"WHILE I THINK the reasons for postmortems are compelling, I know that most people still resist them. So I want to share some techniques that can help managers get the most out of them. First of all, vary the way you conduct them. By definition, postmortems are supposed to be about lessons learned, so if you repeat the same format, you tend to uncover the same lessons, which isn't much help to anyone. Even if you come up with a format that works well in one instance, people will know what to expect the next time, and they will game the process. I've noticed what might be called a "law of subverting successful approaches," by which I mean once you've hit on something that works, don't expect it to work again, because attendees will know how to manipulate it the second time around. So try "mid-mortems" or narrow the focus of your postmortem to special topics. At Pixar, we have had groups give courses to others on their approaches. We have occasionally formed task forces to address problems that span several films. Our first task force dramatically altered the way we thought about scheduling. The second one was an utter fiasco. The third one led to a profound change at Pixar, which I'll discuss in the final chapter. Next, remain aware that, no matter how much you urge them otherwise, your people will be afraid to be critical in such an overt manner. One technique I've used to soften the process is to ask everyone in the room to make two lists: the top five things that they would do again and the top five things that they wouldn't do again. People find it easier to be candid if they balance the negative with the positive, and a good facilitator can make it easier for that balance to be struck. Finally, make use of data. Because we're a creative organization, people tend to assume that much of what we do can't be measured or analyzed. That's wrong. Many of our processes involve activities and deliverables that can be quantified. We keep track of the rates at which things happen, how often something has to be reworked, how long something actually took versus how long we estimated it would take, whether a piece of work was completely finished or not when it was sent to another department, and so on. I like data because it is neutral--there are no value judgments, only facts. That allows people to discuss the issues raised by data less emotionally than they might an anecdotal experience."