There is a moment in nearly every exam where a student meets a question that does not look like the ones they practised. What happens next tells you everything about how they were taught.
A student who memorised the method freezes. The shape of the question is unfamiliar, so the steps they learned no longer obviously apply, and they have nothing underneath to fall back on. A student who understood the idea looks at the same question and recognises what it is really asking, even in disguise.
This is why I build understanding first. It is slower at the start and it can feel less productive than drilling past papers, but it is the only thing that holds when the exam does what exams are designed to do: test whether a student actually understands, not whether they can repeat.
Once that understanding is genuinely solid, the past-paper work becomes enormously effective, because the student is now training a real skill rather than memorising more patterns. That is the whole logic of the Eigenstate Method.