Key moments are the part most sites skip
The basic video result — thumbnail, duration, title — comes from a correctly marked-up VideoObject. Key moments, which let a searcher jump straight to a labelled section, come from adding Clip nodes or a SeekToAction, and almost nobody does.
They are worth the effort because they multiply what one video can rank for. A twelve-minute tutorial with six labelled clips is six answers rather than one, each addressable from search.
The labels have to be honest — a clip described as covering something it does not is a bad experience Google can measure — and the timestamps have to be right, which means maintaining them when the video is re-edited.