There are rough order-of-magnitude differences between the feedback times for build-time errors, automated tests, manual testing, CI, staging, and production. This is useful when thinking about tradeoffs of where you want to catch certain failure classes.
