I attended an interesting session on Risk Management the other day, sponsored by our local Microsoft Project user’s group. The presenter was from Palisade and he demonstrated their @RISK product plug-in for MS Project.

::geek mode on::

@RISK allows you to use probabalistic modeling for any or every task in your schedule. This means that instead of saying that a task will take 5 days, you can use statistical language to describe the likely range of days. Once this has been done for a number of tasks, then the model of the entire schedule can be described using probability.

::geek mode off::

This concept is fundamental to successful project management. Because of the uncertainty inherant in projects, it is insane to try to predict exact task durations. We might as well pretend to be psychics wearing gypsy headscarves. However, we can hone our estimation skills to the point where we can make educated predictions.

The @RISK product is cool because it removes much of the emotion from the calculation of date ranges. However, it is important to note that the assignment of the probability distribution to each task is still a human activity. There’s no magic formula or supercomputer that can tell you exactly how likely a given outcome will be. And thus, garbage in/garbage out.

The take-home lesson is that when you’re planning a project, don’t get trapped into giving your management an exact date when the project will be complete. Provide a range of dates that accurately describes the amount of uncertainty. If it’s early in the project and much is still unknown, say “second half of the year” or “third quarter.” As you refine the project plan and remove some of the uncertainty, you can provide more granularity by stating the month when you expect to complete the project. The only time you can be certain of your completion date is the day you’re done.

Much more information is available in Steve McConnell’s Rapid Development and Software Estimation: Demystifying the Black Art.