The pattern is now well established. A disruption occurs. The organisation responds heroically, at considerable unplanned cost. A post-mortem concludes that resilience must improve. A programme is initiated. Somewhere between twelve and twenty-four months later it is quietly deprioritised in favour of initiatives with a clearer return.
This is not organisational amnesia, though it is often described that way. It is the predictable consequence of framing resilience as a value to be upheld rather than an investment to be justified. Values do not compete well against projects with a calculated payback, and they should not be expected to.
The actuarial framing
Insurance has solved this problem. You do not insure a building because safety is a value; you insure it because the premium is less than the probability of loss multiplied by the size of the loss. The same arithmetic applies to supply chain mitigation and is used far less often than it should be.
For each identified failure point, three quantities are needed: the probability of occurrence over a defined period, the loss given occurrence, and the cost of the mitigation. Where mitigation cost is less than expected loss, buy it. Where it exceeds expected loss, accept the risk explicitly and record that you have done so.
An accepted risk that has been named, quantified and signed off is a managed risk. An unexamined one is an accident waiting for a post-mortem.
Time to recover and time to survive
The most practical framework for quantifying loss given occurrence comes from work done at MIT and subsequently refined in industry, and it rests on two measures.
- Time to recover: how long until supply from this node is restored, whether by repair, requalification of an alternative, or substitution.
- Time to survive: how long you can continue serving customers without it, given inventory, alternative sources and available substitution.
Where time to survive exceeds time to recover, the exposure is covered. Where time to recover exceeds time to survive, the gap between them multiplied by the revenue and margin at risk during that gap gives a defensible loss figure.
The great virtue of this framing is that it decouples the analysis from probability estimation, which is where most risk exercises get stuck. You need not know how likely a fire at a particular plant is in order to establish that its failure would cost eleven weeks of a product line's revenue. That figure alone supports a decision.
The counter-intuitive result
Applied properly, this analysis usually reveals that the largest exposures do not sit with the largest suppliers. High-spend suppliers are visible, actively managed, dual-sourced and closely monitored. The dangerous nodes are typically low-spend, highly specialised, single-source components with long qualification lead times.
A specialised connector representing a fraction of a per cent of spend, made in one plant, requiring nine months to qualify an alternative, can stop a production line as effectively as the loss of a primary contract manufacturer. It is not on anyone's supplier review agenda precisely because its spend is trivial.
| Component | Annual spend | Time to recover | Time to survive | Exposure |
|---|---|---|---|---|
| Primary assembly | High | 6 weeks | 9 weeks | Covered |
| Packaging substrate | Medium | 4 weeks | 7 weeks | Covered |
| Specialised coating | Very low | 22 weeks | 5 weeks | 17 weeks at risk |
| Control module | Low | 14 weeks | 4 weeks | 10 weeks at risk |
Costing the mitigations honestly
Each mitigation carries a real and ongoing cost, and resilience business cases lose credibility when those costs are understated.
- 01Strategic inventory: the carrying cost of the buffer plus obsolescence risk. Straightforward to compute, and effective only for storable items with stable specifications.
- 02Dual sourcing: qualification cost, ongoing management overhead, and (usually the largest component and the most frequently omitted) the volume leverage forfeited by splitting the award.
- 03Flexible or reserved capacity: the option premium paid to hold capacity that is not used, whether internal or contracted.
- 04Nearshoring or regionalisation: the unit cost differential, which is generally permanent and should be capitalised over the horizon rather than treated as a transition cost.
- 05Design substitution: engineering effort and requalification cost, but with the distinctive advantage of eliminating the exposure permanently rather than insuring against it.
Building the frontier
With exposures quantified and mitigations costed, the output is an efficient frontier: for each level of resilience spend, the maximum expected loss removed. This is the artefact that survives a budget review, because it does not ask for a commitment to resilience. It asks a specific question with a specific answer, at this level of spend, this much expected loss is removed, and here is what remains accepted.
Two further disciplines make it stick. The first is to review the register on a fixed cadence rather than after incidents, since post-incident review guarantees over-investment in the failure that just happened and under-investment in the one that has not. The second is to test the plan live, at least annually, with a facilitated scenario exercise.
Every organisation that runs a live simulation discovers something its written plan missed, and it is rarely about supply. It is almost always about decision rights: who can authorise an expensive expedite at two in the morning, who can commit to a customer allocation, and how long the organisation waits for certainty before acting. Those failures are free to fix and invisible until tested.
