How long can the customer wait?

Five parts and one question for the supplier

Preface and contents · supplied PDF, pp. 6–9

Yossi Sheffi looks at the enterprise alongside its suppliers, carriers, factories and customers. Part I follows a disruption as it unfolds. Part II explains how supply chains handle everyday changes in demand. Part III covers security, detection and redundancy. Part IV examines ways to change how work gets done. Part V links that ability to competing in ordinary business.

We read the book from the position of a team that has already promised a customer an outcome. If supply fails, it has to determine how long the customer can wait, choose a recovery route and agree who pays. Here are eight ideas from Sheffi and our applications. Historical cases are attributed to the book. Questions and scenarios involving the companies in this issue are our own.

1. Start with how long the customer can hold out

Chapters 1–4 · PDF pp. 17–24, 32–50, 81–94

Sheffi separates the likelihood of a disruption from its severity, then follows its consequences over time. After the Kobe earthquake, a Sumitomo plant lost gas and water despite escaping physical damage, according to the book. The shortage of brake components reached Toyota’s assembly lines. When US West Coast ports reopened after the 2002 lockout, the backlog of cargo also outlasted the closure.

Put it to work. Pick one critical item. Trace it back from the customer to the operation that makes it usable. That might be testing, packing or calibration. Record the dependencies on the site, utilities, materials and transport. Compare the days of stock available to the customer with the date when the first usable replacement shipment can arrive. Record how long it will then take to clear outstanding orders. Each date informs a different decision.

In this issue. Restoring launch service and meeting the commitment to a satellite customer run on different calendars in the Adaspace case. In a comparable project, the customer needs a date for a usable replacement with testing and acceptance. A reopened launch window alone cannot supply that date.

2. Give a warning an owner and a decision deadline

Chapters 1, 3, 7–9 · PDF pp. 17–24, 58–61, 139–148, 160–166, 180–194

In Sheffi’s account of the Philips fire, Nokia moved affected components to daily monitoring, sought substitutes and pressed for capacity. At Ericsson, the signal took longer to reach people who could move resources. Sheffi also describes Scottish whisky makers combining records of small thefts and finding a common problem route. Linking observations made each one more informative.

The security chapters add two conditions. Test protective measures against different ways they can fail. A shared data source or decision maker can weaken several safeguards together. Near misses become useful when someone follows the report through to a check that the cause has been addressed.

Put it to work. Set an escalation threshold for one risk. After a missed test date, for example, an engineer assesses the impact on the next batch and the programme lead chooses that working day whether to recheck, reschedule or switch. Record the observation, the cause still to be established, the latest decision time and the person responsible. Review each month which alerts led to action and where the team lost time.

3. Agree on the signal that triggers replenishment

Chapters 5–6 · PDF pp. 99–114, 116–134

A distributor’s order can reflect a promotion, batch size or concern about a shortage. Sheffi describes Barilla using distributor shipments to stores as a signal closer to consumption. In another chapter, HP handled base demand and uncertain additional demand with facilities that differed in cost and speed of response.

Put it to work. Before a trial shipment, agree what will trigger replenishment. Possibilities include actual units used by the customer, the customer’s available inventory or a confirmed order from a store. Set an observation window, a decision date and responsibility for volume beyond the agreed base. If the customer shares only purchase orders, ask which promotions and range changes could alter what those orders tell you.

In this issue. In the Karat case, distinguish a supply agreement, the first shipment and actual replenishment. A chronology of those events and data on available inventory would help assess repeat business. The first two milestones leave recurring demand an open question.

4. Size the buffer to cover the time it takes to switch

Chapters 10 and 16 · PDF pp. 197–204, 306–309

For Sheffi, a buffer buys time and requires both funding and release rules. In the Johnson & Johnson example, a protected reserve of medical goods for the Pentagon rotated through ordinary stock. Use below the protected level required the customer’s permission. An agreement on access made the inventory usable.

Our hypothetical calculation. Six hundred usable units at 60 per day cover 10 days. If the decision, qualification and delivery take two, three and seven days in sequence, the alternative arrives after 12 days. With demand unchanged, 120 units cover the gap. Two days saved on the total switching time could also close it. Compare carrying cost, write-offs, expediting and available cash before choosing.

Put it to work. Record who owns and funds the buffer, what releases it, when it expires and how it is replenished. Confirm that the relevant customer can draw on that batch. Stock earmarked for another customer or awaiting quality release leaves the gap open.

5. Test what can move and what can be decided later

Chapters 11–12 · PDF pp. 208–218, 220–233

In Sheffi’s Intel example, a compatible production process allowed work to move between factories. That required compatible equipment, procedures and inputs. Benetton delayed dyeing some garments until sales information became available. In that case, flexibility came from deciding on the product variant later.

Put it to work. Separate these mechanisms. To test a production transfer, run an operation at the alternative site. Measure the time for moving data, setup, testing and release. To test late configuration, see whether a common blank can be stocked and packed, labelled or configured once an order arrives. Compare the reductions in inventory and shortages with the added work and the time required for the final operation.

In this issue. For Firefly and AST, this raises a question about repeatability. What can the receiving team perform with the documents and equipment it has? Which check would demonstrate repeatable results? For a supplier to chains, the comparable test could concern the final operation on a common blank. Its feasibility depends on the buyer’s specification.

6. Check where capacity would come from in a shared disruption

Chapter 13 · PDF pp. 235–251

After the Aisin fire, the secondary source could not meet Toyota’s valve requirements. Sheffi describes a network of manufacturers joining the recovery, with Aisin supplying drawings, engineers and quality control. In a contrasting case, two Motorola chip suppliers were hit by the same earthquake in Taiwan. A supplier count tells you little about whether capacity is available.

Put it to work. For each supply route, identify the factory, critical sub-supplier, production line, transport node, tooling access and required approvals. Ask what volume would be available if several customers called at once. Check who controls the drawings and tools if a supplier stops operating and who pays for transfer. Verify one item with a document or trial shipment.

In this issue. For the proposed Solvay and Viridis chain, usable capacity depends on material, processing and acceptance. A replacement scenario would need a qualified independent facility and a lead time for the first acceptable batch. The announcements leave those questions open.

7. Agree with customers how a shortage will be handled

Chapters 14–15 · PDF pp. 260–271, 273–296

Sheffi describes Continental Teves collecting customer inventory figures after air transport stopped and moving critical shipments by road. The time to a customer shutdown guided allocation. In the chapter on culture, he connects the ability to improvise with shared information, authority to act, accountability and practice.

Put it to work. Agree how to handle partial deliveries. What priority should a contractual minimum, a potential customer shutdown or usable stock receive? Who can change the route, authorise additional spending and agree a substitute? The customer’s first update should give the confirmed scale, volume available and the time of the next update. Explain allocation through a rule the team can apply across customers.

A working test. Give the account and operations leads one delay scenario. Ask them to make a decision without the manager who usually approves exceptions. Record the information and authority they had to find. The exercise exposes gaps while there is still time to fix them.

8. Price what the investment does in an ordinary week

Chapter 16 · PDF pp. 300–316, especially 310 and 314

Sheffi sums up the argument in one sentence. “The best way to achieve supply chain resilience is to create flexibility.” He suggests making the case for security through its accompanying benefits and for flexibility through its role in everyday operations. A fast local source can complement a distant one during a launch, a demand spike or a disruption.

Put it to work. Compare the annual cost with everyday changes in working capital, qualification, downtime, expediting and orders filled. Then work through two disruption scenarios and estimate how many days of the customer’s gap each option closes. Compare at the same required service level. If only one scenario justifies the expense, identify the probability or consequence that would make the decision worthwhile. Keep model inputs distinct from forecasts.

The limit of the argument. The book draws on historical cases and an MIT research project from 2002 to 2005. These explain mechanisms. They cannot supply an average return on flexibility investments. A budget decision needs current evidence from your own chain, including failed attempts to switch as well as successful recoveries.

One test for the next working meeting

Take one item or project commitment. Record how long the customer can wait and when an alternative batch could be accepted. Identify the first action that could close the gap. You might need to approve a sample, reserve a line, obtain a drawing or agree who pays. Give the action an owner, a spending limit and a document that will demonstrate readiness. Run one test in the next week.