I've found that engaging earlier--and attempting these efforts sooner--is a mistake. It tends to cause more conflict than not. To me, engaging too early is like standing there and telling a painter "maybe you should use more green" or telling a musician "try a different chord."
I find I'm most helpful when I can engage with a developer who has realized the initial iteration of his/her vision.
Admittedly, this may result in a little wasted time and some backtracking. But more often than not, that's a small hurdle for the benefits of a) having someone realize their vision and b) reaching a point of being comfortable with, seeking, and accepting criticism and guidance. It just starts the whole relationship out on a much better footing.