Arch,

For myself, it depends. Everyone, I'm sure has an approach that works
for them. Here's what seems to work for me.

If I'm rough turning green blanks the size of the chunk and the figure
(especially the figure) strongly influence the result. But I do make
sure I have the general shape in mind before I start. I've found that
I end up with something "far less satisfying" if I don't have at least
some idea where I'm trying to go.

For (semi?) segmented new designs I usually make a drawing to make sure
the piece is going to look good (at least on paper). I don't get
obsessive about matching the paper dimensions exactly - smooth curves &
form are too important to do that.

Finally, on something I need to work out the production
sequence/technique for or on something that I have multiple (slightly
different) options for, I make one or more prototypes out of cheap wood
or scraps. Usually one (or one of each option) is enough to tell me
what needs modification, both in shape and production sequence. On one
particulary difficult piece I made 5 "trial" pieces before I worked out
all the production problems.

Cheers,
Larry