Tell me about leading a post-mortem after a significant quality issue. How did you run it, and what came out of it?

Behavioral