Signposting is how a listener rejoins after drifting

Anyone who tuned out needs a way back in that does not require rewinding.

Print has headings, paragraphs, white space, page numbers, and a visible sense of how much is left. All of it tells a reader where they are without any of it being read. Audio arrives with none of that. The listener's only sense of position is what they can remember, which is exactly the faculty that drifting has interrupted. Signposting is the medium's substitute for the whole visual architecture of a page, and it has to be spoken, because there is no other channel available for it.

The forms are unglamorous and they work. Saying what is about to be covered before covering it. Marking a transition explicitly instead of letting the topic shift silently. Naming the current section again in the middle of it, not just at the start, so that someone arriving late from their own thoughts finds a handhold within a minute rather than five. Summarising a stretch before moving on. In written prose this level of scaffolding would be condescending. In audio it is the difference between a listener who rejoins and one who gives up.

The underlying principle is that a spoken explanation should be reconstructable from any point in it. Not fully, but enough to know what is being discussed, why, and whether it matters to you. That is a real constraint on structure, and it rules out the kind of long slow build whose payoff depends on having heard every prior minute. Some material genuinely needs that build. Most does not, and pretending it does is usually a way of avoiding the work of making the shape audible.