Why can’t a PDF become an identical Word file?
A PDF says "draw this glyph at this coordinate". A Word document says "this is a paragraph in this style". Going forwards is straightforward; going back means guessing which glyphs formed which paragraph, which lines belonged to a heading, and where a table’s columns were.
Every converter does that guessing. The ones that claim pixel accuracy do it by pinning each line into a fixed text box, which looks right until you type one word and the layout falls apart.
What comes across reliably?
Paragraph text with its bold, italic and monospace runs. Headings, recognized by size and weight, as real Word heading styles so the navigation pane works. Bulleted and numbered lists as lists. Tables as tables, with the rows and columns found from the layout. Pictures, placed in the flow where they sit on the page.
What is deliberately not reproduced: exact line spacing, multi-column layouts, headers and footers, and original fonts (text is set in Calibri at the sizes measured in the PDF).
Which PDFs convert well, and which fight back?
Single-column reports, letters, contracts and CVs convert cleanly. Two-column academic papers, magazine layouts and forms with boxes drawn around every field are the hard cases, because reading order is ambiguous even to a person.
Scanned documents cannot convert usefully at all: they hold a picture of text, so there is nothing to move into Word until OCR has added a text layer.
Should you convert, or edit the PDF directly?
If the change is small — a typo, a date, a name — editing the PDF is less work and keeps the layout exactly. If you need to rewrite paragraphs, restructure sections or reuse the text elsewhere, converting to Word is the right move.
Going back the other way is lossless in the sense that matters: Word to PDF keeps headings, lists, tables, links and bookmarks.