Test allocation of controllable and residual risks
7 min read
Lesson objectives
- Make a documented decision on test allocation of controllable and residual risks with evidence and limits.
Please sign in to use all the tools
Once you sign in, the reading tools are yours: larger or smaller text, typeface, line spacing, focus mode, highlighting and bold, and notes saved to your account and gathered in My study.
Before a service agreement is signed, a team proposes making the supplier bear “all outage risk.” The phrase sounds decisive, but it cannot make the supplier open a site gate controlled by the buyer or repair an electrical network outside its control. Risk allocation is a decision about who can prevent an event, reduce its impact, or detect it early, and which evidence and incentives make the allocation workable. The effect that remains after controls is called residual exposure here. Assigning risk in a management exercise does not resolve legal liability; that depends on the executed terms, applicable law, and actual event record.
Describe a risk as cause, event, and consequence rather than an empty label such as “supplier risk.” For example: “If a cooling component with no local spare fails, restoration at the critical site may take longer.” Component dependence is the cause, failure is the event, and prolonged outage is the consequence. This wording lets you test distinct options: stock, a second supplier, predictive maintenance, or temporary operating capacity. Each has cost, implementation time, and limits. Record who holds stock information and who can make a real decision, then separate prevention from response after the event.
It is not enough to say that the supplier is “best placed” to manage a risk because it has technical expertise. It may be able to buy a spare but only the buyer may approve a pump-design change. A workable allocation might split action: the supplier holds defined stock and reports a reorder threshold; the buyer decides on design change at a recorded gate; both test temporary operation. Each act needs different evidence. A single register row saying “supplier owns continuity” conceals a bottleneck in buyer approval. Transferring a risk that one party cannot influence can also change price or reporting incentives. Treat that as a commercial possibility to examine, not as a universal numerical law.
Assess a risk on two stated dimensions: an approximate likelihood grounded in experience or records, and the consequence if it occurs. An organisation may use a qualitative matrix, but a score is not a natural fact. If “medium by high equals red,” state what the boundaries mean and which action the colour triggers. Do not confuse a future possibility with an issue that has already happened. Two late parts last month are recorded events or issues; stock exhaustion next month is a risk. Connecting them requires stock, demand and replenishment evidence. If they share one row, someone may close tomorrow’s risk merely because yesterday’s late delivery was resolved.
Test a control with two questions: does it reduce likelihood or consequence, and how would we see it working? A spare part reduces the consequence of a failed installed part only if the spare is usable, available, and quick to fit. It does not reduce the chance that the installed part fails. Inspection might lower unexpected-failure likelihood or reveal a fault earlier, but cannot replace a contingency when interruption remains severe. Evidence of effectiveness is more than a written plan: it could include an installation test, inventory record, or safe transition drill. Then describe residual exposure, the person authorised to accept it, and its warning indicator.
Where inputs are uncertain, scenario analysis is more honest than a spurious precise risk score. Suppose a part normally arrives in five days, but could take 12 days under a shipping disruption, while reserve stock supports seven days of operation. In the ordinary case the reserve covers delivery with a two-day margin. In the stressed case the gap is five days. This does not automatically justify buying more stock. Ask about carrying cost, part borrowing, outage consequence and storage life. The calculation shows where residual exposure lies and what information would change the decision. The sound conclusion is that “current cover does not span a 12-day scenario,” not that failure is certain.
Lesson summary
Reference answer: the observed issue is that the seven files were absent from the restorable set, if a restoration test establishes that fact. A “30 of 30 jobs run” report only measures execution against the existing selection list. The future risk is that other unlisted files may be lost or unrecoverable. A shared control could reconcile the approved classification list to the file inventory periodically and test restoration on a sample. To locate the break, obtain the version of the list the buyer sent, the supplier’s receipt, selection settings, and restore-test result. Do not declare the supplier innocent just because the buyer classifies files, or liable simply because files were not copied: the signed wording might impose a warning or checking duty. Protect current files, verify restoration, and send contested allocation questions to qualified specialists. A good answer separates an event from a prospective exposure and gives evidence for control effectiveness.