Failure is an option. Failing is part of going fast. A willingness to accept risk, to iterate quickly, to test hardware, to accept failure. There should be an expectation that failure is highly likely.
 
If you're not at that ratty edge of what's possible, how are you gonna know, until you have a failure?

- That was a failure.
- No, that was a like 80% success.

Sit down and think about something you've been doing the same way for like a year or longer. You've probably missed an opportunity to do it better.