I think they fall into "Let's rewrite this and make it better!" FAIL. You need to be pretty stupid to rewrite something that's actually working unless you got a great confident, team and time, and final client of them proves that they sucked, and now it's going to take them another 6 months to come up with something as good as their previous client.
When startups are going to learn to not to do obviously stupid like "OMG, our client is not good enough, instead of improving it, let's rewrite it, yay!" ?