At 02:14 on a Thursday morning, the main finishing line stops.
The immediate cause is clear: a critical drive assembly has failed. The plant team knows the repair may take three days. The production schedule shows customer orders due across the next two weeks. The first executive question arrives before the first maintenance estimate is complete:
How much customer commitment can we still protect?
That question is more useful than “How quickly can we repair the line?” A fast repair may be the right answer, but it is only one part of the response. Leadership also needs to know whether production can be moved, whether another plant is qualified, whether inventory is usable, whether suppliers can support the alternative, whether transport capacity exists, and which customer promises should receive priority if the network cannot protect them all.
Two networks can experience the same outage and face completely different futures.
The first has an alternate route, qualified capacity at another plant, a supplier who can accelerate a replacement, and an agreed recovery playbook. It loses efficiency, but it protects its most consequential commitments.
The second has a contingency document in a shared folder. The alternate plant has never run the product. The supplier’s lead time is theoretical. The inventory is in the wrong location. Nobody is sure who can approve the change. Its current service metric looked just as strong as the first network’s before the failure.
This is the difference between reliability and resilience.
Reliability describes how consistently the network performs under expected conditions. Resilience describes how well it absorbs disruption, preserves important options, and recovers without abandoning the commitments that matter most. The Network Resilience Index—NRI—provides a way to measure that capability before the outage becomes a customer crisis.
Reliability is not resilience
A reliable supply chain is not automatically a resilient one.
The network may deliver on time every day because demand is stable, the primary supplier is performing, the main plant has available capacity, and the usual transport routes are open. Those conditions can create excellent results while hiding concentration risk.
Reliability asks: “How often does the normal operating system produce the expected outcome?”
Resilience asks: “What happens when the normal operating system cannot produce the expected outcome?”
The distinction matters because disruption changes the decision environment. Under normal conditions, the organization can optimize cost, utilization, inventory, and schedule efficiency. Under stress, it must preserve optionality. It may need to use a more expensive supplier, a less efficient route, a different production sequence, or a temporary specification. The ability to make that transition is a capability, not a hope.
Many organizations confuse preparedness with documentation. They have business-continuity plans, risk registers, escalation trees, and supplier questionnaires. These are useful only if they correspond to real, executable options.
A contingency plan that names an alternate supplier but has no qualification, price, capacity, or logistics pathway is not an available option. It is a possibility. A resilience measure should distinguish between what exists on paper and what can be activated within the time available.
The same principle applies to inventory. A large stock balance may look protective, but the material could be reserved, obsolete, in quarantine, in the wrong geography, or unusable without a missing component. Resilience depends on usable options, not nominal assets.
What NRI measures
NRI answers a practical question:
If a meaningful disruption occurred today, how much important demand could the network protect, for how long, and at what cost?
It should consider the network’s ability to sense the disruption, interpret its consequences, choose an alternative, authorize the response, execute it, and recover. A resilience score that measures only supplier count or spare capacity is incomplete because the option may not be usable in practice.
The NRI lens can include:
- Disruption exposure at plants, suppliers, ports, warehouses, and technology dependencies.
- Qualified alternate sources and the time needed to activate them.
- Alternate routes, carriers, modes, and border options.
- Spare capacity by product, process, asset, and time window.
- Usable inventory and its geographic position.
- Recovery time and recovery point for critical flows.
- Third-party criticality and substitutability.
- Operational playbooks, decision rights, and escalation speed.
- Financial cost of alternate capacity, freight, inventory, and supplier activation.
- Evidence from exercises, incidents, overrides, and actual recovery outcomes.
NRI is not a universal resilience grade. It is a view of the commitments and options connected to a defined network, time horizon, and disruption scenario. A network may be resilient for standard products and fragile for engineered products. It may be resilient for a regional customer and exposed for a global launch. The unit of analysis matters.
The network facing the same outage
Consider two manufacturers with similar size, customers, and current OTIF performance. Both lose a critical finishing line for 72 hours.
Network A has a qualified second plant that can absorb 40 percent of the affected volume. It has a documented but practiced route for transferring semi-finished material. Its quality team has already agreed the release process. A regional supplier can provide the failed component within 24 hours, although at a premium. The logistics team has access to a second carrier and knows which customer orders should move first.
Network B also lists a second plant in its contingency plan. But that plant has no current qualification for the product family. The material specification has changed since the plan was written. The only alternate carrier requires a contract approval that may take a week. The spare component exists, but the inventory record does not identify its exact location. Sales, operations, and quality disagree about which customers should be protected.
Before the outage, both networks may show the same delivery performance. During the outage, their options are not comparable.
Network A will pay more and operate less efficiently, but it can preserve a defined share of strategic commitments. Network B will spend its first two days discovering what it is allowed to do. The disruption becomes larger because decision latency consumes the recovery window.
This is why resilience must be tested as a capability. The question is not whether a contingency plan exists. It is whether the organization can execute the plan under pressure.
Alternate routes and qualified options
Redundancy is often described as having two of something. Two suppliers, two plants, two carriers, two ports. But duplication alone does not create resilience.
The alternate must be able to perform the required work at the required standard within the relevant time window. That means qualification, capacity, specifications, tooling, labor, technology, quality release, commercial terms, and transport feasibility may all matter.
For supply routes, leaders should ask:
- Is the alternate route open for the product, destination, and customer terms?
- Can the carrier provide capacity when the primary route is constrained?
- Are customs, documentation, insurance, and regulatory conditions understood?
- Does the alternate route meet the required temperature, security, or handling standard?
- How quickly can the route be activated?
- What is the cost, and who has authority to approve it?
For alternate sources, the questions include:
- Is the source technically and quality qualified?
- Can it produce the required grade, configuration, and volume?
- Does it have the tooling, raw materials, and labor available?
- Is the customer or regulator required to approve the change?
- How long will first usable supply take?
- What happens to cost, margin, and working capital?
An option should have an activation condition and an owner. Otherwise it becomes an item in a risk register rather than a decision-ready capability.
Spare capacity and recovery time
Spare capacity is not simply unused equipment percentage. It is the ability to produce the right output, at the right quality, within the right period, after the disruption.
A plant may have 20 percent theoretical capacity and still be unable to absorb an outage because the spare capacity is on the wrong line, requires a long changeover, lacks a critical operator, or is already committed to a different product. Capacity must be considered by process step and constraint, not only by aggregate site utilization.
Recovery time is equally important. Leaders need to know how long the network can operate in a degraded state before customer commitments become impossible. A one-day disruption may be manageable if inventory and alternate capacity cover the gap. The same disruption may become critical after three days if the recovery route depends on a weekly sailing or a quality release.
Useful recovery questions include:
- What is the first customer-impacting milestone?
- Which commitments can be protected with current inventory?
- Which decisions must be made within four hours, 24 hours, or 72 hours?
- How much alternate capacity is available in each window?
- What is the recovery time objective for the critical flow?
- Which actions reduce the recovery window, and which only delay the visible failure?
NRI should make the time dimension visible. A network with adequate options next month may still be unable to protect a customer due tomorrow.
Third-party criticality
Resilience is increasingly shaped by dependencies outside the organization’s direct control.
The obvious examples are tier-one suppliers, carriers, ports, utilities, and contract manufacturers. The less obvious dependencies include specialist maintenance providers, testing laboratories, cloud systems, software licenses, packaging suppliers, payment services, and data feeds. A third party becomes critical when its failure can interrupt an important decision or flow and no practical substitute is available within the required time.
Third-party criticality should be assessed by consequence and substitutability, not only spend. A low-spend specialist may be more critical than a large commodity supplier. A service provider may not touch the product but may control the release, authorization, or visibility required to move it.
For each critical third party, the organization should understand:
- What decision or flow depends on it?
- What is the maximum tolerable interruption?
- What alternatives exist, and are they qualified?
- What information will be available during the disruption?
- What contractual, regulatory, or data constraints apply?
- What evidence shows that the recovery option works?
This is where NRI connects to the wider decision intelligence layer. A disruption signal is useful only when the organization can connect it to the commitments, alternatives, and decision rights it affects.
The cost of resilience
Resilience has a cost. Alternate suppliers require qualification. Spare capacity reduces utilization. Buffer inventory ties up cash. Multiple carriers may increase procurement complexity. Exercises consume time. More options can create governance and maintenance work.
The right question is not whether resilience is free. It is whether the investment is proportionate to the consequence it protects.
This is why NRI should be read with other indices. A resilience improvement may reduce working-capital velocity if it adds inventory. It may lower margin integrity if the alternate source is expensive. It may improve ONRI by protecting strategic promises. The trade-off should be explicit.
An executive review can compare:
- The expected cost of resilience.
- The likelihood and consequence of the disruption.
- The share of important commitments protected.
- The recovery time with and without the investment.
- The impact on margin, inventory, quality, and customer trust.
- The reversibility of the decision.
Resilience is not a request to duplicate everything. It is a discipline for deciding where optionality is worth paying for.
Using NRI in disruption exercises
The best way to measure resilience is to exercise it.
An exercise can be simple. Choose one critical flow and introduce a plausible disruption: a plant outage, supplier failure, port closure, cyber incident, quality hold, or transport capacity loss. Freeze the starting conditions. Then ask the teams to make real decisions using the data, permissions, and contacts they would have during an actual event.
The exercise should test more than the existence of a plan. It should reveal:
- How quickly the disruption is detected and understood.
- Whether the affected commitments can be identified.
- Whether inventory and capacity data are trustworthy.
- Whether alternate options are genuinely available.
- How long approval takes.
- Whether functions agree on customer priorities.
- Whether the alternative creates quality, regulatory, or safety risk.
- What the recovery action costs.
- Whether the organization captures learning afterward.
The outcome should be an evidence-based NRI view. How much critical demand could be protected in the first 24 hours? What share could be protected after 72 hours? Which options failed because they were not qualified, not available, not authorized, or not known?
Exercises also expose social weaknesses. A team may have a good supplier alternative but no agreed rule for allocating its limited output. It may have inventory but no confidence in the records. It may know what to do but lack the authority to act without a meeting. These are resilience gaps even when the physical network appears redundant.
Connecting NRI to other indices
NRI should not become another isolated score.
It connects naturally to the On-Time and In-Full Network Reliability Index. ONRI shows which customer promises matter most; NRI shows how much of those promises the network can protect when conditions change. A high ONRI with low NRI may mean the business is delivering well today but is exposed tomorrow.
NRI also connects to Supply Chain Exposure or concentration measures. A network may have alternate capacity in theory but still depend heavily on one supplier, route, geography, or technology. Exposure explains where resilience investment may be needed.
Working Capital Velocity matters because resilience often uses inventory. The key question is whether the inventory is targeted, usable, and positioned for the commitments that need protection. Undifferentiated stock can create cost without creating meaningful recovery capability.
Margin Integrity matters because resilience actions can be expensive. Premium suppliers, expedited logistics, overtime, and small production runs may protect service while consuming profit. The decision is not “resilience or margin”; it is “which resilience investment protects enough value to justify its cost?”
Together, these relationships create a more complete picture:
- ONRI: Which promises matter most?
- NRI: How much disruption can we absorb while protecting them?
- Exposure: Where are we concentrated or dependent?
- WCVI: What cash and inventory does resilience consume?
- MII: What does protection cost?
Governance and versioned definitions
Resilience scores are vulnerable to optimistic assumptions. If an alternate supplier is counted before qualification, the score will exaggerate capability. If spare capacity is counted without labor or tooling, the score will exaggerate capability. If a contingency plan is treated as executable without an exercise, the score will exaggerate capability.
NRI needs a versioned definition and an evidence standard. Each option should have a status such as identified, assessed, qualified, exercised, activated, or retired. The status should include an owner, last validation date, expected activation time, constraints, and cost.
The model should also distinguish confidence from capability. A team may believe an alternate route is available but have no current confirmation. That is a low-confidence option. NRI should not treat belief as evidence.
Definitions must state the scenario and time horizon. “Can we absorb a disruption?” is too broad. “Can we protect strategic customer schedule lines for seven days after a single-site outage?” is measurable. Different scenarios will produce different results, and that is useful.
Governance should prevent goalpost-moving. A resilience score must not improve merely because the organization changed what counts as an available option. Changes to assumptions, weights, or evidence thresholds should be recorded and communicated.
Questions for leaders
In a disruption review, leaders should ask:
- Which important customer commitments are protected in the first 24 hours?
- Which commitments become exposed after 72 hours or seven days?
- What is the first decision that must be made, and who can make it?
- Which alternate options are qualified and tested rather than merely listed?
- Where is the network dependent on one supplier, asset, route, person, or system?
- Which inventory is genuinely usable and correctly positioned?
- What capacity can be activated, and what constraints limit it?
- What does the recovery option do to margin, quality, working capital, and customer trust?
- When did the organization last exercise this scenario?
- What evidence would make us revise the NRI result?
These questions turn resilience from a policy document into an operating capability.
Start with one critical flow
Organizations do not need to model the entire global network before beginning. Choose one flow where a disruption would create visible customer or business consequence.
Map its dependencies from supplier through customer. Identify the important promises, the first irreversible milestones, the usable inventory, the qualified alternatives, the recovery time, the owners, and the approval boundaries. Run a tabletop exercise, then a more realistic test if the stakes justify it.
The first NRI can be a simple evidence-based assessment. It can show the share of weighted commitments protected under a defined scenario, the time to activate alternatives, and the gaps that prevent execution. The score matters less than the decisions it prompts.
Over time, connect the result to actual events. Did the exercise predict what happened? Did the alternate work? Did the team act within the required window? Did the recovery protect the intended customers? Did it create an unacceptable quality or financial consequence?
That learning improves both the index and the network.
Resilience is the ability to keep important promises
Disruption is not a sign that a supply chain has failed. Disruption is a condition every serious network must expect.
The real test is what happens next. Can the organization identify the commitments that matter? Can it see the dependency chain? Can it activate a qualified alternative? Can leaders make a priority decision before the recovery window closes? Can the network absorb the cost without creating a larger failure somewhere else?
NRI provides a structured way to ask those questions before the crisis.
It does not reward a large inventory balance, a thick risk register, or a contingency plan that nobody has tested. It rewards usable options, clear authority, credible information, and evidence that the organization can recover under pressure.
Resilience is not the absence of disruption. It is the ability to absorb disruption without abandoning important promises.
Disclaimer
Industry situations in this chapter are composite illustrations unless explicitly attributed to a public source. They are not claims about any particular company, plant, vendor, or incident. External standards, research, and public case studies should be verified before publication. Implementations must be validated against local safety, quality, cybersecurity, regulatory, contractual, labour, privacy, and data-governance requirements. AI recommendations and autonomous actions should remain within clearly defined human authority, operational controls, and tested recovery procedures.
#SupplyChain #SupplyChainResilience #BusinessContinuity #SupplyChainRisk #DisruptionReadiness #OperationsManagement #Procurement #Logistics #Manufacturing #DecisionIntelligence
Takeaways
Excerpt | Practical point / context |
|---|---|
“How much customer commitment can we still protect?” | The first useful question after a disruption is about protected commitments, not only repair speed. |
“This is the difference between reliability and resilience.” | Normal execution and disruption absorption are different capabilities. |
“A contingency plan that names an alternate supplier but has no qualification, price, capacity, or logistics pathway is not an available option.” | Plans should be measured by executability, not documentation. |
“Resilience depends on usable options, not nominal assets.” | Inventory and capacity count only when they can support the required flow. |
“An option should have an activation condition and an owner.” | Every recovery path needs a trigger and accountability. |
“Resilience is not a request to duplicate everything.” | Investment should be proportional to the consequence being protected. |
“The best way to measure resilience is to exercise it.” | Exercises reveal gaps in data, authority, qualification, and decision speed. |
“A disruption signal is useful only when the organization can connect it to the commitments, alternatives, and decision rights it affects.” | Detection must lead to context and action. |
“It does not reward a large inventory balance, a thick risk register, or a contingency plan that nobody has tested.” | NRI focuses on credible capability rather than appearances. |
“Resilience is not the absence of disruption. It is the ability to absorb disruption without abandoning important promises.” | The central takeaway of the article. |
Further reading
- The Decision-Centric Supply Chain: Why AI Should Optimize Decisions, Not Dashboards
DATTS
NRI is valuable only when a disruption signal is connected to affected commitments, alternatives, decision rights, and a timely response. The decision-centric supply-chain article provides the broader principle that AI and metrics should optimize choices rather than merely display alerts.
- Intelligent Supply Chains · Chapter 8 · The Rise of Decision Products
DATTS
The ingredients of a usable recovery capability: qualified options, owners, activation conditions, constraints, and action. The article shows how to maintain those elements around a recurring operational decision
- AI-Native Projects · Chapter 16 · From Data Lakes to Decision Lakes
DATTS
NRI exercises should capture what the team knew, which alternatives it considered, what it chose, how long approval took, what the intervention cost, and whether recovery worked. The decision-lake article is directly relevant because it explains how to preserve this context for audit and learning.
- ONRI: Measuring the Delivery Promises That Really Matter
DATTS
NRI becomes useful when analysed along with ONRI
