Start at the top and work your way down. Don't consider how you should build it (design and technical) till you feel comfortable you're solving the right problem the right way.
Note: This post was part of my LinkedIn series from last year. I'll be expanding on it in the future. Subscribe to the newsletter to receive updates.