I'd like to see customer involvement much earlier. You could start out solving a specific problem for a customer who will pay you to do it. If that particular problem seems to be relevant for other customers, you can make it into a product. This only requires that your initial contract with the first paying customer allows you to do this, but if you allow the price to go down this should not be a major obstacle. You should probably still go through the intuition and market analysis phases, but the benefit with this model is that the idea will have passed the testing phase already.
I started writing this as an ask HN post, but it became so long that I decided to write a blog post instead. Still, what I am really after is feedback from you guys :-)