Great question!
If this is a money venture, my first task would be to research the idea and see how it's been done before, and then I would analyze how it could be made better. If I feel it can be done better, then I just completed my first step: research.
If I'm using this for learning purposes (and even if it's for $$), my first couple of steps would be to first write out overall specs on basic features I'd like to see built in. These would be based off of more research, keeping in mind things ARE going to change (nobody gets the specs right the first time). Now that I have an overall idea of what I want, I start thinking about architecture. What language(s) do I use? Do I want to use something that's quick and dirty, and get an MVP stood up as soon as possible? Or do I want to learn a new technology? Perhaps time is on my side and I want to do everything "right" the first time through...What's the best database to use? Once the general technologies are chosen, that's when I begin the coding process.
TL;DR - Research is the most important thing to me when diving into a new project. The more knowledge you have, the easier it'll be to do the actual work :)