I am surprised that the estimates are actually pretty reasonable and reliable. After all, the guys who do the work create the estimates. I think there is a problem on how the estimates are being used. If an estimate for a change comes out to 80 hours, you cannot schedule the thing to be done in two calendar weeks.
Nobody works eight hours a day on a change. There are interruptions and other activities going on. I would hope that management 101 dictates that you use some percentage of your resources' time to estimate progress toward using up the hours from an estimate. Ha. Maybe nobody else heard of this novel idea.
Currently we are adding the tracking of actual hours worked. We track how many hours we spend each day on which tasks. That will give us a clear realistic picture of how many hours of the workday are spent on the changes we produce. You would think that with this information, plus accurate estimates, one could reliably predict our project schedule.
Then again, I may be expecting too much.
