In a post of quite some years ago, I talked about the drawbacks of the trial and error method. It often took me a long time to realize that a piece was defended, and that the logical conclusion was to eliminate that defender. In order to speed that up, I invented the tree of scenarios. With 23 standard scenarios, if I remember correctly.
But thinking about scenarios is a system 2 exercise, it is no skill. So it didn't work. But slowly we are finding out how to transfer this knowledge to system 1. It is about the concepts and the analogies we talked about.
I work with a database of 748 selected tactics. It is very important that these tactics are selected well. I found these careful selected tactics in two books at Chessable.com: The checkmate patterns manual by CraftyRaf and 1001 chess exercises for club players by New in Chess.
Right now, I'm internalizing all 748 tactical exercises. The spaced repetition method is a great help for that. Once the moves are internalized, I focus on the technique behind the moves. The tree of scenarios. Which scenarios are at work? Once I found out, I simply memorize them. Once the scenarios are memorized, I try to conceptualize them. To generalize the ideas. What is needed for the tactic to work? I work bottom up, that is to say, I start with the position and generalize from there. That is the other way around as the top down approach with the tree of scenarios. What do we need? An undefended back rank, with a rook that can enter at will to sac itself in the corner, a bishop on the long diagonal to pin the pawn on g2, b2, g7 or b7. And a Queen that can deliver mate on those very squares after entering via the rook file. See the comments here
Besides this conceptualization, the phantom auras need of course to be imagined, and these kind of exercises are perfectly suited for that:
|White to move|
8/8/3p2k1/2q5/3N4/1B1K4/5R2/8 w - - 0 1