Maybe you could save some time by having more than one testing book; make several, ensure they are totally distinct internally, and move them all over at the same time; test 5 or 10 versions, each subtly different in the SVG code, and then, based on what you see, change and repeat?
|