Two thoughts: The best startups are masters of asymmetric warfare -- they don't fight on the battlefield the incumbent selects. And, users often understand their problems well but are horrible at suggesting features to solve these problems.
When startups start taking off, the founders usually can't tell what's changed. It's some combination of all the random things they did to make it take off, but it's hard to say which ones.