The Prophet's Dilemma

No real substantive post today; what started as a busy week has turned into a nightmare, complete with some very bad news about a very old friend.

Well, life goes on, and with it, work. On that front, my company is in the process of our first ever project post-mortem. One of the questions that's come up is the amount of hours spent unit testing vs. improvements to code quality. We had to fight to get hours in the schedule for UTs, and we got them. But now we're faced with the "Prophet's Dilemma": we predicted quality would suffer without unit tests, and we got unit tests, and now we can't prove that code quality would have suffered without them.

Another concern is how much UTs are enough. The ideal is supposed to be that if you can change the meaning of a line of code, and none of the tests break, you're lacking cases. How many of you have that kind of coverage on an enterprise system, and how did you get it (in a major UT/development push, or slowly over time)?