What to read first: Worst case analysis: When, why, and how
You have identified a risk. The consequences of that risk are unpredictable. The main post recommended finding the ‘reasonable worst case’ and analysing that first. You could then proceed to other ‘cases’, each one less severe but more likely than the worst case.
You might prefer to start with the most likely case. You might favour credibility and realism over scaring your stakeholders. Even the reasonable worst case might seem a bit too scary to talk about.
If you were to recognise the risk only from the most likely case, you would miss the risk of worse cases. The risk from those worse cases is less likely to be acceptable, and if you ignore them, you won’t even know that.
In the product launch example, the ‘Extreme’ case was Actual average sales fall below 50% of break-even volume. Its consequence for profitability was rated as ‘worst imaginable’ and its likelihood of realisation was indicated at 5%. So by itself, this ‘Extreme’ case was a 5% chance of a ‘worst imaginable’ outcome.
The risk from this ‘Extreme’ case alone might well have been unacceptable to the enterprise, even if the risk from the reasonable worst case had been considered acceptable. The risk from the reasonable worst case was a 15% chance of a ‘disappointing’ outcome.
You might also consider the ‘average’ case as a single representative case.
At best, you will come up with an ‘average’ case with a single consequence and likelihood. That best kind of ‘average’ case has the same mathematically expected consequence as the varied cases modified by their varying likelihoods.
This ‘average case’ might generate a ‘level of risk’ with which you feel comfortable.
But in the meantime, you have thrown away all of the effects of uncertainty around the size and consequence of the risk event. ‘Risk’ is the effect of uncertainty, and risk management exists to recognise it. That recognition includes understanding that a 5% chance of an un-survivable event is not the same as a 20% chance of something that just nasty, but survivable. Neither of those are equivalent to a 45% chance of a disappointing result for the year. The problem with the ‘average case’, and with a scale for ‘level of risk’, is that they treat all risks with the same ‘expected consequence’ or ‘level of risk’ as identical. They are not. The difference is, precisely, risk. Risk is the effect of uncertainty on objectives. Averaging throws away the uncertainty.
To model risk as the average case you first need to confuse ‘risk’ with the long-term incidence of bad stuff. Don’t do that.