Supply chain resilience: designing for shocks you cannot forecast

Four mechanisms, four different bills

Buffering holds stock or capacity in reserve and pays for it continuously. Redundancy maintains alternative sources or routes and pays in lost scale and duplicated qualification. Flexibility builds the ability to switch products, sites or modes and pays in equipment, standardisation effort and training. Speed shortens the interval between an event and an effective response, and pays in monitoring and readiness. Most organisations lean heavily on one of the four and neglect the others, which leaves them protected against a narrow class of events.

Efficiency and resilience trade against each other honestly

Consolidating volume, removing buffers, sourcing from the lowest-cost location and running assets close to their ceiling all improve unit economics and all reduce the system's tolerance for disturbance. This is not a mistake to be corrected but a trade to be made deliberately, item by item and node by node. The failure is not choosing efficiency; it is choosing it everywhere without recording where the tolerance has been spent.

Exposure is concentrated in a few places

Resilience effort should follow the small number of positions where a single failure would stop a large share of output: a sole-source component, a plant with no alternative, a corridor that carries most of the inbound volume, a sub-tier input shared by several suppliers. Identifying those positions and treating each explicitly delivers far more protection per unit of spend than uniform improvement across a whole network.

Time to recover is the measure worth holding

For each critical node, the useful pair of figures is how long the business could continue if that node stopped, and how long restoring supply would take by the best available route. Where the second exceeds the first, an exposure exists that no amount of monitoring removes. Expressed this way, resilience becomes a comparable measure across the network and a basis for deciding where to spend, rather than an abstract quality.

Test the design before an event does

Assumptions about alternatives are unreliable until exercised. Running a scenario against a specific node — a supplier failure, a site closure, a corridor interruption — and following it through to the customer impact reliably finds contingencies that were never resourced and contacts nobody has spoken to in years. The output should be a small number of funded changes, not a document that circulates and expires.

Frequently asked questions

Is holding more stock the simplest route to resilience?

It is the fastest to implement and the least discriminating. Stock protects against interruptions shorter than the cover it provides and does nothing about longer ones, so it works best combined with mechanisms that shorten recovery, such as a qualified alternative source.

How is resilience justified to a finance function?

By expressing it as the cost of a named exposure rather than as a general quality. Stating what would stop, for how long, and what that would cost, against the annual price of the protection, turns the conversation into an insurance decision that can be accepted or declined on its merits.

Does regionalising supply automatically improve resilience?

It shortens some exposures and concentrates others. A regional structure reduces dependence on long corridors but increases sensitivity to events within the region, so the benefit depends on which hazards the business is actually exposed to.