@
Education - follow Hitch's advice and use Word Styles. Before saving I search for tabs (^t), double paragraph marks (^p^p) etc and eliminate via styling.
I save as DOCX and convert via calibre's conversion facilities (ebook-convert), I see very little cruft from that process. The Word Styles convert pretty much 1:1 with CSS styles. The only thing I might do is cosmetic and move things around a bit.
I guess what I'm saying is that KG's DOCX to HTML converter is better than MS's is
BR