A record-breaking quarter and a collapsing service level at a fulfilment operator
Harbourline Fulfilment runs two distribution centres and picks, packs and ships orders for 40 e-commerce clients. Its QMS dashboard carries three headline indicators: sales revenue, orders picked per labour hour (measured per shift), and picking accuracy (measured per picker, reported monthly). A new warehouse layout went live in March. Since then picking accuracy has fallen from 99.6% to 98.1% and on-time dispatch from 97% to 89%, while orders per labour hour has risen 12% - which the operations director is presenting as the site's best quarter on record. Client complaints about wrong items have tripled and two clients have withheld volume. Each function reviews its own figures in its own meeting; nothing brings sales, inbound receiving, picking and dispatch together. The layout change was approved by operations alone, with no documented consideration of its effect on receiving or on the customer. Management review minutes list each indicator separately, with no commentary on how they interact.
1. Using the process approach, explain how an indicator can improve while the outcome the customer receives gets worse. What does the standard require Harbourline to determine about its processes?
Read the worked answer
The process approach is about the system of interacting processes, not the sum of departmental scores. A local optimum is not a system optimum. Picking speed rose because the new layout shortened travel paths; the same layout appears to have moved similar-looking SKUs closer together and reduced the space available for inbound put-away checks, so pickers now reach a location faster and pick the wrong item more often. Speed and accuracy were never independent, but the two indicators sit in different functions' reports, so the trade-off is invisible.
What clause 4.4.1 requires the organization to determine for each process:
- inputs required and outputs expected;
- sequence and interaction of the processes;
- criteria and methods needed, including monitoring, measurement and related performance indicators;
- resources, responsibilities and authorities;
- risks and opportunities;
- evaluation and improvement methods.
The word that matters here is interaction. Harbourline defined indicators per function, so 4.4.1 c was satisfied in form and defeated in substance: no indicator describes the handover between receiving and picking, and no indicator is anchored to the customer's outcome. The 12% efficiency gain is therefore an unverified claim about the whole system - it may be a real gain, or it may be a cost transferred to rework, returns and churn. Only a measure that spans the process can tell the difference, and until then the operations director's conclusion is unsupported.
2. What would you change in how performance is monitored, analysed and reviewed? Be specific about indicators, review structure and the clauses that support each change.
Read the worked answer
Monitoring (9.1.1). 9.1.1 requires the organization to determine what needs monitoring and measuring, the methods, when, and when results are analysed and evaluated. Harbourline should add end-to-end indicators that no single function owns: order accuracy measured at dispatch and confirmed by the customer, dock-to-dispatch cycle time, rework and return rates attributable to wrong picks, and first-contact resolution. Accuracy should also be reported by location and SKU cluster, not only per picker - the per-picker cut disguises a layout effect and points blame at individuals.
Customer perception (9.1.2). Complaint tripling and withheld volume are the strongest signals available; they must enter the same review as the internal metrics, not a separate customer-service meeting.
Review structure (9.3). Inputs under 9.3.2 include process performance and product conformity, customer satisfaction and feedback from relevant interested parties, and nonconformities and corrective actions. A review that examines each indicator in isolation does not analyse that input - it reads it aloud. Require one page per quarter that states the trade-off explicitly: what changed in the system, which indicator moved in which direction, and which customer outcome moved with it.
Accountability (5.1.1 c, 5.3). Someone must own the dock-to-dispatch process as a process, with authority over the layout decision. Without that, the same local optimisation recurs at the next change.
3. Is the layout change evidence that risk-based thinking was inadequate? Justify your answer against the relevant requirements rather than against good practice alone.
Read the worked answer
Yes, but the finding must be stated against a requirement, not against a feeling that "more analysis would have been nice".
The applicable requirements are 6.1 and 6.3:
- 6.1.1–6.1.2 require the organization to determine risks and opportunities considering 4.1 and 4.2, and to plan actions to address them, integrate those actions into QMS processes, and evaluate their effectiveness. The relevant risk is not "the warehouse might be inefficient" in the abstract; it is the potential effect on the ability to meet customer and applicable requirements - wrong-item dispatch and late dispatch.
- 6.3 (planning of changes) requires changes to be carried out in a planned manner, considering the purpose of the change and its potential consequences, the integrity of the QMS, availability of resources, and allocation or reallocation of responsibilities and authorities. A layout change is a change to the QMS.
A change approved by one function, with no recorded consideration of consequences for receiving, picking accuracy, or the customer, is a 6.3 failure with a 6.1 root. The strongest NC statement is not "no risk assessment was performed" - the standard permits many forms - but: "Change to warehouse layout implemented March 2025 without consideration of potential consequences for receiving, picking accuracy or dispatch, and without reallocation of responsibilities for the affected processes."
Contrast with the invalid version of this finding: a demand for a formal risk matrix. 6.1 does not impose a methodology, so the observable defect is the absence of consequence analysis and integration, not the absence of a template.