The mistake that many people make though is becoming too wedded to that original business plan. A process of research and analysis can be productive, but it must lead quickly to a rapid prototyping phase. We have to take action, and learn from that action. However, we can't just leap without any spade work.
For the Wall Street Journal, London Business School Professor John Mullins has written a great column exposing some of the many problems with traditional business plans. For those interested in learning more about this topic, I strongly suggest reading his essay.