but could not be edited, this is why. It is not a bug in the converter you used. It is a consequence of what a PDF actually is, and it is worth ten minutes to understand, because it tells you what to expect from any tool that offers the conversion.
A PDF does not contain paragraphs
A Word document is a structured thing. It contains a heading, then a paragraph, then a list, and the software works out where those land on a page when it draws them. Change a word near the top and everything below shuffles down. That is what "reflow" means, and it is the whole reason a document format is useful.
A PDF is the opposite by design. It is a description of a finished page: put this glyph at this coordinate in this font at this size, several thousand times over. There is no paragraph in the file. There is no heading. There is not even a word — only a series of letters that happen to sit close together, and a space is often not a character at all but a gap left by moving the drawing position along.
This is not a flaw. It is what PDFs are for: a page that looks identical on every machine, for ever, whether or not the reader has the fonts. Everything good about PDFs and everything painful about converting them come from the same decision.
So "convert to Word" really means "guess"
To produce a Word document, a converter has to put back the structure that was thrown away. It has to group glyphs into words, words into lines, lines into paragraphs, work out which lines are a heading because they are larger and bolder and followed by a gap, and notice that a grid of short runs aligned in columns was probably a table.
Every one of those steps is inference. A good converter is right most of the time. None is right always, and any tool promising a perfect conversion is either not doing this or not telling you the truth.
The three symptoms, and what each one means
Bad conversions fail in recognisable ways, and each tells you something about the tool that produced them.
1. Every line is in its own floating box
You open the file, the words are in the right places, and then you try to fix a typo and discover that nothing moves. Each line sits in its own text frame, pinned to a coordinate. Adding a sentence pushes nothing down.
This means the converter did not reconstruct anything. It traced the page — took every text run and placed it at the position it occupied in the PDF. This is what you get from tools built on a graphics importer rather than a document analyser, and it is the worst outcome, because the file looks accurate and is useless. The only reason to want a Word document is to change it.
A quick test: click into the middle of the document and press Enter. If the text below does not move down, you have a traced file rather than a converted one.
2. Headings cut off mid-word
A heading that reads EDUCATION in the PDF arrives in Word as EDUCATIO. The last letter is simply gone.
The text is usually not gone at all — it is in the file, and it is being clipped. This follows directly from the first symptom. Each floating frame was sized to the exact width the text measured in the PDF, using the font embedded in the PDF. PDFs commonly embed subsets of fonts you do not have installed. Word substitutes something else, the substitute is a fraction wider, and the fixed-size frame clips whatever no longer fits.
It is always the last character, always in the same direction, and it is invisible in a thumbnail. Converters that produce flowing paragraphs cannot have this problem, because there is no fixed-width frame to overflow.
3. Grey or white bars sitting over the text on a phone
The document is fine on a laptop. Open it in Word on a phone and bands of colour are painted across the words.
Those bands are the background shapes from the original page — a coloured header, a sidebar tint. The converter brought them through as floating shapes anchored behind the document text. Desktop Word honours that ordering and draws them underneath. Mobile viewers do not reliably, so the shapes paint on top.
If you see this, always check the file on a phone before sending it to anyone. It is the failure most likely to reach a recipient unnoticed, because the person who made it was looking at a desktop.
What a good conversion actually looks like
A reconstruction will not be pixel-identical to the PDF, and you should be suspicious of one that claims to be. What it should be is:
- Editable — type in the middle of a paragraph and the rest moves.
- Made of real paragraphs, not text boxes. In Word, the Navigation pane will show a document structure rather than nothing.
- Free of floating shapes over the text, on a phone as well as a laptop.
- Roughly the same length as the original. A two-page PDF that becomes six has had its layout shattered into fragments.
That trade — slightly less faithful, actually editable — is the right one almost every time, because if you wanted something that looked exactly like the PDF, you already had the PDF.
When no converter can help you
If your PDF is a scan — a photograph of a page — there is no text in it to recover. Every converter will either hand you back an empty document or a picture of a page pasted into Word.
The fix is a separate step: optical character recognition, which reads the letter shapes in the image and produces text. Run OCR first, then convert the result. Any tool that silently converts a scan into an empty document without telling you why is wasting your time.