I think your approach would be the optimum balance. Build something rather quickly that fits your problem domain. You don't have to build something perfect, jus

I think your approach would be the optimum balance. Build something rather quickly that fits your problem domain. You don't have to build something perfect, just something that is usable and is relevant to the problem your attacking (use common sense). Release something to your target audience and see what they think of it. Then you can iterate from there mixing in your academic market research and your real life market research. Oh before I forget, build something that you yourself will definitely use. And don't forget to have fun in the process!