Converting to or from PDF is a change of container, not a change of design. That single distinction explains most of the disappointment people feel with any converter, including this one.
A PDF describes fixed positions: this glyph, at this coordinate, on this page. A word processor document describes flowing content that is positioned when it is opened. Going from flowing to fixed is reliable, because the layout is being frozen. Going the other way means guessing which frozen positions were once paragraphs, and the guess is imperfect on anything complex.
Practically: expect text and images to survive well. Expect multi-column layouts, text wrapped around figures and elaborate tables to need attention afterwards.
Text, images, page order and simple structure usually do. Multi-column layouts, text wrapping and complex tables are the elements most likely to shift, because they depend on layout decisions that do not transfer cleanly between formats.
Conversion changes the file format. It does not set the trim size, margins or bleed that a print interior needs, which is a separate formatting step.
Not with a format converter. A scan stores a picture of a page, so there is no text to extract. Optical character recognition is the tool for that, and its output always needs proofreading.