In my many years of estimating, the ONLY thing I have found to work consistently is to meticulously break work into a series of tasks taking no longer than a fe

In my many years of estimating, the ONLY thing I have found to work consistently is to meticulously break work into a series of tasks taking no longer than a few hours. Real time, not points. Don't estimate in a meeting. Let people go back to their desks to research and document the issue. Give them time. It takes longer, but it yields more accurate estimates. Have the team meet up again afterward and let everyone discuss the estimates and how they arrived at each piece. Let the team come to a group consensus about each estimate. By breaking it out into a series of small tasks, you also have a daily schedule and a daily benchmark to work against: [x] X (2 hours est., 2 hours actual) [x] Y (3 hours, 5 hours actual) [ ] Z (1 hour) Now you have a daily update: "I went 2 hours over on Y yesterday and didn't get to Z, but I stayed a little later to get Y finished. I'm at least 1 hour behind schedule, but I think I can make it up today and tomorrow." This makes you super dependable to management and keeps you on task. Super dependable means you make them look good--which means you're first to get the choice assignments, first to get a bonus or raise, first to get a promotion. Managers know that schedules occasionally slip; what they want is to know as early as humanly possible so they can make a decision about the project.