Distribution for risk impact

[Ken]

When should you use a probability distribution instead of a single value for impacts? Are there any rules or guidelines about when you should use statistical distributions instead of singe values for risk impacts?

[Intaver Support]

There really are no established rules about when you should use on over the other. However, when we do a risk assessment generally for impacts we look at the answer to the question “If risk x occurs, what will happen”. If this a well known risk and impacts a specific WBS or activity and the answer is “we will have to replace this widget and it will cost this and take this long to fix.

If the answer is, “We have ordered these parts many times and the cost has not varied in the past couple of years and we have performed this activity many times” the you can feel very confident in have a single impacts for cost or schedule.

However, if the answer is, “ Last time we went to order the parts from procurement, they weren’t available from our regular supplier, so we had to source them elsewhere, They cost a bit more and the procurement process caused delays. We also had to revise some of our procedures because the equipment was not exactly the same”. Then this is a good opportunity to try estimate best case and worst case for both schedule and cost and use this to define a probability distribution for what he impacts of the risk could be.

In our experience, if you ask enough questions, most risk impacts will be assessed as probabilistic