In an iterative industry, I think there is too much emphasis on "launch". From day 1, you should have someone interested in your product/service. If you haven't got at least one early adopter, someone sold on the idea, why bother?

The real "launch" is the tipping point where there is enough functionality that it fully solves the problem for the intended audience. But you won't get there without early usage and feedback.

Rather than think of chicken and egg problems, start adding cells together and see which one evolves first.

I'd add it:

* user observation and interviews

* a/b testing

to the list of tools.

(Indeed - the whole discussion is framed in the context of 'the product launch'. With a chunk of my clients that's just a dead concept. They're trying out new features with small chunks of their market at many different levels of fidelity all of the time. Not everybody can do this - outside of the context of web applications/sites it gets much harder - but if you can escape from the idea of the product launch everybody's lives seems to get much simpler)