The Psychology of Quality and More
Advanced Risk Analysis
When assessing how likely a risk is to occur, the first problem that may be met is that you do not know or are not sure. A probability figure can consequently be moderated by a confidence score. For example you may conclude that there is a 50% probability of the risk happening, plus or minus 20%, giving a possibility range of 30% to 70%, which effectively says ‘I don’t think it’s very likely or unlikely – it’s in the middle somewhere.
Probabilities do not necessarily stay the same. The chance of process failure may be assessed as low, providing certain preconditions are met, such as having sufficient trained people available. These often relate to trigger events, such as key people resigning or late deliveries from suppliers, which significantly change the probability of the risk occurring. When key trigger events are identified, then these can be treated as risks in themselves and actions taken to reduce the chance of them occurring. It is also important to monitor trigger events as these should result in re-assessment of the risks.
When a risk occurs, problems may occur in a number of different ways. It can
be likened to a bomb landing, the metaphor of which provides useful
After depth, you can ask about breadth. Does the bomb do local damage or is it widespread? Is it one customer who is affected or is it all? Would problems be limited to this process or is there a significant knock-on effect? Significant breadth in impact multiplies the severity. This is one reason why containment action is often important.
Another concern is manageability of the situation. Sometimes when risks occur you can do little to contain or control things, and simply have to wait and mop up afterwards. Knowing this significantly changes how you plan.
A further dimension is time, and the probability of a risk occurring can change with time, such as when the time moves beyond trigger event dates.
In terms of impact, some risks are short and sharp, whilst other cause problems over a longer period, like the difference between explosive and atomic weapons. Understanding this time dynamic is a further task of analysis, where effects such as delayed effects and ripples can occur. Recovery also has a time element to it, which may need to be balanced against cost.
Risks can also have a different impact depending when they occur, for example a parts shortage may well have a greater impact closer to a delivery date as opposed to if it occurs early in the process cycle.
Another dimension is uncertainty, which can occur in many places during analysis beyond the basic confidence figure in probability identified above.
A common problem is visibility of the risk, where you may be able to see the risk and trigger events coming from a long way off as opposed to the more desirable state of knowing when it will happen. Common action to increase the visibility of risks includes use of measurement, such as monitoring tool wear to be able to predict out-of-specification parts.
When the manageability of the situation is low, then uncertainty also increases. This includes for mitigation and monitoring actions before the risk occurs, as well as in contingency action after the risk has occurred (as described above).
Putting deep attention into risk analysis as described above takes a lot of time and effort. This effectively means more advanced analysis is often only feasible for a limited number of situations. However, for critical risks, taking time to truly understand the multi-faceted nature of the risk can pay dividends in quality.
Next time: Assumption Management
This article first appeared in Quality World, the journal of the Chartered Quality Institute
And the big