Why PDF to Word loses formatting
A PDF records where things sit, not how the document was built. That gap is what you see in the converted file.
A Word document is a set of instructions: this is a heading, this paragraph follows it, this table has four columns. A PDF is the result of following those instructions, and the instructions are thrown away. What remains is a record of where every character was placed on the page. Converting back to Word means reading those positions and guessing at the structure that produced them.
Good guesses are possible when the layout is regular. Body text in a single column, headings that are clearly larger, tables with visible rules: these usually come back as real paragraphs, real headings and real tables. The conversion has something to work from.
It struggles when the layout is doing something clever. Multiple columns, text wrapped around images, headers and footers that repeat, tables held together by alignment rather than lines. Here the converter often falls back on text boxes, which look right on screen and behave badly the moment you edit them, because each box is pinned to a position rather than flowing with the text.
A scanned PDF is the hardest case, because there are no character positions at all, only a picture. Run OCR first to put a text layer in, then convert. And if what you want is the words rather than the layout, PDF to Text gives you them cleanly without the fight.
Tools for this