There is no reliable way and it will always be a hunch. Some questions are hard to nail down and need a bunch of subquestions to define them. Others only need their title and even filling in the text field below that is unnecessary redundancy. But even those questions that need a fence of subquestions to circumscribe them all point at a single focus of interest.
If a question reads like an essay and asks you to cover all aspects of a field in your answer, it is too much. I think Taal's questions on hypnotherapy and NLP are good examples of this. I'd love to see a poll asking who of the 111 visitors actually read through the whole question. A question that makes you want to go tl;dr, is too big. The problem is not that any part of the question is wrong or off-topic, but that Taal wants to understand everything about hypnotherapy and NLP through one single question. And that is not what SE can do, that would be a chapter in a psychotherapy handbook, and those chapters exist and wait in the next university library for Taal to read.
My personal preference in natural sciences is: one question per question.
In psychology, journal articles have a scope of one question per article. My reason for that is, that after an answer, there are many open paths that your research might take. If you try to foresee all the related questions, you are preliminarily closing all those open doors, denying you the chance to have surprising ideas and make unexpected discoveries. How do you know what your next question will be after you have heard the answer to your first? And even if you still want to know the same thing, why not give your conversational partner a chance to chime in? Fun science is a dialogue, not a series of mind-numbing monologues. A question that is boring you while you read it, or makes you impatient with wanting to answer to the first sentence, is too big.