There is what Steve Blank calls the stage where you are searching for a scalable business model. Then, there is the stage when you have found that model and need to scale it. In the former stage you have to have a "beginner's mind," be in learning mode, and expect to learn things you didn't anticipate.

Of course, it is worth it to take the time to think carefully through your assumptions, and ensure you at least have hypotheses around how you will create value. But use the analysis as a way to focus attention on the most critical assumptions, rather than spend a ton of time massaging the numbers.

Innovation is doubly hard inside big companies.

First you document your idea. You should be comprehensive, but that doesn't mean you have to produce a doctoral thesis length plan. Rather you want to make sure you have touched all the different things that have to happen to succeed. Then, you evaluate your approach. The goal here isn't to figure out if your idea is good or bad, but rather to begin to figure out what are some of its weakest elements.

People who copy what exists copy a point-in-time artifact, and if you are managing the process correctly you are already hard at work on the next thing.

Almost every disruption starts at the perceived fringes of today's market.

The need to be thoughtful about experiment design is particularly acute within large companies, since some of the behaviors, such as having small teams and tapping into low-cost resources to maximize flexibility, won't come naturally to many people inside huge companies.