A text-based PDF (created in Word, LibreOffice, or another editor) contains real text you can select and copy. Converting these files preserves paragraphs, headings, and tables fairly well. Minor spacing or font deviations may occur with non-standard typefaces.
A scanned PDF is an image – no text, only pixels. Without OCR, it converts to an empty DOCX. For correct conversion, run the OCR tool first, then convert to Word.
Complex layouts – multiple columns, merged-cell tables, shapes – may not transfer precisely. This is a PDF format limitation. Review the result and adjust manually as needed.