The dashboard turns red.
On-Time and In-Full Network Reliability has fallen from 94 to 86. The number is visible to everyone in the operating review, but the room does not agree on what it means.
Sales says the problem is an unrealistic promise date. Planning says the forecast changed too late. Operations says a constrained line failed. Procurement points to a supplier delay. Quality says released inventory was not usable. Logistics says the carrier window was missed because the order was not ready.
Everyone sees the same red number. Everyone proposes a different cause.
This is the point at which an index can either improve decision quality or create another argument. A score is a signal, not a diagnosis. It tells leaders that something deserves attention. It does not tell them what happened, why it happened, what will happen next, or which action is justified.
The value of an index is measured by the quality and speed of the conversation it enables.
This article provides a practical sequence for moving from index movement to disciplined investigation and action. The method applies to ONRI, NRI, SCEI, DFAI, QRI, YMEI, TEEI, MII, WCVI, and CAI. The score starts the conversation. The decision sequence turns it into operating value.
A score is a signal, not a diagnosis
Indexes are useful because they compress complexity. Leaders cannot review every order, batch, supplier event, inventory position, or production deviation in a single meeting. A score creates focus.
But compression creates risk. The score hides the detail that explains it. A falling ONRI may result from one strategic customer miss or many small routine misses. A lower NRI may reflect one concentrated supplier or a broad deterioration in alternatives. A falling WCVI may mean inefficient stock, deliberate resilience inventory, or delayed invoicing.
The first discipline is to resist the urge to explain the number immediately.
The review should ask:
1. What exactly changed?
2. Compared with which baseline?
3. Where is the movement concentrated?
4. How confident are we in the data?
5. What decision window remains?
This prevents the meeting from becoming a contest between the loudest functional narrative and the most familiar explanation.
The common decision sequence
A practical operating review can use six questions:
1. What happened?
2. Why did it happen?
3. What is likely to happen next?
4. What options remain?
5. What action is justified?
6. What did we learn?
The questions should be answered in order. Jumping to action before understanding the signal creates expensive corrections. Staying in diagnosis after the decision window closes creates a different failure: the organization knows what happened but loses the ability to change the outcome.
The sequence is deliberately simple. Its value comes from making the reasoning visible and assigning ownership at each step.
What happened?
The first task is to describe the movement without explanation.
For a falling ONRI, the team should state the period, population, baseline, and scope. Did the result fall across the network or in one plant? Did it affect all customers or one strategic account? Did original-promise reliability fall, or only latest-date performance? Were schedule lines short, late, or both?
Useful evidence includes:
- The index value and prior trend.
- Raw measure alongside weighted measure.
- Number and value of affected commitments.
- Customer, product, plant, supplier, route, and schedule-line distribution.
- Original and revised dates.
- Quality, inventory, capacity, and logistics state.
- Data freshness and missing fields.
The output should be a fact pattern, not a story. For example: “ONRI fell six points because 12 high-criticality schedule lines in two product families missed their original dates. Eight were late, three were short and late, and one was held for quality. Seven affected lines relate to one plant.”
The fact pattern should include data confidence. If the customer criticality field is incomplete, if quality status is delayed, or if the inventory balance includes stock that has not been released, the team should say so. Uncertainty is part of the situation. Hiding it creates false confidence in the diagnosis.
The same discipline applies when the score improves. A better result may reflect genuine recovery, a favorable mix, revised dates, a definition change, or delayed recognition of an issue. The review should understand improvement as carefully as decline.
That statement creates a better starting point than “service was poor because planning missed the forecast.”
Why did it happen?
Once the fact pattern is clear, the team can investigate causes.
Cause should be traced through the decision chain rather than assigned to the last visible failure. A late shipment may begin with a customer mix change, an unqualified material assumption, a quality hold, a missed production sequence, or a transport cut-off.
The investigation should distinguish:
- Trigger: what changed first?
- Constraint: what prevented the normal response?
- Decision: what choice was made?
- Execution: what happened after the choice?
- Consequence: which customer, cost, cash, or risk effect followed?
This structure prevents “supplier delay” from becoming a complete root cause when the real issue was that the business had no qualified alternative or did not escalate early enough.
Evidence should be gathered at the level of the decision. If the score moved because of a schedule-line miss, investigate that line’s promise, inventory, quality, production, supplier, route, and owner. Do not rely only on the aggregate plant report.
What is likely to happen next?
Diagnosis looks backward. Decision-making must look forward.
After understanding the cause, leaders need to estimate the next consequence. Will the issue affect more orders? Will inventory cover the gap? Will a supplier recover? Will a quality hold release? Will a customer accept a revised date? Which intervention window is closing?
The forecast does not need to be perfect. It should make uncertainty visible.
Useful forward-looking questions include:
- Which commitments are exposed in the next 24 hours, week, or month?
- What assumptions must remain true for recovery to work?
- What is the confidence in the supplier, quality, inventory, and capacity evidence?
- What will become irreversible if no action is taken?
- Which options disappear after a cut-off or approval delay?
This step is where the operating review connects to NRI, SCEI, DFAI, and CAI. Resilience, exposure, demand confidence, and change agility shape the future consequence of the current signal.
What action is justified?
An action is justified when it is proportionate to consequence, evidence, uncertainty, and reversibility.
Possible actions may include:
- Reallocate inventory.
- Expedite material or transport.
- Resequence production.
- Qualify or activate an alternate source.
- Release or reclassify inventory.
- Change a customer promise.
- Protect a strategic order.
- Pause a lower-value commitment.
- Fix a data or decision-rights problem.
The team should state why the action is justified, what it costs, which risk it creates, who owns it, and when it will be reviewed. “Expedite” is not a decision narrative. “Expedite because three strategic schedule lines are inside the carrier cut-off, the customer impact is high, and the premium cost is below the approved recovery threshold” is closer.
Not every red score requires an emergency action. Some movement reflects noise, a planned change, or a definition issue. The decision sequence protects leaders from overreacting to low-consequence variation while ensuring that high-consequence signals receive timely attention.
Different roles, one decision narrative
The same index should support different questions at different levels.
The executive team needs consequence, exposure, options, cost, and authority. The supply-chain leader needs dependency, timing, alternatives, and cross-functional ownership. The planner needs the affected orders, quantities, dates, constraints, and feasible actions. The plant leader needs sequence, capacity, quality, and labor evidence. Customer service needs the promise, customer impact, communication path, and confidence in recovery. Finance needs the cost, margin, cash, and approval boundary.
These are not competing views. They are different altitudes of one decision narrative. The underlying facts should remain consistent while the level of detail changes.
This prevents two common failures. Executives do not drown in operational detail, and operators do not receive a vague instruction to “protect service” without knowing which commitments or trade-offs matter. A decision-ready layer preserves the connection between intent, constraint, action, and outcome.
When not to escalate
Good operating reviews also define what does not require escalation.
A small low-criticality variance with a clear owner and a normal recovery path may be handled locally. A score movement caused by a planned shutdown may need documentation but not executive intervention. A data-quality issue may need correction before the business takes an operational action.
Escalation should be based on consequence, urgency, uncertainty, reversibility, and authority, not on the color alone. If every red score becomes an executive emergency, leaders stop distinguishing signals. If no red score crosses a clear threshold, the organization loses time.
Thresholds should be versioned and tested. They should identify who is notified, what evidence is required, what decision window applies, and when the issue returns to normal operating ownership.
Evidence quality and investigation speed
Investigation slows when each function uses a different definition of the same object. Sales may refer to an order, planning to a schedule line, production to a campaign, quality to a batch, and logistics to a shipment. The team may be discussing one customer outcome through five identifiers.
A decision-ready review needs shared relationships between those objects. It should be possible to move from the customer promise to the schedule line, material, batch, production step, carrier, and invoice without rebuilding the case manually. This is not only a technology benefit. It is a design choice about what the organization considers a connected decision episode.
The review should also measure how long investigation takes. If the index moves every Monday but the cause is not understood until Friday, the business has lost most of the response window. Decision latency is evidence about the operating model and should be improved alongside the index itself.
Drilling down by business object
An index should allow leaders to move through the network without losing the meaning of the original score.
Product
Which product, grade, size, or configuration is driving movement? Is the problem isolated to a specification, campaign, or material route?
Plant
Which site, line, shift, or process is affected? Is the issue capacity, quality, energy, yield, maintenance, or sequence?
Customer
Which customer, segment, promise, or contract is exposed? What is the commercial and relationship consequence?
Supplier
Which supplier, region, material, or sub-tier dependency is involved? Is the source reliable, exposed, or unqualified for alternatives?
Schedule line
Which exact quantity and date is at risk? Was the original promise changed? Is the line usable only when delivered with other lines?
Decision episode
What did the team know, what options existed, who chose, what action occurred, and what outcome followed?
These drill-downs should be designed before the dashboard is built. If the system shows a red number but cannot connect it to the object and decision that created it, the index will remain descriptive.
The falling ONRI example
Start with the network view. ONRI has fallen from 94 to 86. The raw OTIF result has fallen less, suggesting that the misses are concentrated in higher-criticality commitments.
Drill into customer. One strategic account represents a large share of the weighted decline. Several schedule lines were originally due within the same week.
Drill into product. The lines share one premium grade that requires a constrained finishing route.
Drill into plant. The line lost two production windows because a quality hold delayed the previous batch. Rework was completed, but the schedule did not recover before the customer cut-off.
Drill into supplier. The input material came from the primary source. A nominal alternate exists but is not technically qualified for the premium grade.
Drill into decision. The team used available inventory for a lower-criticality order because the strategic priority was not visible in the allocation rule. Premium freight was considered but rejected before the severity of the customer impact was understood.
Now the cause is clearer. The ONRI movement is not simply a forecast problem or a transport problem. It reflects quality release, constrained capacity, qualification exposure, allocation governance, and delayed decision-making.
The justified action may include protecting remaining inventory, expediting a feasible quantity, initiating a customer conversation, reserving the next production slot, and starting alternate qualification. The review should also assign improvement work across QRI, NRI, SCEI, and CAI.
Designing the weekly operating review
A weekly review should be short enough to sustain and deep enough to change action.
Before the meeting
Publish index movement, baseline, data confidence, and the top exceptions. Include the affected customer, product, plant, supplier, and decision owner. Avoid asking the meeting to discover basic facts live.
In the meeting
Use the six-question sequence. Spend the most time on signals with high consequence, short intervention windows, or conflicting evidence. Record decisions, owners, thresholds, and review dates.
After the meeting
Track whether the action occurred, whether the outcome matched the expectation, and what assumption changed. Close the loop with the index and the decision record.
The review should not become a tour of every metric. It should be a forum for choosing what deserves attention and what action is justified.
Avoiding metric theatre
Metric theatre occurs when the organization spends more time explaining scores than improving outcomes.
Warning signs include:
- Repeated debate over definitions with no version control.
- Long lists of exceptions with no prioritization.
- Scores presented without data confidence or consequence.
- Actions with no owner or deadline.
- Root causes copied from old reports.
- Improvement projects disconnected from index movement.
- Targets achieved through reclassification or goalpost changes.
The antidote is a decision record. Capture the signal, evidence, interpretation, options, choice, owner, outcome, and learning. The record can be simple. Its value comes from linking the number to the action and result.
Governance of index use
Indexes need governance not only in calculation but in use.
Leaders should agree who owns the definition, who can change a weight, who approves exceptions, and who decides when an index movement requires escalation. The operating review should preserve original data, current status, and historical definitions.
The organization should distinguish fact, model interpretation, recommendation, and human decision. This is especially important when analytics or AI generate explanations. A recommendation should show evidence, confidence, major uncertainty, and the action that would improve confidence.
Governance should also protect psychological safety. If teams fear that a red score will be used for blame, they will hide uncertainty and delay escalation. A decision-ready review holds people accountable for choices while making system constraints visible.
Questions for leaders
Leaders can ask:
- What exactly changed, and compared with which baseline?
- Where is the movement concentrated?
- How confident are we in the evidence?
- What is the likely next consequence?
- Which options remain, and which will disappear soon?
- What action is proportionate to the consequence and cost?
- Who owns the action and when will it be reviewed?
- What index or business object should we drill into next?
- What did we learn that should change a rule, process, or investment?
- Did the outcome match the decision’s expectation?
The final question makes the review a learning system rather than a recurring status meeting.
Start with one index and one decision
Choose one index that leaders already discuss and one recurring decision it should improve. ONRI and customer-promise recovery are a practical starting point.
Define the movement, drill-downs, forward-looking questions, action thresholds, and outcome measures. Run the sequence manually for several cycles. Record what evidence was missing and which decisions were delayed.
Then improve the data product. Add the views that actually changed the conversation: schedule lines, quality status, inventory usability, supplier exposure, capacity, customer criticality, or margin.
The goal is not to automate the meeting immediately. It is to understand the reasoning pattern well enough that technology can accelerate it without hiding the judgment.
A compact decision brief
A weekly decision brief can fit on one page. It should contain:
- Signal: what changed and when.
- Scope: affected customers, products, plants, suppliers, and lines.
- Consequence: what is at risk and by when.
- Evidence: facts, confidence, and missing information.
- Cause: trigger, constraint, decision, and execution chain.
- Options: feasible actions, cost, downside, and reversibility.
- Recommendation: the preferred response and why.
- Authority: who approves, executes, and escalates.
- Outcome: what happened after the action.
The brief is not a replacement for operational systems. It is a shared object for a decision. It reduces meeting time because people can review the situation before the conversation and preserves learning afterward.
From score to decision
An index is valuable because it helps an organization notice what matters before the outcome is fixed. It creates a shared signal across functions and gives leaders a way to focus attention.
But the score is only the beginning. Leaders must ask what happened, why it happened, what is likely next, what options remain, and what action is justified. They must drill from network to customer, product, plant, supplier, schedule line, and decision episode without losing the business context.
The value of an index is not the elegance of its formula or the color on its dashboard. It is measured by the quality and speed of the conversation it enables, and by whether that conversation improves the next decision.
That is the difference between a performance report and a decision system: one describes movement, while the other helps the organization respond while options still exist.
It turns measurement into a shared operating habit rather than a periodic exercise in explanation.
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. A decision-lake implementation must be validated against local safety, quality, cybersecurity, regulatory, contractual, labour, privacy, and data-governance requirements. AI recommendations should remain within clearly defined human authority and operational controls.
#SupplyChain #DecisionIntelligence #SupplyChainAnalytics #OperationsManagement #ExecutiveLeadership #SAndOP #OperationalExcellence #BusinessTransformation #SupplyChainStrategy #DecisionMaking
Takeaways
Excerpt | Practical point / context |
|---|---|
“Everyone sees the same red number. Everyone proposes a different cause.” | A metric without a shared investigation method creates debate instead of action. |
“A score is a signal, not a diagnosis.” | Index movement should trigger disciplined inquiry. |
“The score starts the conversation. The decision sequence turns it into operating value.” | Metrics become useful when connected to action and learning. |
“The first task is to describe the movement without explanation.” | Establish the fact pattern before assigning cause. |
“Diagnosis looks backward. Decision-making must look forward.” | Leaders must estimate future consequence and remaining options. |
“An action is justified when it is proportionate to consequence, evidence, uncertainty, and reversibility.” | Action thresholds should be explicit and balanced. |
“The first objective is not to automate the meeting immediately.” | Understand the reasoning pattern before building technology. |
“Metric theatre occurs when the organization spends more time explaining scores than improving outcomes.” | Avoid reporting rituals disconnected from action. |
“The final question makes the review a learning system rather than a recurring status meeting.” | Compare decisions with outcomes. |
“The value of an index is measured by the quality and speed of the conversation it enables.” | The article’s central takeaway. |
