To get work done, you've got to do the work.
It's that simple.
The idea of estimating effort and duration that is as far apart as 40 hours and 12 days for a single task is anathema to effective project performance. If you set up an effective project environment -- one in which resources are allowed/encouraged/expected to focus on the single most important task until complete, then the vagaries that get added by multitasking will be minimized.
On the other hand, you do have to recognize the possibility that Murphy's Law has not yet been repealed. Using a 2-point estimate, let's say 30 hours aggressive and 60 hours safe, is far more useful than a single point 40 hour estimate. With such a pair of estimates, you can develop a buffered critical chain schedule that will reflect both the possible and the possible. Using such a methodology to schedule, and then a simple frequent and regular reporting of estmates to completion allows one to track consumption of buffer, which provides guidance 1) for identification of the "most important task," 2) possible needs for corrective action, and 3) general health of the project promise.
But the behaviors -- relay race behaviors -- are the key to making it happen on time.
