- We went from closed beta to complete public launch. This means that without a signup code you were not able to interact with the application other then log in or entering a signup code. If I would have to do this again, I’d go from closed beta to open beta but limited signups and then to fully open signup.
- Starting with just normal signups and Facebook login was probably a good idea (rather than adding Twitter login and potentially other login services too). Facebook gives us enough stuff to deal with right now.
- We made changes to the sign up process up until the day before launch. While this worked out mostly fine, I wouldn’t recommend this to anyone.
- The closed beta rounds have really helped us to improve the product. However, the most valuable thing I experienced personally in the last couple of months was sitting with a couple of new users to see how they signed up.
- Make sure to have top notch serverguys. While we were mentioned on national television and radio stations several times, we had zero down time.
- Make at least one person responsible for handling support during launch. We have 4 people doing a phenomenal job there.
- Our frontend application is a single page app. Most people love this, but people using an iOS device are confronted with crashes due to memory leaks way too much. We had to balance between building features and debugging memory issues during the last couple of weeks. Loading up to 178 (!) webfonts at the same time isn’t helping either. Points to take away: if you are building a single page app, develop “memory first”. Make it a habit to check for memory problems during development and test your app on iPhones and iPads. Part of these issues can be appointed to iOS7, there are a lot of people complaining about Safari being unstable. When we removed CSS transitions, almost all crashes disappeared immediately (we're still working on iPad animations).