The slow start approach doesn't work well for novel, or large-scale fast growing businesses. The former case, one needs repeated iteration and trial and error

The slow start approach doesn't work well for novel, or large-scale fast growing businesses. The former case, one needs repeated iteration and trial and error to discover/create a new market. You will deviate several degrees from your original direction within the first 2-3 years. In the later case, you have no choice to grow "smart" if you have something viral. You don't choose your scale or the speed of your growth once a few people discover you. You will end up working insane hours just to keep the infrastructure alive. (this is strictly for software businesses, btw. but I can think of an instance of D.C. hole in the wall sandwitch joint that got reviewed on NPR. Nothing like a flesh-DoS, and you can't tell that guy he doesn't need to put in insane hours.)