I think it's very easy to over-index on this point. Indeed I think it's now super fashionable to do so. The whole "ideas are cheap, execution is everything" argument.

Ideas aren't everything but they are important. The crucial distinction is that the initial idea isn't that important, iteration to an idea you can scale (and the execution that follows) is.