Was excited to see Variance in the Inspector panel for Task:Schedule... until I tried editing it. I understand it's for reporting the variance from a baseline, but here's what I need:
If Bart says he estimates that a task will take 20 hrs, but it could be as little as 10 if he's lucky, or 30 if he's not, then I want to be able to enter in all three values as Best Case, Expected, Worst Case. Then the schedules in the Gantt can reflect this with essentially error bounds. I can also play with the values a bit and see how problems or lucky breaks propagate across the schedule for that project. (Or, optimally other projects assuming multi-project support.) This would let me determine critical paths not only for completion, but also determine critical tasks for schedule slippage.
A day lost on Project X may cause Project Y to slip *badly* if they share a resource that cannot be re-allocated in time.
Actual work entered would of course override this as the project progresses, and a final report could let us see exactly how close we were in our original estimates, at all three points. If we consistently hit our targets, great. If we consistently miss our targets, but our estimates for variance are decent, that's okay, at least we have the bounds correct. If we consistently miss our targets *and* our variance estimates are way off, then we have some learning to do.
If Bart says he estimates that a task will take 20 hrs, but it could be as little as 10 if he's lucky, or 30 if he's not, then I want to be able to enter in all three values as Best Case, Expected, Worst Case. Then the schedules in the Gantt can reflect this with essentially error bounds. I can also play with the values a bit and see how problems or lucky breaks propagate across the schedule for that project. (Or, optimally other projects assuming multi-project support.) This would let me determine critical paths not only for completion, but also determine critical tasks for schedule slippage.
A day lost on Project X may cause Project Y to slip *badly* if they share a resource that cannot be re-allocated in time.
Actual work entered would of course override this as the project progresses, and a final report could let us see exactly how close we were in our original estimates, at all three points. If we consistently hit our targets, great. If we consistently miss our targets, but our estimates for variance are decent, that's okay, at least we have the bounds correct. If we consistently miss our targets *and* our variance estimates are way off, then we have some learning to do.
Linear Mode
