Most players review a solve by checking whether they got the answer. That is a pass or fail question and it teaches nothing. A review should answer a different question: where did the solve stop being planned?
I use a four-question script that takes about five minutes. One: what was my plan at rung one, in one sentence? Two: at which rung did the sentence stop being true? Three: was the wrong turn a word problem or a frame problem? Four: what word would I have needed to know to keep the frame alive? The answers go in a notebook, three or four lines total.
The third question is the one that changes behaviour. A word problem is a vocabulary gap and it is fixed by exposure. A frame problem is a planning gap and it is fixed by slowing down at the start. Most players assume all their losses are word problems and drill vocabulary, which is why so many people plateau while feeling like they are improving.
The fourth question turns each review into a small piece of research. You end up with a list of words that keep appearing at the moment a plan breaks, and that list is more valuable than a generic vocabulary list because every entry is tied to a real failure. If you keep the notebook for a month, you will see the same four or five words repeating, and those are your actual curriculum. Browse the archive for older routes when one of those words shows up, and check whether the published solution used the same pivot or found a different one. That comparison is the closest thing this hobby has to a coach.
The last thing I would say about the script is that it should not be a performance. Nobody reads the notebook, and the four questions are not a template to fill in properly. Scribble the answers, close the book, go do something else. The point is the five minutes of honest attention while the puzzle is still fresh.