A Word document describes structure: headings, paragraphs, tables. A PDF describes appearance: put this glyph at this position. Converting means guessing the structure back, and guesses are sometimes wrong.
What converts well
Text-heavy documents with a simple single-column layout usually come through cleanly.
What drifts
Multi-column layouts, complex tables, and text wrapped around images. Expect to tidy up.
What will not work at all
Scanned documents. A scan is a picture of text, with no text layer to recover. Run it through the OCR tool first, then convert.