The Single Line That Stopped
Arclight Diagnostics runs a single qualified production line for the reagents used in its screening kits. The risk register carries one entry for supplier failure: probability 30%, impact $400,000, response recorded as accept, with a $150,000 contingency reserve held at programme level. Nine days ago the supplier's only qualified line shut down unplanned; the shipment arrived three days late and was absorbed by the four-day inventory buffer. The risk owner, the operations director, has not touched the entry since it was created four months ago. Separately, procurement has just signed a 12-month volume commitment with the same supplier to secure a 7% discount worth about $95,000. The supplier's contract caps liability for late delivery at the value of the affected order. A notified labelling regulation will also require a package redesign within eight months, and that work is not on the register at all.
1. Assess whether the register entry and the contingency reserve are adequate for this risk. Show your reasoning, including the expected monetary value.
Read the worked answer
What the entry gets right. It exists, it is owned, and it is quantified. What it gets wrong:
- EMV is being used as if it were a reserve. EMV = 0.30 × $400,000 = $120,000. That is the long-run average loss, not the amount needed if the event occurs. The reserve of $150,000 is 37.5% of the impact - it covers the expected value but leaves the programme $250,000 short if the risk materialises. For a low-probability, high-impact, fat-tailed risk, size the reserve against a confidence level (for example P80 of a Monte Carlo over the register), not against EMV.
- The probability is stale. A nine-day unplanned shutdown of the only qualified line is the very event the register described in weaker form. It is a near-miss and it is free information: both the 30% estimate and the four-day buffer need re-testing. A three-day delay consumed 75% of the buffer - what is the buffer against a nine-day outage?
- Accept is recorded with no trigger. Acceptance is legitimate only when there are defined early-warning indicators and a decision rule. There are none.
- The impact is one-dimensional. $400,000 ignores the regulatory and clinical consequence of a reagent shortage in a diagnostics supply chain, which may dominate the financial figure.
Judgement: the register is compliant in form and inadequate in substance.
2. Procurement argues the 7% discount is a clear win and the volume commitment reduces price risk. How do you evaluate that judgement?
Read the worked answer
Evaluate it on three questions, not on the discount.
- Does it reduce or transfer risk? A 12-month volume commitment converts a supply risk into a contractual obligation. If the supplier fails, committed spend cannot be redirected, and liability is capped at the value of the affected order - far below a $400,000 impact. The commitment increases correlated exposure while the discount is roughly 24% of a single realised impact event.
- What is the counterfactual? A discount is only a win relative to the alternative of dual-sourcing or spot pricing. If qualifying a second source costs less than $95,000, the discount is buying convenience at the price of resilience.
- Who owns the decision? This is a commercial decision with risk consequences. It should go to the risk owner jointly with procurement on a documented risk assessment, not be signed as a procurement saving.
Recommendation: accept the discount only with (a) a renegotiated liability and service-credit schedule, (b) a contractual exit for sustained non-performance, and (c) a funded second-source qualification running in parallel. If the supplier refuses (a) and (b), the discount is compensation for concentration risk rather than a saving, and should be priced and reported as such - a $95,000 benefit against a $400,000 unhedged exposure is not a win.
3. What should happen in the next two weeks, and who owns each action?
Read the worked answer
Two weeks, four owners.
- Risk owner (operations director), this week: re-score probability and impact using the near-miss; document the nine-day outage, the buffer consumption and the missed signal. Define triggers - days of finished-goods inventory, supplier line availability, order-book lead time - and the decision rule that fires at each.
- PM, week 1: run a reserve adequacy review (scenario analysis or Monte Carlo at P80) and take any top-up request to the programme board. Re-baseline the inventory buffer against a nine-day outage rather than a three-day one.
- Procurement, weeks 1–2: open the liability cap and service credits; price a dual-sourcing option and bring it to the board with the discount comparison.
- Quality and regulatory, week 2: add the labelling change as a separate regulatory risk with a hard eight-month deadline. A package redesign carries validation and approval lead time that will not compress, so its trigger date is nearer than it appears.
- PM, week 2: present a one-page risk review to the board - near-miss analysis, revised scores, reserve position, decision requests.
The lesson to record is that a near-miss is a control test, not a lucky escape; the buffer worked, but only because the outage was three days rather than nine.