You just need the ability to make things that work. And by "work" I mean actually work, and in a way that others want to use it (an intuitive interface that needs almost no training/explanation helps a lot here). It doesn't have to be polished, pretty, or anywhere near feature complete (see: Apple 1). It just has to pass the "will people actually use this beyond the novelty phase" test. I can't stress enough how important intuitive function is above form at this stage.
That's step 1.
Step 2 is to have a management team that can effectively monetize that thing and manage the risks (financial/legal/market) that all businesses have to deal with. Engineers are usually not good at these things.
I've succeeded at step 1 many times but I'm personally not good at Step 2.