I agree entirely with Iratus about the grid fill and the necessary element of guesswork to get started. I almost always start a Listener in pencil, recognising that the final entries may change (whether by the endgame or to make progress with a carte blanche). In practice, I was able to make steady progress on the grid fill without having to change any of my tentative pencil entries.
On 25A, yes if you hazard that there are only 2 parts contributing to the grid entry, then in theory there are only 4 possible answers. But we also know all grid entries are real words/names and on that basis there is only one feasible answer (in BRB at least). For me, there is no point in worrying about the possibility of there being 3 contributing parts to 25A until I have discovered that the only feasible 2-part entry doesn't work with other crossers, something which never arose. Doing that also helped me to pencil in which of the 2 feasible grid entries to pencil in for 22D (again hazarding that was also a 2-part entry). And that in turn gave me an idea about what to expect of 20A etc.
Careful tracking of what parts of clue solutions I'd used where was of course also important, but all-in-all the puzzle gave me enough to keep making progress and be rewarding.