This is what I've figured out working on this startup so far. It's one of those 'social' startups which always have a high chance of failure but... we'll see.
Lessons I've learned and which we are re-building around:
1. Make it work for 50 people and it can then scale up. We have not done this yet. I should have done this from the beginning. There are no words for how dumb I was
2. It is right to dream of a big crazy idea because the 'concept' or 'direction' is important and the details do fall into place e.g. learning to make it work for 50 people. It's not an A to Z path and it's not comfortable.
3. We should have been making what we would have loved to use rather than what we thought others needed. However, the time taken to research what others wanted has paid off because the first way is suspect to introversion. Chicken <> egg
4. F**k the sign up processes first time. It should have been 'We have a problem. How would we want this app to solve it in the fastest way possible without pissing us off the way most apps do'. (this is still slightly defensive thinking though). Visualise and be creative around raw instinct for what works or does not.
5. This is one thing we have done but did not quite get it right: "Have a core". The objective has always been the same, we've just never expressed it well enough. Read this: http://andyswan.com/post/16525016414/having-a-core
6. Following on from number 5. This is what I've always kept in my head: It takes 70% instinct, 30% technique and the passmark is 95%. We're re-sitting the exam. While technique only makes up 30%, it takes up 90% of the total time. However, it has to filter though instinct (love of solving the problem)
7. There is defensive thinking and attacking, passionate thinking. Defensive: How do we improve our messaging to help the user understand the site? Passionate: This whole site will be like a great movie that the user is going to love to watch. You're in a relationship with the user, which personality would they rather date?
Chief muppet of LikeOurselves.com