The track plan to me is an evolutionary thing, up to (and even after) the ballast is laid. A deeper discussion of this aspect can be found here. I tend to get into the scale 1:1 planning- that is, playing around with different arrangements with my actual track after the space is defined. I have an 80% solution in mind, but for that last 20% I use test-fitting and some actual operation on tacked-down track.