Something I learned from working at a large competitor at Zynga (where there was some employee cross-pollenation): A/B test the hell out of your ideas. Going from "making my app more social" to "Put vacant lots on users' farms to get them to add friends" means playing with lots of different ideas. What exactly clicks depends heavily and all the myriad details of your app and userbase.