Keeping a native copy of the word documents would be ideal but unfortunately we're working with 20,000-30,000 already exported pages so that ship has sailed.

Our "tagging" not a word feature, we use an HTML element that's been added to the document outside of word. It looks something like this:

then we have a script in our LMS that will inject the requested content via an ajax request. Since it's an empty element no content appears without our script present, thus it's easy to be deleted from within word.

Thanks a lot for your help thus far by the way!