Identify and describe the global market forces, risks, development chain, supply chain, and strategies from a perspective

Hewlett-Packard: DeskJet Printer Supply Chain

Brent Cartier, manager for special projects in the Materials Department of Hewlett-Packard (HP) Company’s Vancouver Division, clicked off another mile. It had been a long week and it looked like it would be a long weekend as well, based on the preparation that needed to be done for Monday’s meet- ing with group management on worldwide inventory levels for the DeskJet Printer product line. Even when he was busy, he always took time for the 25-mile bike ride to work—it helped reduce stress in times like this.

The DeskJet printer was introduced in 1988 and had become one of HP’s most successful products. Sales had grown steadily, reaching a level of over 600,000 units in 1990 ($400 million). Unfortunately, inventory growth had tracked sales growth closely. Already, HP’s distribution centers had been filled with pallets of the DeskJet printer. Worse yet, the organization in Europe was claiming that inventory levels there needed to be raised even further to main- tain satisfactory product availability.

Each quarter, representatives from the produc- tion, materials, and distribution organizations in Europe, Asia-Pacific, and North America met to discuss “the I-word”—as they referred to it—but their conflicting goals prevented them from reach- ing consensus on the issues. Each organization had a different approach to the problem. Production had

not wanted to get involved, claiming it was “just a materials issue,” but had taken the time to rant about the continued proliferation of models and options. The distribution organization’s pet peeve was forecast accuracy. They didn’t feel that the dis- tribution organization should have to track and store warehouses of inventory, just because the Vancouver Division couldn’t build the right prod- ucts in the right quantities. The European distribu- tion organization had even gone so far as to suggest that they charge the cost of the extra warehouse space that they were renting back to the Vancouver Division directly, instead of allocating it among all the products that they shipped. Finally, Brent’s boss, David Arkadia, the materials manager at the Vancouver Division, had summarized the perspec- tive of group management at the last meeting when he said, “The word is coming down from corporate: We can’t run our business with this level of unpro- ductive assets.We’re just going to have to meet cus- tomer needs with less inventory.”

As Brent saw it, there were two main issues. The first issue was to find the best way to satisfy customer needs in terms of product availability while minimiz- ing inventory. The second and stickier issue involved

CHAPTER 11

Coordinated Product and Supply Chain Design

331

C A S E

Source: Copyright © 1994 by the Board of Trustees of the Leland Stanford Junior University. All rights reserved. Used with permission from the Stanford University Graduate School of Business. This case was written by Laura Kopczak and Professor Hau Lee of the Department of Industrial Engineering and Engineering Management at Stanford University.

sim8239X_ch11_331-364.qxd 5/9/07 12:01 PM Page 331

THE RETAIL PRINTER MARKET

Worldwide sales of small workgroup/personal printers in 1990 were about 17 million units, amounting to $10 billion. The market tracked personal computer sales closely; the market was mature in the United States and western Europe but was still developing in eastern Europe and in the Asia-Pacific region. Small workgroup/personal printers were sold almost exclu- sively through resellers. The reseller channels were changing rapidly, particularly in the United States. Traditionally, printers had been sold through computer dealers, but as personal computers became commodity products, more and more sales were flowing through superstores and consumer mass merchandisers such as Kmart and Price Club.

The retail printer market was composed of three technology segments: impact/dot matrix (40 percent), ink-jet (20 percent), and laser (40 percent). Dot matrix was the oldest technology and was viewed as noisy and of lower print quality compared to the other two types. The dot-matrix printer market share was expected to fall to 10 percent during the next few years as the technology was replaced by either ink-jet or laser printers in all applications except multipart forms and wide-carriage printing. Prior to 1989 most customers were not aware of ink-jet technology. However, customers were dis- covering that ink-jet print quality was almost as good as laser print quality—and at a much more affordable price. Sales had increased dramatically. In the monochrome market, it remained to be seen which technology would eventually dominate at the low end. Much would depend on the pace at which technology developed in both areas and on the rela- tive costs.

HP and Canon pioneered ink-jet technology sepa- rately at their respective corporate laboratories during the early 1980s. The key technological breakthroughs had been ink formulation and the disposable print- head. HP had introduced its first disposable head model, the ThinkJet printer, in the late 1980s, while Canon had just introduced one in 1990.

HP led the ink-jet market in the United States, while Canon led the market in Japan. European com- petitors included Epson, Mannesmann-Tally, Siemens, and Olivetti, though only Olivetti had introduced a printer with a disposable print head by 1991. Some dot-matrix printer companies were also starting to offer ink-jet printer products.

332 DESIGNING AND MANAGING THE SUPPLY CHAIN

how to get agreement among the various parties that they had the right level of inventory. They needed to develop a consistent method for setting and imple- menting inventory goals and get everyone to sign off on it and use it. It was not going to be easy. The situa- tion was especially urgent in Europe. His mind was still filled with the faxed picture that he had received the previous day, showing the dip in product avail- ability levels for some versions of the product at the European Distribution Center (DC), yet he was sure that loads and loads of DeskJets had been shipped to Europe in the past months. His voice mail had been filled with angry messages from the sales offices, and yet the European DC was telling Vancouver that it had run out of space to store Vancouver’s products.

Brent parked his bike and headed for the company showers. His morning shower was another ritual— this was the time he had to review his plans for the day and play out different scenarios. Perhaps a solu- tion would come to him . . .

BACKGROUND

Hewlett-Packard Company was founded in 1939 by William Hewlett and David Packard, with head- quarters in Palo Alto, California. It grew steadily over the next 50 years, diversifying from its base in electronic test and measurement equipment into computers and peripherals products, which now domi- nated its sales. In 1990 HP had over 50 operations worldwide, with revenues of $13.2 billion and net income of $739 million.

HP was organized partially by product group and partially by function. The Peripherals Group was the second largest of HP’s six product groups, with 1990 revenues of $4.1 billion. Each of the group’s divi- sions acted as a strategic business unit for a specific set of products. Products included printers, plotters, magnetic disk and tape drives, terminals, and net- work products.

The Peripherals Group had set technological stan- dards with many of its products, with innovations such as the disposable print head used in its ink-jet printers and moving-paper plotters. While these inno- vations contributed to its success, the Peripherals Group was also recognized for its ability to identify and profitably exploit market opportunities, as in the case of its most successful product, the LaserJet printer.

sim8239X_ch11_331-364.qxd 5/9/07 12:01 PM Page 332

recognized an opportunity to have his ideas tested in the United States. They decided to work together.

Within a year Vancouver had converted the factory to stockless production just-in-time (JIT) and had reduced inventory from 3.5 months to 0.9 month, with a drastic reduction in cycle time. Vancouver became a showcase factory for the kanban process; between 1982 and 1985 more than 2,000 executives from within and outside HP toured the process. Vancouver impressed visitors by having them sign a raw printed circuit board as they arrived, then pre- senting them with a finished printer, made with that PC board using the standard process, an hour and a half later.

There was one key element missing, however. As Bob Foucoult puts it, “We were all dressed up but had no one to take us to the dance.” Vancouver had not yet introduced a successful, high-volume prod- uct that would take full advantage of the advanced production line. Vancouver had introduced products based on HP’s latest ink-jet technology but, as with any new technology, they had to gain experience to work the bugs out. The early models had poor reso- lution and required special paper for printing, resulting in limited success in the marketplace. In 1988 things started to change. Vancouver intro- duced the DeskJet printer, a new model with near- letter-quality resolution that used standard paper. The introduction was a wild success. Since the manufacturing process had been in place and had been thoroughly exercised, all that was needed was to “flip the switch.” HP’s knowledge and imple- mentation of the ink-jet technology, combined with its streamlined manufacturing process, gave it the edge needed to become the market leader in the ink-jet printer market.

THE DESKJET SUPPLY CHAIN

The network of suppliers, manufacturing sites, distri- bution centers (DCs), dealers, and customers for the DeskJet product comprised the DeskJet supply chain (Figure 11-1). Manufacturing was done by HP in Vancouver. There were two key stages in the manufac- turing process: (1) printed circuit board assembly and test (PCAT) and (2) final assembly and test (FAT). PCAT involved the assembly and testing of electronic components such as ASICs (application-specific inte- grated circuits), ROM (read-only memory), and raw printed circuit boards to make logic boards and

CHAPTER 11: COORDINATED PRODUCT AND SUPPLY CHAIN DESIGN 333

Ink-jet printers were rapidly becoming commod- ity products. The end customer, choosing between two ink-jet printers of equal speed and print quality, increasingly used general business criteria such as cost, reliability, quality, and availability to decide. Product loyalty continued to decrease.

THE VANCOUVER DIVISION AND ITS QUEST FOR ZERO INVENTORY

In 1990 the Vancouver Division’s mission statement read: “Our Mission Is to Become the Recognized World Leader in Low-Cost Premium-Quality Printers for Printed Communications by Business Personal Computer Users in Offices and Homes.”

The Vancouver Division, located in Vancouver, Washington, was established in 1979. HP saw an opportunity to provide personal printers for the rela- tively new, fast-growing personal computer market. HP consolidated personal printer activities from four divisions (Fort Collins, Colorado; Boise, Idaho; Sunnyvale, California; and Corvallis, Oregon) to the Vancouver site. The new division became part of HP’s Peripherals Group and was chartered with the design and manufacturing of ink-jet printers.

As Bob Foucoult, the production manager and one of Vancouver’s first employees, recalled, “Manage- ment was pulled from all over HP and plopped down in Vancouver. There was no cohesive staff and no cohesive set of business practices—perhaps that’s why we were so open to new ideas.”

The manufacturing organization realized early on that a fast, high-volume manufacturing process would be required for success in the printer market. With the current (1979) 8- to 12-week manufacturing cycle time and 3.5 months of inventory, the Vancouver Division would be doomed to fail. They looked within HP for knowledge of high-volume processes, but found none. HP, being an instrument company, only had experience building low-volume, highly customized products using batch processes.

One day in mid-1981 two Vancouver managers happened to take seats on a plane next to two profes- sors: Richard Schoenberger (Nebraska University) and Robert Hall (Indiana University). Schoenberger had just written a rough draft for a paper called “Driving the Productivity Machine” about a manu- facturing process being used in Japan: kanban. Vancouver’s management recognized the promise of this “new” manufacturing concept and Robert Hall

sim8239X_ch11_331-364.qxd 5/9/07 12:01 PM Page 333

printhead driver boards for the printers. FAT involved the assembly of other subassemblies such as motors, cables, keypads, plastic chassis and “skins,” gears, and the printed circuit assemblies from PCAT to produce a working printer, as well as the final testing of the printer. The components needed for PCAT and FAT were sourced from other HP divisions as well as from external suppliers worldwide.

Selling the DeskJet in Europe required customiz- ing the printer to meet the language and power sup- ply requirements of the local countries, a process known as “localization.” Specifically, the localiza- tion of the DeskJet for different countries involved assembling the appropriate power supply module, which reflected the correct voltage requirements (110 or 220) and power cord terminator (plug), and packaging it with the working printer and a manual

334 DESIGNING AND MANAGING THE SUPPLY CHAIN

written in the appropriate language. The design of the product was such that the assembly of the power supply module had to be done as part of the final assembly and test process, and therefore the localiza- tion of the printer was performed at the factory. Hence, the finished products of the factory consisted of printers destined for all of the different countries. These products were then sorted into three groups destined for the three distribution centers: North America, Europe, and Asia-Pacific. Figure 11-2 details the bill of materials and the various options available.

Outgoing products were shipped to the three distri- bution centers by ocean. In Vancouver, inventories of the components and raw materials were maintained to meet production requirements, but otherwise, no sig- nificant buffer inventories between the PCAT and FAT

FIGURE 11-1 The Vancouver supply chain.

Supplier

Supplier

Supplier Supplier

IC Mfg

PCAT FAT EuropeanDC Customer

Customer

US DC Customer

Print Mech Mfg

Far East DC

Key: IC Mfg = Integrated circuit manufacturing PCAT = Printed circuit assembly and test FAT = Final assembly and test Print Mech Mfg = Print mechanism manufacturing

FIGURE 11-2 Bill of material in the Vancouver supply chain.

Raw wafers ASIC

Raw PCB PCB

Raw head driver board

Cables Keypad Motors Plastics

Print mechanism

DeskJet printer

Power supply

Manuals

Finished product

Versions: A AA AB AQ AU AY AK

sim8239X_ch11_331-364.qxd 5/9/07 12:01 PM Page 334

stages were kept. Management had continued to pre- fer to maintain no finished goods inventory at the fac- tory, a tradition that was started in 1985 as described in the previous section.

The total factory cycle time through the PCAT and FAT stages was about a week. The transportation time from Vancouver to the U.S. DC, located in San Jose, California, was about a day, whereas it took four to five weeks to ship the printers to Europe and Asia. The long shipment time to the DCs in Europe and Asia was due to ocean transit and the time to clear customs and duties at ports of entry.

The printer industry was highly competitive. Customers of HP’s computer products (resellers) wanted to carry as little inventory as possible, yet maintaining a high level of availability to end users (consumers) was critical to them. Consequently there had been increasing pressure for HP as a manufac- turer to provide high levels of availability at the DCs for the resellers. In response, management had decided to operate the DCs in a make-to-stock mode in order to provide very high levels of availability to the dealers. Target inventory levels, equal to the fore- casted sales plus some safety stock level, were set at the three DCs.

As mentioned earlier, Vancouver prided itself as an almost “stockless” factory. Hence, in contrast to distribution, manufacturing of the DeskJet printer operated in a pull mode. Production plans were set weekly to replenish the DCs “just in time” to main- tain the target inventory levels. To ensure material availability, safety stocks were also set up for incom- ing materials at the factory.

There were three major sources of uncertainty that could affect the supply chain: (1) delivery of incoming materials (late shipments, wrong parts, etc.), (2) internal process (process yields and machine downtimes), and (3) demand. The first two sources of uncertainties resulted in delays in the manufac- turing lead time to replenish the stocks at the DCs. Demand uncertainties could lead to inventory buildup or back orders at the DCs. Since finished printers were shipped from Vancouver by ocean, the consequence of the long lead time for the European and Asian DCs was that the DC’s ability to respond to fluctuations in the demand for the different versions of the product was limited. In order to assure high availability to customers, the European and Asian DCs had to maintain high lev- els of safety stocks. For the North American DC

CHAPTER 11: COORDINATED PRODUCT AND SUPPLY CHAIN DESIGN 335

the situation was simpler; since an overwhelming majority of demand was for the U.S. version of the DeskJet printer, there was little localization-mix fluctuation.

THE DISTRIBUTION PROCESS

At HP, while a typical DC shipped hundreds of differ- ent peripheral and computer products, a small number of products accounted for a large share of the unit volume. The DeskJet printer was one of these high-volume products.

The Operations Manager of each regional DC reported to a Worldwide Distribution Manager, who reported directly to HP’s Vice President of Marketing, and by dotted line to the Peripherals Group Manager (peripherals made up the bulk of shipments through distribution centers). Each Operations Manager had a staff of six functional managers, representing Finance, MIS, Quality, Marketing, Physical Distribution, and Distribution Services. The first three functions were similar to their respective functions in a manufactur- ing organization. Marketing was responsible for inter- actions with customers. Physical Distribution was responsible for the “physical process,” that is, from receiving through shipping. Distribution Services was responsible for planning and procurement.

The major performance measures for a typical DC included line item fill rate (LIFR) and order fill rate (OFR). LIFR was calculated as the total number of customer order line items filled on time divided by the total number of customer line items attempted. (Each time HP tried to pull material for a line item, it was counted as an attempt.) OFR was a similar measure, but was based on orders com- pleted, where an order contains multiple line items. Secondary performance measures included inven- tory levels and distribution cost per gross shipment dollar. The two major costs were outbound freight and salaries. Freight was charged back to the prod- uct lines based on the actual number of pounds of product shipped. In addition, the DC estimated the “percentage of effort” required to support a particu- lar product line and charged that percentage of non- freight costs back to that product line. The system was somewhat informal, and major negotiations took place between the DCs and the major product lines during the budget-setting process to determine the percentage allocation that was appropriate for each product line.

sim8239X_ch11_331-364.qxd 5/9/07 12:01 PM Page 335

The DCs had traditionally envisioned their process as a simple, straight-line, standardized process. There were four process steps:

1. Receive (complete) products from various suppli- ers and stock them.

2. Pick the various products needed to fill a customer order.

3. Shrink-wrap the complete order and label it. 4. Ship the order by the appropriate carrier.

The DeskJet printer fit well into the standard pro- cess. In contrast, other products, such as personal computers and monitors, required special processing, called “integration,” which included addition of the appropriate keyboard and manual for the destination country. Although this extra processing didn’t require much additional labor, it was difficult to accommodate in the standard process and disrupted the material flow. Furthermore, the DCs’ materials management systems supported distribution (passthrough process- ing of “end-items” in the form of individual models and options) and did not support manufacturing (assembly of components into a final product). There were no MRP (material resource planning) or BOM (bill of materials) explosion systems, and the DCs did not have adequate people trained in component pro- curement.

There was considerable frustration within the dis- tribution organization regarding the support of assem- bly processes. In general, top management stressed the DC’s role as a warehouse and the need to continue to “do what they were best at—distribution.” Tom Beal, the U.S. DC materials manager, expressed the general concern when he said, “We have to decide what our core competency is and what value we add. We need to decide whether we are in the business of warehousing or integration, then adopt strategies to support our business. If we want to take on manufac- turing processes (here), we have to put processes in place to support them.”

THE INVENTORY AND SERVICE CRISIS

To limit the amount of inventory throughout the DeskJet supply chain and at the same time provide the high level of service needed had been quite a challenge to Vancouver’s management. The manufacturing group in Vancouver had worked hard on supplier management to reduce the uncertainties caused by delivery variabilities of incoming materials, on improving process yields, and

336 DESIGNING AND MANAGING THE SUPPLY CHAIN

on reducing downtimes at the plant. The progress made had been admirable. However, improvement of forecast accuracy remained a formidable task.

The magnitude of forecast errors was especially alarming in Europe. It was becoming quite common to have product shortages for model demands from some countries, while inventory of some other models kept piling up. In the past, the target inventory levels at the DCs were based on safety stocks that were a result of some judgmental rule of thumb. It seemed like the increasing difficulty of getting accurate fore- casts meant that the safety stock rules would have to be revisited.

David Arkadia had solicited the help of a young inventory expert from corporate HP, Dr. Billy Corrington, to help him put in place a scientifically based safety stock system that would be responsive to forecast errors and replenishment lead times. Billy had formed a team consisting of Laura Rock, an industrial engineer, Jim Bailey, the planning supervisor, and José Fernandez, the purchasing supervisor from Vancouver, to overhaul the safety stock management system. They were to recom- mend a method for calculating appropriate safety stock levels for the various models and options at the three DCs. Gathering appropriate data turned out to be a task that the team spent a lot of time at. They now felt that they had a good sample of demand data (see Table 11-1) and were developing the safety stock methodology. Brent was hoping that this new methodology would solve the inventory and service problem. It would be nice if he could tell his management that all this inventory and service mess was due to their lack of a sound safety stock methodology, and Billy’s expertise would then be their savior.

One issue that continually came up was the choice of inventory carrying cost to be used in safety stock analyses. Estimates within the company ranged from 12 percent (HP’s cost of debt plus some warehousing expenses) to 60 percent (based on the return on investment [ROI] expected of new product develop- ment projects). Another issue was the choice of tar- get line item fill rate to be used. The company target was 98 percent, a number that had been “developed” by marketing.

As faxes and phone calls about the worsening situ- ation at the European DC kept pouring in, Brent also began receiving other suggestions from his colleagues that were more aggressive in nature. Talks about

sim8239X_ch11_331-364.qxd 5/9/07 12:01 PM Page 336

Vancouver’s setting up a sister plant in Europe had surfaced. Would the volume in Europe be large enough to justify such a site? Where should it be located? Brent knew that the European sales and mar- keting folks would like such an idea. He also liked the idea of having a European plant to take care of the inventory and service problem in Europe. Maybe that would put a halt to his recent loss of sleep.

There was certainly a group that advocated more and more inventory. It was simple logic, according to them. “When it comes down to real dollars, inventory costs do not enter into the P&L statements, but lost sales hurt our revenues. Don’t talk to us about inven- tory–service trade-offs. Period.”

Kay Johnson, the Traffic Department supervisor, had long suggested the use of air shipment to trans- port the printers to Europe. “Shortening the lead time means faster reaction time to unexpected changes in product mix. That should mean lower inventory and higher product availability. I tell you, air freight is expensive, but it is worth it.”

Brent recalled his conversation at lunch with a summer intern from Stanford University. The enthusiastic student was lecturing Brent that he should always try to tackle the “root of the prob- lem.” Going to the root of the problem, according to the intern, is what the professors taught at school, and was also what a number of quality gurus preached. “The root of the problem is that you have a horrible forecasting system. There is no

CHAPTER 11: COORDINATED PRODUCT AND SUPPLY CHAIN DESIGN 337

easy way out. You’ve got to invest in getting the system fixed. Now, I know this marketing profes- sor at Stanford who could help you. Have you ever heard of the Box-Jenkins method?” Brent also remembered how he lost his appetite at that lunch, as he was listening to the student who was so eager to volunteer his advice.

WHAT NEXT?

Brent reviewed his schedule for the day. At 11:00 he planned to meet with Billy, Laura, Jim, and José to review the recommended inventory levels they had calculated using the safety stock model. He was somewhat concerned about what level of change the model would recommend. If it suggested small changes, management might not feel the model was useful, but if it suggested large changes, they might not accept it either.

After lunch he would meet briefly with the materi- als and manufacturing managers to review the results and sketch out their recommendations. At 2:00 he would talk with the U.S. DC materials manager by phone. That night he could reach Singapore and Saturday morning he could reach Germany. Hopefully he could get buyoff from everyone. He wondered, too, if there wasn’t some other approach that he should be considering. He knew that whatever numbers he came up with would be too high.

By the end of this chapter, you should be able to answer the following questions: • What are the frameworks, tools and concepts that companies can use as they think

about the product engineering process and its impact on supply chain performance? • How can design for logistics concepts be used to control logistics costs and make

the supply chain more efficient?

TABLE 11-1

SOME SAMPLE DESKJET DEMAND DATA: EUROPE

Option Nov. Dec. Jan. Feb. Mar. Apr. May June July Aug. Sept. Oct.

A 80 0 60 90 21 48 0 9 20 54 84 42 AA 400 255 408 645 210 87 432 816 430 630 456 273 AB 20,572 20,895 19,252 11,052 19,864 20,316 13,336 10,578 6,096 14,496 23,712 9,792 AQ 4,008 2,196 4,761 1,953 1,008 2,358 1,676 540 2,310 2,046 1,797 2,961 AU 4,564 3,207 7,485 4,908 5,295 90 0 5,004 4,385 5,103 4,302 6,153 AY 248 450 378 306 219 204 248 484 164 384 384 234 Total 29,872 27,003 32,344 18,954 26,617 23,103 15,692 17,431 13,405 22,692 30,735 19,455

sim8239X_ch11_331-364.qxd 5/9/07 12:01 PM Page 337

• What is delayed differentiation and how can Hewlett-Packard use delayed differen- tiation to address the problems described in the case above? How can the advan- tages of delayed differentiation be quantified?

• When should suppliers be involved in the new product development process? • What is mass customization? Does supply chain management play a role in the

development of an effective mass customization strategy?

For many years, manufacturing engineering was the last stop in the product engi- neering process. The researchers and design engineers worked on developing a prod- uct that worked, and perhaps one that used materials as inexpensively as possible. Then manufacturing engineers were charged with determining how to make this design efficiently. In the 1980s, this paradigm began to change. Management began to realize that product and process design were key product cost drivers, and that taking the manufacturing process into account early in the design process was the only way to make the manufacturing process efficient. Thus, the concept of design for manufac- turing (DFM) was born.

Recently, a similar transformation has begun in the area of supply chain manage- ment. We have discussed appropriate strategies for supply chain design and opera- tion, assuming that product design decisions were already made by the time the supply chain is designed. Designing the supply chain, we have assumed, involves determining the best way to supply existing products using existing manufacturing processes. In the last few years, however, managers have started to realize that by taking supply chain concerns into account in the product and process design phase, it becomes possible to operate a much more efficient supply chain. Obviously, this is analogous to the DFM practice of taking manufacturing into account during the product design phase.

For most of this chapter, we discuss various approaches that leverage product design in order to manage the supply chain more effectively. Before we get to specific design issues, we begin with a general framework that integrates the development chain introduced in Chapter 1 with the supply chain.

11.1 A GENERAL FRAMEWORK

Recall that in Chapter 1, we introduced the concept of the development chain, the set of activities and processes associated with new product introduction. Indeed, although for much of this text we have focused on the supply chain, in many organizations we find two interacting chains:

• The supply chain, which focuses on the flow of physical products from suppliers through manufacturing and distribution all the way to retail outlets and customers, and

• The development chain, which focuses on new product introduction and involves product architecture, make/buy decisions, earlier supplier involvement, strategic partnering, supplier footprint, and supply contracts.

Clearly, these two chains will intersect as products move from development to pro- duction, and just as clearly, decisions made in the development chain will impact the efficiency of the supply chain. Unfortunately, in most organizations, different man- agers are responsible for the different activities that are part of these chains. Typically, the VP of engineering is responsible for the development chain, the VP of manufactur- ing for the production portion of the chains, and the VP of supply chain or logistics for

338 DESIGNING AND MANAGING THE SUPPLY CHAIN

sim8239X_ch11_331-364.qxd 5/9/07 12:01 PM Page 338

the fulfillment of customer demand. What’s more, each of these managers frequently has performance incentives that focus on his or her individual responsibilities, often ignoring the impact of his or her decisions on the other portion of the development and supply chains. Unless carefully addressed, the typical impact of these organiza- tional and incentive structures is a misalignment of product design and supply chain strategies.

Keep in mind that each chain has different characteristics. For example, key char- acteristics of the supply chain include

• Demand uncertainty and variability, in particular, the bullwhip effect discussed in Chapter 5.

• Economies of scale in production and transportation (see Chapters 2 and 3). • Lead time, in particular due to globalization (see Chapters 9 and 10).

Of course, each of these dimensions has a significant impact on the appropriate supply chain strategy, and, thus, in Chapter 6, we developed a framework to match these dimensions with supply chain strategies (see Sections 6.2 and 6.3).

The development chain provides a different set of challenges. It can be character- ized by

• Technology clockspeed, that is, the speed by which technology changes in a partic- ular industry. This clearly has an impact on product design and hence on the devel- opment chain.

• Make/buy decisions, or decisions regarding what to make internally and what to buy from outside suppliers. (For more details, see our discussion in Chapter 9.)

• Product structure, that is, the level of modularity or integrality that a product must have. Later in this chapter, we discuss the concept of modularity in products in more detail, but for the purposes of this section, it is sufficient to say that a highly modu- lar product is assembled from a variety of modules, and for each module there may be several options. In this way, the bulk of manufacturing can be completed before the selection of modules and assembly into the final product takes place.

Clearly, each of these characteristics has a significant impact on the supply chain strategy that the firm must use. Indeed, the supply chain strategy for a fast clockspeed product, for example, PCs or laser printers, is quite different than the supply chain strategy for a slow clockspeed product, for example, airplanes. Similarly, the level of outsourcing, the supplier footprint, and product architecture also have an impact on the supply chain strategy.

Concepts such as the development supply chain and technology clockspeed are directly related the work of Marshall Fisher. In his seminal article, “What Is the Right Supply Chain for Your Product?” [72], Professor Fisher distinguishes between two extreme product types, innovative products and functional products. Functional products are characterized by slow technology clockspeed, low product variety, and typically low profit margins. Examples include grocery products such as soup and beer, tires, office equipment, and so forth. Innovative products, on the other hand, are characterized by fast technology clockspeed and short product life cycle, high product variety, and relatively high margins.

So, what is the appropriate supply chain strategy and the product design strategy for each product type? Obviously, products with fast clockspeed (innovative products) require a different approach than products with slow clockspeed (functional products). At the same time, both the supply chain strategy and the product design strategy must take into account the level of demand uncertainty.

CHAPTER 11: COORDINATED PRODUCT AND SUPPLY CHAIN DESIGN 339

sim8239X_ch11_331-364.qxd 5/9/07 12:01 PM Page 339

Figure 11-3 provides a framework for matching product design and supply chain strategies with the characteristics of the development chain (clockspeed) and the sup- ply chain (demand uncertainty). The horizontal axis provides information on demand uncertainty while the vertical axis represents product introduction frequency, or product clockspeed.

As we have already observed in Chapter 6, everything else being equal, higher demand uncertainty leads to a preference for managing the supply chain based on a pull strategy. Alternatively, smaller demand uncertainty leads to an interest in managing the supply chain using a push strategy. Recall the characteristics of push and pull supply chain strategies. In a push strategy, the focus is on predictable demand, leveraging high economies of scale, and achieving cost efficiency. In contrast, in a pull supply chain, the focus is on reacting to unpredictable demand, dealing with low economy of scale, and achieving responsiveness, which in part is achieved by aggressively reduc- ing lead times.

Similarly, everything else being equal, high product introduction frequency (fast clock speed) suggests a focus on modular product architecture since this allows the independent development of product subcomponents so that final feature set selection and product differentiation are postponed as much as possible, sometimes until demand is realized. (We discuss these ideas in significantly more detail later in this chapter in Section 11.2.4.) On the other hand, speeding up product development and postponing differentiation, and thus product modularity, is not that important when product introduction frequency is low (i.e., for slow clockspeed products).

In Figure 11-3, we partition the region spanned by these two dimensions, demand uncertainty and product introduction frequency, into four boxes. Box A represents products that are characterized by predictable demand and slow product introduction frequency. Examples include products such as diapers, soup, and pasta. Our framework suggests that, in this case, the focus is on a push strategy, supply chain efficiency, and high inventory turns.

Box B represents products with fast clockspeed and highly unpredictable demand. Many high-tech products such as PCs, printers, and cell phones as well as fashion

340 DESIGNING AND MANAGING THE SUPPLY CHAIN

FIGURE 11-3 The impact of demand uncertainty and product introduction fre- quency on product design and supply chain strategy.

Product introduction

frequency Product

architecture

Modular

Integral

Supply chain strategy Push Pull

Demand uncertaintyHL

L

H

C

A

B

D

sim8239X_ch11_331-364.qxd 5/9/07 12:01 PM Page 340

EXAMPLE 11-1

To test our framework, consider products such as televisions. In general, the technology in this industry does not change very frequently, although manufacturers change models frequently and recently we have seen a significant move from old technology such as cathode-ray tubes to flat-screen panels. Thus, product introduction frequency is high but perhaps not as high as that of PCs. Customer demand uncertainty is not very high, and, hence, demand is predictable, except for promotional events and the fact that recently product prices dropped by as much as 10 percent a month and this affected the level of demand. Thus, televisions fit somewhere on the line between boxes B and D in Figure 11-3, perhaps closer to the central vertical line. What are the product design and supply chain strategies used in this industry? Interestingly, depending on the manufacturer and the market area, a modular product architecture and lead time reduc- tion strategies. Indeed, while most manufacturing is done in China, the business strategy depends on the final destination. Sophisticated TV manufacturers ship components to poor countries and assemble the products at the market based on customer demand. This push–pull strategy requires modular design and allows manufacturers to reduce cost and satisfy legal requirements that sometime demand assembly by local companies. On the other hand, for the U.S. market, the focus recently has been on lead time reduction from the old 90-day lead time from manufacturing to arrival at the stores to about 30 days. Such short lead times reduce inventory significantly, and the supply chain is thus less vulnerable to the impact of a 10 percent monthly drop in product price [16].

items belong to this category. Here, the focus is on responsiveness, on pull strategy as well as modular product architecture. Thus, products in this category require a supply chain that values responsiveness over cost—the Zara case study in Chapter 9 illustrates how this can be done through excess capacity and reduction of lead time to customers. In many cases, dynamic pricing strategies also are utilized to better match supply and demand, as we discuss in Chapter 13.

Box D represents products with slow clockspeed but high demand uncertainty. These are the products and industries where a combination of push and pull is essential. And, similarly to box B, these are also situations where lead time reduction, if possible, is important. Examples of products in this category include high-end furniture, chem- ical products such as agrochemicals, commodity and specialty chemicals, and prod- ucts such as (large-diameter) tires used in the mining industry where volume is relatively small and, hence, demand is highly unpredictable.

Finally, box C represents products with fast clockspeed and low demand uncer- tainty. There are not many products with these characteristics, but one example that comes to mind is the cell phone engine. Indeed, many cell phone manufacturers use the same engine in all their phones, so that demand for the engine is an aggregation of demand for all their phones. Thus, demand uncertainty is low. The cell phone engine by itself does not have modular product architecture, but it is part of a modular product. Again, similarly to box A, the focus here is on a push supply chain emphasizing effi- ciency or cost reduction.

CHAPTER 11: COORDINATED PRODUCT AND SUPPLY CHAIN DESIGN 341

In the next section, we discuss a series of concepts introduced by Professor Hau Lee [118] and known collectively as design for logistics (DFL). These concepts sug- gest product and process design approaches that help to control logistics costs and increase customer service levels.

Following that, we discuss the advantages of including suppliers in the product design process. This discussion is based on an extensive report issued by the Global Procurement and Supply Chain Benchmarking Initiative at Michigan State

sim8239X_ch11_331-364.qxd 5/9/07 12:01 PM Page 341

University, which is titled “Executive Summary: Supplier Integration into New Product Development: A Strategy for Competitive Advantage”[145].

Finally, we discuss the concept of mass customization, developed by Joseph Pine II with several co-authors. In particular, we focus on the ways in which advanced logis- tics and supply chain practices help to enable this exciting new business model.

11.2 DESIGN FOR LOGISTICS

11.2.1 Overview Transportation and inventory costs, as we have seen, are often critical supply chain cost drivers, particularly when inventory levels must be kept fairly high to ensure high service levels. These are exactly the issues that DFL addresses, using the following three key components [118]:

• Economic packaging and transportation. • Concurrent and parallel processing. • Standardization

Each of these components addresses the issue of inventory or transportation costs and service levels in complementary ways. They are discussed in detail in the follow- ing subsections.

11.2.2 Economic Packaging and Transportation Of the various DFL concepts, perhaps the most obvious involves designing products so that they can be efficiently packed and stored. Products that can be packed more compactly are cheaper to transport, particularly if delivery trucks “cube out” before they “weigh out.” In other words, if the space taken up by a product and not its weight constrains how much can fit in a delivery vehicle, products that can be stored more compactly can be transported less expensively.

342 DESIGNING AND MANAGING THE SUPPLY CHAIN

EXAMPLE 11-2

Swedish furniture retailer Ikea, with about $18 billion in sales, is the world’s largest furniture retailer. Started in Sweden by Ingvar Kamprad, Ikea currently has 220 stores in 33 countries [102, 222]. It has grown so dramatically by “reinventing the furniture business” [130]. Traditionally, furniture sales were split between department stores and small, locally owned shops. Typically, customers would place an order, and delivery could take place up to two months after the order was placed.

Ikea changed that formula by displaying all of its 10,000 products in large warehouse-like spaces in out-of-town stores and keeping all of these items in the warehouse. This was accom- plished by designing products so that they can be packed compactly and efficiently in kits, which customers take from the stores and assemble at home. These kits are easy and cheap to transport, so products can be manufactured efficiently in a small number of factories and then shipped rela- tively cheaply to stores all over the world. Since Ikea has so many stores, each of which is very large, the company is able to take advantage of vast economies of scale. This has enabled the firm to sell good-quality furniture at prices lower than that of its competitors [130].

Ikea continues to work toward improved design and packaging to continue its dramatic growth— “recently the company figured out how to shave one-third off the width of bookcase packing boxes by making the back panels a separate assembly piece” [164].

There are other reasons to design products to pack compactly. For example, many major retailers favor products that take up less storage space and stack easily. Efficient storage reduces certain components of inventory cost because handling costs typically

sim8239X_ch11_331-364.qxd 5/9/07 12:01 PM Page 342

CHAPTER 11: COORDINATED PRODUCT AND SUPPLY CHAIN DESIGN 343

EXAMPLE 11-4

The Hawaiian sugar industry switched over to bulk transportation after World War II, when costs began to increase. They estimate that the cost of transporting a bulk ton of sugar is about $0.77 today, whereas the cost of transporting the same quantity of sugar in bags would be about $20.00 [56].

decrease, space per product (and thus rent per product) decreases, and revenue per square foot can increase. For example, many of the large plastic items available in dis- count stores, such as garbage pails, are designed to stack, so that they take up less shelf (or floor) space in the store. Thus, while it might not be enough to design pack- aging efficiently after the product design is completed, it may be valuable to redesign the product itself in order to take these issues into account.

Similarly, it is often possible to ship goods in bulk and only complete final packag- ing at the warehouse or even at the retailer. This may save on transportation costs because bulk goods tend to be shipped more efficiently.

EXAMPLE 11-3

Recently Rubbermaid won several design awards from BusinessWeek magazine. When describing why the Clear Classics food storage containers won an award, the writers mention that “Wal-Mart loves products designed to fit 14-by-14-inch shelves,” which is one of the reasons these products were so successful. In addition, when describing the children’s Icy Rider sled designed by Rubbermaid (which also won the award), the writers state, “Of course, not all products sold in Wal-Mart can fit into 14-by-14 shelving. But if designers create them to stack and save space, they have a shot of selling toWal-Mart . . . After researching Wal-Mart’s needs, Rubbermaid made the Icy Rider thin and stackable” [152].

In some cases, final packaging can even be delayed until the goods are actually sold. For example, many grocery stores now sell flour, cereal, honey, liquid soap, rice, beans, grains and many other goods in bulk, allowing consumers to package as much as they want.

Recall that cross-docking (see Chapter 7) involves moving goods from one truck (e.g., from the supplier) to another set of trucks (e.g., perhaps going to individual retail stores). In some cases, boxes or pallets are taken off an incoming truck and moved directly to an outgoing one. However, it is often necessary to repackage some of the products. In many cases, bulk pallets of single items come in from suppliers, but mixed pallets with many different items have to go out to individual retailers. In this case, goods must be repacked at the cross-dock point, so more identification or label- ing also might be needed if packages are broken up [187]. In general, packaging and products that are designed to facilitate this type of cross-docking operation by making repacking easier will clearly help to lower logistics costs.

11.2.3 Concurrent and Parallel Processing In the previous section, we focused on simple ways that redesign of the product and packaging could help control logistics costs. In this subsection, we will focus on mod- ifying the manufacturing process—which also may require modification of the product design.

sim8239X_ch11_331-364.qxd 5/9/07 12:01 PM Page 343

EXAMPLE 11-5

A European manufacturer produces network printers for the European market in alliance with a manufacturer in the Far East. The main printer PC board is designed and assembled in Europe. It is then shipped to Asia, where it is integrated with the main printer housing in a process that involves building the printer, including the motor, printhead, housing, and so forth, around the board. The fin- ished product is then shipped to Europe. The manufacturer is concerned with the long production and transportation lead times, which make it essential to maintain a large safety stock in Europe. However, much of the long manufacturing lead time is due to the sequential manufacturing process. Redesigning the printer manufacturing process and product so that the board can be integrated with the rest of the printer at the end of the manufacturing process will decrease lead times by allowing parallel manufacturing in Europe and the Far East. In addition, moving final assembly to Europe can serve to further increase responsiveness and decrease lead times. The two manufac- turing processes are diagrammed in Figure 11-4 [118].

We have seen that many difficulties in operating supply chains are due to long manufacturing lead times. Most manufacturing processes consist of manufacturing steps performed in sequence. The requirements of short start-up times and ever- shorter product life cycles often dictate that certain manufacturing steps be per- formed in different locations to take advantage of existing equipment or expertise. Concurrent and parallel processing involves modifying the manufacturing process so that steps that were previously performed in a sequence can be completed at the same time. This obviously helps reduce manufacturing lead time, lower inventory costs through improved forecasting, and reduce safety stock requirements, among other benefits.

A key to keeping the manufacturing process parallel is the concept of decoupling. If many of the components of the product can be decoupled, or physically separated, dur- ing manufacturing, it is possible that these components can be manufactured in parallel. If manufacturing each of the individual components takes the same amount of time in the newly decoupled design, but the manufacturing steps are performed in parallel, lead time will decrease. Even if some of these modular components take slightly more time to manufacture, the overall lead time may still decrease since various components are being manufactured in parallel. An added advantage of this manufacturing strategy of decoupling is that it may be possible to design different inventory strategies for the various decoupled components. If the supply of raw materials or manufacturing yield is uncer- tain for a particular component, a higher inventory level can be held of that single com- ponent, rather than for the entire end product.

344 DESIGNING AND MANAGING THE SUPPLY CHAIN

FIGURE 11-4 Concurrent processing.

Board Printer

Printer

Board

Housing

Europe

Europe

Asia

Asia

Serial processing

Parallel processing

Customers (Europe)

Customers (Europe)

sim8239X_ch11_331-364.qxd 5/9/07 12:01 PM Page 344

11.2.4 Standardization As we have discussed above, it is possible in some cases to shorten lead times (e.g., by taking advantage of parallel processing) in order to reduce inventory levels and increase forecast accuracy. Sometimes, however, it is impossible to reduce the lead time beyond a certain point. In these cases, it may be possible to achieve the same objectives by taking advantage of standardization.

Recall the third principle of forecasting described in Chapter 2: aggregate demand information is always more accurate than disaggregate data. Thus, we can better fore- cast demand for a continent than a country or for a product family (e.g., ski jackets) than a specific product (or style). Unfortunately, in a traditional manufacturing envi- ronment, aggregate forecasts are not of much use—the manufacturing manager has to know exactly what needs to be made before starting the process. However, by effec- tively using standardization, it may be possible to make effective use of the informa- tion in aggregate forecasts. Specifically, approaches based on product and process commonality make it possible to delay decisions about which specific product will be manufactured until after some of the manufacturing or purchasing decisions have been made. Thus, these decisions can be made on an aggregate level, using the more accurate aggregate forecasts.

Professor Jayashankar Swaminathan has developed a wide-ranging framework for effective implementation of standardization through the use of the correct operational strategy [201]. Swaminathan suggests that product modularity and process modularity are the key drivers that enable a standardization strategy that lowers inventory costs and increases forecast accuracy.

Following Swaminathan, we define the following concepts:

A modular product is a product assembled from a variety of modules such that, for each module, there are a number of options. The classic example of a modular product is the personal computer, which can be customized by combining different video cards, hard drives, memory chips, and so forth. Recall that this concept of modularity is also important for the implementation of concurrent and parallel processing, which was described in the previous subsection. A modular process is a manufacturing process consisting of discrete operations, so that inventory can be stored in partially manufactured form between operations. Products are differentiated by completing a different subset of operations during the manufacturing process. Observe that modular products are not necessarily made of modular processes, as it may not be possible to store intermediate, or semifinished, inventories.

Swaminathan identifies four different approaches to standardization:

• Part standardization • Process standardization • Product standardization • Procurement standardization

In part standardization, common parts are used across many products. Common parts reduce required part inventories due to risk pooling and reduce part costs due to economies of scale. Of course, excessive part commonality can reduce product differ- entiation, so that less expensive customization options might cannibalize sales of more expensive parts. Sometimes, it is necessary to redesign product lines or families to achieve commonality.

CHAPTER 11: COORDINATED PRODUCT AND SUPPLY CHAIN DESIGN 345

sim8239X_ch11_331-364.qxd 5/9/07 12:01 PM Page 345

346 DESIGNING AND MANAGING THE SUPPLY CHAIN

EXAMPLE 11-6

Benetton is a major supplier of knitwear, Europe’s largest clothing manufacturer, and the world’s largest consumer of wool in the garment sector [223], supplying hundreds of shops. The nature of the fashion industry is that consumer preferences change rapidly. However, because of the long manufacturing lead time, store owners frequently had to place orders for wool sweaters up to seven months in advance of when the sweaters would appear in their stores. The wool sweater manufac- turing process typically consists of acquiring yarn, dyeing it, finishing it, manufacturing the garment parts, and then joining those parts into a completed sweater. Unfortunately, this left little flexibility to respond to the changing tastes of consumers. To address this issue, Benetton revised the manufac- turing process, postponing the dyeing of the garments until after the sweater was completely assembled. Thus, color choices could be delayed until after more forecasting and sales information were received. Hence, because of the postponement of the dyeing process, yarn purchasing and manufacturing plans could be based on aggregate forecasts for product families, rather than fore- casts for specific sweater/color combinations. This revised process made sweater manufacturing about 10 percent more expensive and required the purchasing of new equipment and the retraining of employees. However, Benetton was more than adequately compensated by improved forecasts, lower surplus inventories, and, in many cases, higher sales [192].

Process standardization involves standardizing as much of the process as possi- ble for different products, and then customizing the products as late as possible. In this case, products and manufacturing processes are designed so that decisions about which specific product is manufactured—differentiation—can be delayed until after manufacturing is under way. The manufacturing process starts by making a generic or family product that is later differentiated into a specific end-product. For this rea- son, this approach is also known as postponement or delayed product differentiation [118]. By delaying differentiation, production starts can be based on aggregate fore- casts. Thus, design for delayed product differentiation can be effectively used to address the uncertainty in final demand even if forecasts cannot be improved.

It is usually necessary to redesign products specifically for delayed differentiation. For example, it may be necessary to resequence the manufacturing process to take advantage of process standardization. Resequencing refers to modifying the order of product manufacturing steps so that those operations that result in the differentiation of specific items or products are postponed as much as possible. One famous and dra- matic example of a firm utilizing resequencing to improve its supply chain operation is Benetton Corporation.

EXAMPLE 11-7

A major U.S. manufacturer of mass storage devices makes different unique hard-drive products for each of a variety of customers. Orders are placed to be delivered by a certain time and, since lead times are very long, the manufacturer has to keep a variety of products in process in order to meet promised delivery dates. Since variability of demand is high and each product is unique, the manu- facturer has to maintain high levels of in-process inventory to meet demand reliably. The manufac- turing process involves a brief generic segment, through which products intended for all customers must go, and then an extensive customization portion. Clearly, the ideal point to hold inventory is before customization begins. Unfortunately, however, the majority of manufacturing time, due par- ticularly to time-consuming testing, occurs after differentiation has started. This testing has to take place after differentiation starts because a particular circuit board has to be added to the assembly for the testing to take place, and this circuit board is different for each customer. In order to delay

A U.S. disk drive manufacturer provides another notable example. Notice in this example that although lower levels of inventory need to be held to achieve specific service levels, the per-unit inventory cost tends to be more expensive.

sim8239X_ch11_331-364.qxd 5/9/07 12:01 PM Page 346

CHAPTER 11: COORDINATED PRODUCT AND SUPPLY CHAIN DESIGN 347

EXAMPLE 11-8

A major printer manufacturer was preparing to introduce a new color printer into the market. Demand for the new printer and an existing printer was expected to be highly variable and nega- tively correlated. The manufacturing processes for the two products were similar, except that differ- ent circuit boards and printhead assemblies were used. Differences in head assemblies and circuit boards led to very different manufacturing processes. To implement process standardization, that is, delayed differentiation, it is necessary to ensure that the manufacturing processes are similar until the final step. To do this, the printers have been redesigned so that both products share a com- mon circuit board and printhead. This ensures that differentiation can be delayed as much as possi- ble. Thus, part standardization enables process standardization in this case [118].

Part and process standardization are frequently connected. Sometimes part stan- dardization is necessary for implementing process standardization.

EXAMPLE 11-7 C o n t i n u e d

differentiation, it is possible to insert a generic circuit board into the assembly, complete much of the testing, remove the generic circuit board, and add the customer-specific boards later. In this way, disk drive differentiation can be delayed until more order information is available. Clearly, this will decrease the level of required in-process inventory needed to meet demand reliably. However, this will add some additional manufacturing steps. In particular, the generic board has to be added and removed. Thus, it is necessary to compare the manufacturing inefficiencies caused by adding and removing this circuit board with the gains in inventory savings. The manu- facturing processes are illustrated in Figure 11-5 [118].

FIGURE 11-5 Delaying differentiation.

Before redesign

After redesign

PCB inserted

Generic PCB PCB inserted

Testing

Testing End product

End product

In some cases, the concepts of resequencing and commonality allow the final man- ufacturing steps to be completed at distribution centers (DCs) or warehouses instead of at the factory. One of the advantages of this approach is that if DCs are much closer to the demand than the factories, products can be differentiated closer to the demand, thus increasing the firm’s ability to respond to rapidly changing markets. This is one of the approaches we will discuss in more detail in Section 11.2.8, when we analyze the case from the beginning of this chapter.

Sometimes, processes can be redesigned so that the differentiating steps don’t have to be performed in a manufacturing facility or distribution center at all, but can take place in the retailer after the sale is made. Often this is accomplished by focusing on modularity during the design phase, placing functionality in modules that can be easily added to a product. For example, many laser printers and copiers are packaged in their most basic version. Along with the printer, each retail store stocks separately packaged modules that add features to the printer or copier, such

sim8239X_ch11_331-364.qxd 5/9/07 12:01 PM Page 347

as advanced paper handling, stapling, and so forth. Obviously, this can greatly lower required inventory since only extended features can be stocked in module form, instead of entire printers.

In product standardization, a large variety of products may be offered, but only a few kept in inventory. When a product not kept in stock is ordered, the order may be filled by a product that offers a superset of the features required by a cus- tomer. This process, known as downward substitution, is common in many indus- tries. For example, in the semiconductor industry, it is quite common to sell a higher-speed or a higher-functionality chip as a lower-speed/functionality chip when the low-end chip is out of stock. Similarly, car rental agencies and hotels fre- quently fill reservations with higher-end vehicles or rooms when the lower-end vehicles or rooms are not available. Sometimes, it may be possible to redesign products so that one product can be adjusted to meet several end-customer require- ments. For example, as we have seen previously, many products are similar, except that power supplies have to be different for different markets. Instead of manufac- turing two versions of a product, however, manufacturers can utilize a standardized product, with a switchable power supply. We discuss this issue further in the case at the end of this chapter.

Finally, procurement standardization involves standardizing processing equip- ment and approaches, even when the product itself is not standardized. This is particu- larly valuable when processing equipment is very expensive. In the production of application-specific integrated circuits (ASICs), for example, very expensive equip- ment is required. Although end-products are highly customized and demand is unpre- dictable, the same equipment is used to produce each of the possible end-products. Thus, equipment procurement can be managed independent of the final demand.

11.2.5 Selecting a Standardization Strategy To help with the selection of the appropriate standardization strategy, Swaminathan [201] proposed a framework based on the observation that the firm’s choice of standardization strategy is a function of the firm’s ability to modularize its products and processes. Table 11-2 illustrates the proposed strategic choices under different conditions.

• If process and product are modular, process standardization will help to maximize effective forecast accuracy and minimize inventory costs.

• If the product is modular, but the process is not, it is not possible to delay differenti- ation. However, part standardization is likely to be effective.

• If the process is modular but the product is not, procurement standardization may decrease equipment expenses.

• If neither the process nor the product is modular, some benefits may still result from focusing on product standardization.

348 DESIGNING AND MANAGING THE SUPPLY CHAIN

TABLE 11-2

OPERATIONAL STRATEGIES FOR STANDARDIZATION

Process

Nonmodular Modular

Part Process Modular standardization standardization

Product Product Procurement Nonmodular standardization standardization

sim8239X_ch11_331-364.qxd 5/9/07 12:01 PM Page 348

11.2.6 Important Considerations The various strategies described above are designed to deal with inaccurate forecasts and product variety; frequently, it may not be possible or cost-effective to implement these strategies in the context of a particular product or a specific supply chain. Even if implementing a particular strategy is theoretically possible, in many cases the expenses resulting from product and packaging redesign will exceed the savings under the new system. In addition, capital expenditures are likely to be required to retool assembly lines. Sometimes, as we discussed above, it may even be necessary to add manufacturing capability at distribution centers. Typically, the value of these types of changes is higher at the start of the product life cycle, when expenditures can be amortized over the entire life of the product. It is possible that DFL initiatives that make a great deal of sense at the start of the product life cycle don’t pay for them- selves when implemented later [118].

It also may be more expensive to manufacture a product with a new process. In many of the examples mentioned above, the products and manufacturing processes became more expensive. It is therefore necessary to estimate the savings produced by a more effectively designed product or process, and compare these savings to the increased cost of manufacturing. Many of the benefits of implementing such a system are very difficult to quantify; increased flexibility, more efficient customer service, and decreased market response times may be hard to place a value on, which only serves to make the analysis more difficult. To add to the difficulty, engineers are often forced to take a broader perspective than they have been trained to take when they are making these kinds of decisions.

To add to these complications, process modifications such as resequencing will cause the level of inventory in many cases to go down, but the per unit value of inven- tory being held will be higher. In the sweater example, it may be possible to hold less wool in inventory because it doesn’t have to be dyed before it is assembled. However, much of this wool will be held in the form of sweaters, which have a higher value than dyed wool. Of course, if manufacturing or customizing steps are postponed, the generic products may have a lower value than customized products, so value is added later in the supply chain than it would be otherwise.

Finally, in some cases, tariffs and duties are lower for semifinished or nonconfig- ured goods than for final products [118]. Thus, implementing a strategy of completing the manufacturing process in a local distribution center may help to lower costs asso- ciated with tariffs and duties.

All of these issues have to be taken into consideration when implementing a specific design for logistics strategy. Nevertheless, it is clear that in many cases, DFL can help to improve customer service and greatly reduce the costs of operating the supply chain.

11.2.7 The Push–Pull Boundary Recall our discussion of the push–pull boundary in Chapter 6. In push-based systems, production decisions are based on long-term forecasts, while in pull-based supply chains, production is demand driven. We listed many advantages of pull-based sys- tems and concluded that compared to push-based systems, pull-based systems typi- cally lead to a reduction in supply chain lead times, inventory levels, and system costs, while simultaneously making it easier to manage system resources.

Unfortunately, it is not always practical to implement a pull-based system through- out the entire supply chain. Lead times may be too long, or it may be necessary to take advantage of economies of scale in production or transportation. The standardization

CHAPTER 11: COORDINATED PRODUCT AND SUPPLY CHAIN DESIGN 349

sim8239X_ch11_331-364.qxd 5/9/07 12:01 PM Page 349

350 DESIGNING AND MANAGING THE SUPPLY CHAIN

strategies we have been discussing in this section can be viewed as a method to com- bine push and pull systems within a single supply chain into what we called, in Chapter 6, a push–pull system. Indeed, that portion of the supply chain prior to prod- uct differentiation is typically a push-based supply chain. That is, the undifferentiated product is built and transported based on long-term forecasts. In contrast, differentia- tion occurs as a response to market demand. Thus, the portion of the supply chain starting from the time of differentiation is a pull-based supply chain.

For instance, in the Benetton example, Example 11-6, uncolored sweaters are made to forecast, but dyeing takes place as a reaction to customer demand. We call the point of differentiation the push–pull boundary, since this is the point where the system changes from a push-based system to a pull-based system.

One way to view the push–pull boundary concept is through the third rule of inven- tory management, discussed in Chapter 2. Since aggregate demand information is more accurate than disaggregate data, the push portion of the supply chain includes only activities and decisions made prior to product differentiation. These activities and decisions are based on aggregate demand data.

Clearly, then, an additional advantage of postponement is that it allows firms to realize many of the advantages of pull-based systems, while at the same time allowing for the economies of scale inherent in push-based systems. Often, when implementing a standardization strategy, if there is more than one possible differentiation point, it may be useful to think in terms of locating the push–pull boundary in order to achieve a balance between the advantages of the push- and the pull-based systems.

11.2.8 Case Analysis Consider the Hewlett-Packard case that you read at the beginning of this chapter. Although several problems and issues are outlined in the case, we will focus on analyzing the inventory problems in the European distribution center. In particular, HP faces long delivery lead times of about four to five weeks from its production facility in Vancouver, Washington, to Europe. The Vancouver plant is a high-speed, high-volume facility where manufacturing takes about a week.

In particular, HP is concerned with high inventory levels and inventory imbalance in Europe. One of the characteristics of the DeskJet product line is that it is cus- tomized for local markets, a process called localization. This involves adding labeling and documentation in the correct language and customizing the power supply for the correct voltage level and plug. Customization is done in Vancouver many weeks before the products arrive in Europe. Furthermore, once the printers arrive in Europe, inventory imbalance might occur in the following sense: The European DC often finds itself with too much inventory of printers customized for certain markets, and not enough inventory of printers customized for others.

What are the causes of these problems? Based on the case and material we have discussed in previous chapters, the following issues are clear:

• There is significant uncertainty about how to set the correct inventory level. • The many different localization options make inventory difficult to manage. • Long lead times lead to difficulty in forecasting and high safety stocks. • Uncertainty in the many local markets makes forecasting difficult. • Maintaining cooperation between the various HP divisions is challenging.

In the short term, the first issue can be addressed by rationalizing safety stock uti- lizing the methods we discussed in Chapter 2. To address these problems in the longer term, the following solutions have been proposed:

sim8239X_ch11_331-364.qxd 5/9/07 12:01 PM Page 350

• Switch to air shipments of printers from Vancouver. • Build a European factory. • Hold more inventory at the European DC. • Improve forecasting practices.

Unfortunately, there are significant problems with each of these suggestions. Air shipments are prohibitively expensive in this competitive, low-margin business. European volumes are not sufficient to justify building a new factory. Inventory is already a problem; more would simply magnify the problem. Finally, it is unclear how to improve forecasts.

Thus, HP management is motivated to consider another option: process standard- ization or postponement. Specifically, this option involves shipping “unlocalized” printers to the European DC and localizing them after observing local demand. The question is, what are the inventory savings of such a strategy? To address this issue, we utilize the inventory management policies detailed in Chapter 2.

Recall that we can calculate required safety stock for each of the customized prod- ucts by noting that safety stock must equal , where z is selected to maintain the required service level (see Table 2-2). In the analysis below, we assume that lead time is five weeks, and we require a 98 percent service level. By dividing this quantity by average demand, we determine the number of weeks of safety stock required. Thus, the first six rows of Table 11-3 contain the results of these calculations for each of the customization options specified in Table 11-1. The second-to-last row totals all of the required safety stock. We see that by utilizing effective inventory man- agement strategies and the current distribution system, HP needs over three-and-a-half weeks of safety stock on hand to meet the 98 percent service level requirement. The table also shows the effect of postponing localization until after demand is observed. In this case, the DC keeps safety stock of only the generic printer, customizing the printers as demand is realized. This allows the DC to focus on aggregate demand lev- els and, therefore, as we saw in the section on risk pooling in Chapter 2, aggregate demand has a much smaller standard deviation than individual demand. The standard deviation of the aggregate demand is calculated in the last row of the table. This new standard deviation is used to determine safety stock for the generic model. Observe that this new system, in which localization is postponed, requires less safety stock than the currently existing system.

The dollar savings in inventory carrying cost obviously depend on the rate of carry- ing cost used. For example, if carrying cost is taken to be 30 percent and a product

z ! STD ! 2L

CHAPTER 11: COORDINATED PRODUCT AND SUPPLY CHAIN DESIGN 351

TABLE 11-3

INVENTORY ANALYSIS

Standard Standard Average deviation Average deviation Weeks monthly of monthly weekly of weekly Safety of safety

Parameter demand demand demand demand stock stock

A 42.3 32.4 9.8 15.6 71.5 7.4 AA 420.2 203.9 97.7 98.3 450.6 4.6 AB 15,830.1 5,624.6 3,681.4 2,712.4 12,433.5 3.4 AQ 2,301.2 1,168.5 535.1 563.5 2,583.0 4.8 AU 4,208.0 2,204.6 978.6 1,063.2 4,873.6 5.0 AY 306.8 103.1 71.3 49.7 227.8 3.2 Total 23,108.6 5,373.9 20,640.0 3.8 Generic 23,108.6 6,244 5,373.9 3,011.1 13,802.6 2.6

sim8239X_ch11_331-364.qxd 5/9/07 12:01 PM Page 351

value of $400 is assumed, annual savings are about $800,000. In addition, there are other benefits to implementing a postponement strategy. These include

• The value of inventory in transit, and thus insurance costs, goes down. • It may be possible to reduce freight handling costs. • Some of the localization materials can be locally sourced, reducing costs and meet-

ing “local content” requirements.

On the other hand, there are costs associated with implementing this process stan- dardization strategy. First, the product and packaging have to be redesigned so that localization can be delayed. This entails expense and requires research-and-development work on a product that is already working well. Also, the European distribution center has to be modified to facilitate localization there. Recall that, in addition to capital investments, the mind-set of the distribution operation—“Distribution, not manufac- turing, is our core competency”—has to be changed.

Hewlett-Packard did indeed implement such a strategy, with great success. Inventory declined while service levels rose, leading to significant cost savings and increased profitability. To achieve these results, the printer was redesigned for local- ization and the distribution center took on more work and responsibilities.

11.3 SUPPLIER INTEGRATION INTO NEW PRODUCT DEVELOPMENT

Another key supply chain issue involves the selection of appropriate suppliers for components of the new product. Traditionally, this has been done after design and manufacturing engineers have determined the final design for a product. Recently, a study by the Global Procurement and Supply Chain Benchmarking Initiative, at Michigan State University [145], found that firms often realize tremendous benefits from involving suppliers in the design process. Benefits include a decline in pur- chased material costs, an increase in purchased material quality, a decline in develop- ment time and cost and in manufacturing cost, and an increase in final product technology levels.

In addition to the competitive forces that drive managers to seek out all types of supply chain efficiencies, several competitive forces are specifically encouraging managers to find opportunities to work with suppliers during the product design pro- cess. These forces include the continuing focus on strategies that encourage compa- nies to focus on their core competencies and outsource other business capabilities, and to continually reduce the length of product life cycles. Both of these forces encourage companies to develop processes that make the design process more efficient. Taking advantage of supplier competencies is certainly one way to do this.

11.3.1 The Spectrum of Supplier Integration The supplier integration study [145] notes that there is no single “appropriate level” of supplier integration. Instead, they develop the notion of a spectrum of supplier integration. In particular, they identify a series of steps from least to most supplier responsibility as follows:

None. The supplier is not involved in design. Materials and subassemblies are sup- plied according to customer specifications and design. White box. This level of integration is informal. The buyer “consults” with the supplier informally when designing products and specifications, although there is no formal collaboration.

352 DESIGNING AND MANAGING THE SUPPLY CHAIN

sim8239X_ch11_331-364.qxd 5/9/07 12:01 PM Page 352

CHAPTER 11: COORDINATED PRODUCT AND SUPPLY CHAIN DESIGN 353

Grey box. This represents formal supplier integration. Collaborative teams are formed between the buyer’s and the supplier’s engineers, and joint development occurs. Black box. The buyer gives the supplier a set of interface requirements and the supplier independently designs and develops the required component.

Of course, just because the black-box approach is at one end of the continuum doesn’t mean that it is the best approach in all cases. Instead, firms must develop a strategy that helps them determine the appropriate level of supplier integration for dif- ferent situations. The Global Procurement and Supply Chain Benchmarking Initiative has developed a strategic planning process to help firms make this determination [145]. The first several steps of the process are summarized below:

• Determine internal core competencies. • Determine current and future new product developments. • Identify external development and manufacturing needs.

These three steps help management determine what is going to be procured from suppliers and what level of supplier expertise is appropriate. If future products have components that require expertise that the firm does not possess, and development of these components can be separated from other phases of product development, then taking a black-box approach makes sense. If this separation is not possible, then it makes more sense to use the grey-box development. If the buyer has some design expertise but wants to ensure that the supplier can adequately manufacture the compo- nent, perhaps a white-box approach is appropriate.

11.3.2 Keys to Effective Supplier Integration Simply selecting an appropriate level of supplier integration is not sufficient. Much work goes into ensuring that the relationship is a success. The next steps of the strate- gic planning process [145] help to ensure this success:

• Select suppliers and build relationships with them. • Align objectives with selected suppliers.

Selecting suppliers in general involves various considerations such as manufactur- ing capacity and response time. Since supplier integration partners typically supply components (in addition to cooperating in their design), all of the traditional consider- ations still apply. In addition, the special nature of supplier integration presents an additional set of supplier requirements.

The same study identifies many of these, including

• The capability to participate in the design process. • The willingness to participate in the design process, including the ability to reach

agreements on intellectual property and confidentiality issues. • The ability to commit sufficient personnel and time to the process. This may include

colocating personnel if appropriate. • Sufficient resources to commit to the supplier integration process.

Of course, the relative importance of these requirements depends on the particular project and type of integration. Once suppliers are identified, it is critical to work on building relationships with them. For example, firms have found it useful to involve suppliers early in the design process. Companies that do so report greater gains than those that involve suppliers only after design concepts have been generated. Sharing future plans and technologies with suppliers helps to build this relationship, as does a

sim8239X_ch11_331-364.qxd 5/9/07 12:01 PM Page 353

354 DESIGNING AND MANAGING THE SUPPLY CHAIN

joint continuous improvement goal. Separate organizational groups dedicated to man- aging the relationship are also useful. In all of these cases, the goals of the purchasing firm revolve around building long-term, effective relationships with trusted suppliers. These will naturally lead to the alignment of buyer and supplier objectives, which will result in more effective integration.

11.3.3 A “Bookshelf” of Technologies and Suppliers The Michigan State group also developed the idea of a “bookshelf” of technologies and suppliers within the context of supplier integration. This involves monitoring the development of relevant new technologies and following the suppliers that have demonstrated expertise in these technologies. Then, when appropriate, a buyer firm can quickly introduce these technologies into new products by integrating the supplier design team with its own. This enables a firm to balance the advantages and disadvan- tages of being on the cutting edge of new technology. On one hand, there is no need to use the technology immediately in order to gain experience with it: suppliers are developing this knowledge with other customers. On the other hand, the danger of being slow to introduce cutting-edge technology and concepts is lessened. The book- shelf concept is a dramatic example of the power of supplier integration.

11.4 MASS CUSTOMIZATION

11.4.1 What Is Mass Customization? In his book Mass Customization [165], Joseph Pine II introduced a concept that is becoming important to more and more businesses: mass customization. In this section, we will first review the concept and then discuss how logistics and supply chain net- works play an important role in the implementation of related ideas.

Mass customization has evolved from the two prevailing manufacturing paradigms of the 20th century: craft production and mass production. Mass production involves the efficient production of a large quantity of a small variety of goods. Spurred by the Industrial Revolution, so-called mechanistic firms developed in which management put a high priority on automating and measuring tasks. A very bureaucratic manage- ment structure, with rigid, functionally defined groups and tasks, and tightly super- vised employees, is common. This kind of organization enables tight control and predictability, which tends to lead to high degrees of efficiency. The quality of a small number of items can be quite high and prices can be kept relatively low. This is partic- ularly critical for commodity products, where firms have typically competed on price and, more recently, on quality.

Craft production, on the other hand, involves highly skilled and flexible work- ers, often craftsmen in the manufacturing setting, who are governed by personal or professional standards, and motivated by the desire to create unique and inter- esting products or services. These workers, found in so-called organic organiza- tions, are typically trained through apprenticeships and experience; the organization is flexible and continually changing. This type of organization is able to produce highly differentiated and specialized goods, but it is very difficult to regulate and control. As a consequence, the quality and production rates of these goods are hard to measure and reproduce, and they are typically much more expensive to manufacture [166].

In the past, managers often had to make a decision between these two types of organizations with their inherent trade-offs. For some products, a low-cost, low-vari- ety strategy was appropriate while for others, a higher-cost, higher-variety, more

sim8239X_ch11_331-364.qxd 5/9/07 12:01 PM Page 354

CHAPTER 11: COORDINATED PRODUCT AND SUPPLY CHAIN DESIGN 355

adaptable strategy was more effective. The development of mass customization demonstrates that it is not always necessary to make this trade-off.

Mass customization involves the delivery of a wide variety of customized goods or services quickly and efficiently at low cost. Thus, it captures many of the advantages of both the mass production and craft production systems described above. Although not appropriate for all products (e.g., commodity products may not benefit from dif- ferentiation), mass customization gives firms important competitive advantages and helps to drive new business models.

11.4.2 Making Mass Customization Work Pine points out [166] that the key to making mass customization work is highly skilled and autonomous workers, processes, and modular units, so that managers can coordinate and reconfigure these modules to meet specific customer requests and demands.

Each module continually strives to upgrade its capabilities; a module’s success depends on how effectively, quickly, and efficiently it completes its task, and how good it is at expanding its capabilities. Managers are charged with determining how these capabilities “fit together” efficiently. Thus, management’s success depends on how effectively it can develop, maintain, and creatively combine the links between modules in different ways to meet different customer requests, and on the creation of a work environment that encourages the development of a variety of different modules.

Since each unit has highly specialized skills, workers can develop expertise and efficiency in the manner of mass production. Since these units or modules can be assembled in many ways, the differentiation of craft production is achievable. Pine calls this type of organization a dynamic network.

There are several key attributes that a company, or, more specifically, the systems within a company that link different modules, must possess to implement mass cus- tomization successfully [166]. They are

Instantaneousness. Modules and processes must be linked together very quickly. This allows rapid response to various customer demands. Costless. The linkages must add little if any cost to the processes. This attribute allows mass customization to be a low-cost alternative. Seamless. The linkages and individual modules should be invisible to the cus- tomer, so customer service doesn’t suffer. Frictionless. Networks or collections of modules must be formed with little overhead. Communication must work instantly, without taking time for the team building; this is necessary in so many other types of environments.

With these attributes in place, it becomes possible to design and implement a dynamic, flexible firm that can respond to varying customer needs quickly and efficiently.

EXAMPLE 11-9

National Bicycle is a subsidiary of Matsushita that sells bicycles under the Panasonic and National brand names in Japan. Several years ago, management found that sales were not at acceptable levels, primarily because the company was unable to predict and satisfy varying customer demand. In the year before beginning the mass customization efforts, 20 percent of bicycles from the previ- ous year remained in inventory. Rather than market to a particular niche or try to improve forecasts, National became a mass customizer.

The company developed a highly flexible bicycle frame manufacturing facility, noting that paint- ing and the installation and tuning of components were separate functions that could be performed by other “modules” in its manufacturing facility. Next, they installed a sophisticated custom-order

sim8239X_ch11_331-364.qxd 5/9/07 12:01 PM Page 355

356 DESIGNING AND MANAGING THE SUPPLY CHAIN

EXAMPLE 11-10

Dell Computer has become one of the dominant players in the PC industry—it sells more systems globally than any computer company [224]—by adopting a unique strategy based on mass cus- tomization. Dell never builds a PC for a customer until the customer’s order has been placed. This allows the customer to specify unique requirements, and Dell builds the computer to these require- ments. A growing majority of orders come in over the Internet. The order-taking system interfaces with Dell’s own supply chain control system, which ensures that inventory is where it needs to be for the computer to be quickly manufactured. In addition, Dell stores very little inventory. Instead, Dell’s suppliers have built warehouses close to Dell’s facilities, and Dell orders parts on a just-in- time basis. By implementing these strategies, Dell has been able to provide customers with exactly what they want very quickly. In addition, inventory costs are low and Dell minimizes the danger of parts obsolescence in the rapidly changing computer industry. In this way, Dell has become one of the dominant players in the desktop PC, laptop, and server markets.

Dell has utilized many of the important concepts we have discussed to achieve its goals. The company is driven by advanced information systems that do everything from taking many of the orders (over the Web) to managing inventory in the supply chain. Strategic partnerships have been established with many of Dell’s suppliers. Dell is even establishing supplier integration partnerships with some of its key suppliers (e.g., 3Com, the network equipment supplier) to ensure that new computers and networking devices are compatible. Finally, Dell has utilized the concept of post- ponement, deferring final assembly of computers until orders have been received, to achieve mass customization [139].

11.4.3 Mass Customization and Supply Chain Management Clearly, many of the advanced supply chain management approaches and techniques that we have discussed in this and earlier chapters are essential if mass customization is to be successfully implemented. This is particularly true if the components in the network stretch across several companies.

The same information technology that is so critical for effective supply chain manage- ment is also critical for coordinating the different modules in the dynamic network and ensuring that together they meet customer requirements. The required system attributes listed above make effective information systems mandatory. Similarly, in many cases, the modules in the dynamic network exist across different firms. This makes concepts such as strategic partnerships and supplier integration essential for the success of mass cus- tomization. Finally, as many of the printer-related examples indicate, postponement can play a key role in implementing mass customization. For instance, postponing regional differentiation until products have reached regional distribution centers facilitates regional customization. As the following example illustrates, postponing differentiation until orders have been received allows customer-specific customization.

EXAMPLE 11-9 C o n t i n u e d

system called the Panasonic Order System at retailers. This system includes a unique machine that measures customer weight and size, and the appropriate dimensions of the frame, position of the seat, and extension of the bar stem. The customers also can select model type, color patterns, and various components. Information from the dealer is instantaneously transmitted to the factory, where a computer-aided design (CAD) system produces technical details in three minutes. The information is transmitted automatically to the appropriate modules, where manufacturing is com- pleted. The bike is then delivered to consumers two weeks later.

Thus, by noting that the production process could be separated into independent production modules in a seamless and essentially costless manner, and by installing sophisticated information systems, National Bicycle was able to increase sales and customer satisfaction without significantly increasing manufacturing costs [71].

sim8239X_ch11_331-364.qxd 5/9/07 12:01 PM Page 356

SUMMARY

In this chapter, we focused on various ways that product design interacts with supply chain management. First, we considered various design for logistics concepts, in which product design is used to lower the costs of logistics. Products designed for effi- cient packaging and storage obviously cost less to transport and store. Designing products so that certain manufacturing steps can be completed in parallel can cut down on manufacturing lead time, leading to a reduction in safety stocks and increased responsiveness to market changes. Finally, standardization enables risk pooling across products, leading to lower inventories, and allows firms to use the information contained in aggregate forecasts more effectively.

Another critical design/supply chain interaction involves integrating suppliers into the product design and development process. We discussed different ways that suppli- ers can be integrated into the development process and considered guidelines for man- aging this integration effectively.

Finally, advanced supply chain management helps to facilitate mass customization. Mass customization involves the delivery of a wide variety of customized goods or services quickly and efficiently at low cost. Obviously, this approach helps to pro- vide firms with important competitive advantages and, just as obviously, effective supply chain management is critical if mass customization is to be successful.

DISCUSSION QUESTIONS

1. List two low clock speed products, two medium clock speed products, and two fast clock speed products.

2. How does a low clock speed impact the product design strategy? How about a fast clock speed?

3. Give an example of a product appropriate for each of the boxes in Figure 11-3. 4. Discuss some examples of products that are designed to lower shipping and storage

costs. 5. How does the proliferation of products, models, and options make the supply chain

more difficult to manage? 6. What are the advantages of downward substitution? What are the disadvantages? 7. What are some products or industries that have been damaged by excessive part

standardization? 8. Discuss some examples of modular and nonmodular products and processes. 9. How do standardization strategies help managers deal with demand variability and

the difficulty of making accurate forecasts? 10.What are the advantages and disadvantages of integrating suppliers into the prod-

uct development process? 11.You are the CEO of a medium-sized apparel manufacturer, and you are considering

a mass customization strategy for some of your products. How will you decide which, if any, of your products are appropriate candidates for mass customization?

CHAPTER 11: COORDINATED PRODUCT AND SUPPLY CHAIN DESIGN 357

sim8239X_ch11_331-364.qxd 5/9/07 12:01 PM Page 357

358 DESIGNING AND MANAGING THE SUPPLY CHAIN

Source: Copyright 1996 by the Board of Trustees of the Leland Stanford Junior University. All rights reserved. Used with permission from the Stanford University Graduate School of Business. This case was written by Professor Hau L. Lee, based on an original case writ- ten by Steven Pious and Toni Cupal. It is intended as the basis for class discussion rather than to illustrate either effective or ineffective handling of an administrative situation. The product and individuals’ names have been disguised.

Hewlett-Packard Company: Network Printer Design for Universality

INTRODUCTION

Sarah Donohoe, manufacturing engineering manager of the network laser printer division at Hewlett- Packard Company (HP), listened intently to her col- leagues at the project review meeting for the development of their latest new product. With Sarah at the meeting were Jane Schushinski, marketing manager; Leo Linbeck, head of product design; and David Hooper, the controller of the division.

The main topic for this meeting was the decision of whether or not to use a universal power supply for the next generation of network laser printer, code-named Rainbow. Previously, printers in the North American and the European market have dis- tinct power supplies and the associated fusers in the main engine of the printer. For North American printers, a 110-volt power supply was installed. For European printers, a 220-volt power supply was added. This printer engine was built by HP’s manu- facturing partner in Japan. Due to the long lead time for engine manufacturing, HP had to specify the requirements of the two types of printers at least 14 weeks ahead. The time that it takes the Japanese partner to commit the printers for shipment, the transportation times, and customs clearance totals about four weeks. Hence, if a universal power sup- ply is used, then HP would have the flexibility of postponing the specification of the printer engine by at least two months in the planning process. Consequently, the production team believed that a universal power supply can enable HP to better respond to the changing demand in the individual markets and reduce its inventory costs.

Linbeck had begun the meeting by reviewing a fax he had received from the Japanese partner. “We have been asking our partner for a universal power supply and fuser for a long time, and now, when we are about to finalize our design of the next generation network printer, they are telling us that designing the new power supply is finally feasible and can be

completed within the time constraints we have set for delivering the product to market on time. However, we must make the decision within the next two weeks so our Japanese partner can line up its design engi- neers to work on the project.” Hooper summed up finance’s position as follows, “I do not know what other costs or benefits to the supply chain will be derived from this new change, but what I do know is that our Japanese partner quoted that universal power supply would increase costs by $30 per unit.”

As the conversation progressed around the room, Hooper’s words became more and more indicative of the group’s feelings as a whole. The only hard number available for analyzing the costs and bene- fits of the change was the $30 increase as quoted by the Japanese partner. If the team was to implement the change, they would have to convince manage- ment that the benefits outweighed the costs. Unfortunately, as the meeting went on, quantifying the advantages and disadvantages appeared more and more difficult.

THE HEWLETT-PACKARD COMPANY

Hewlett Packard was one of Silicon Valley’s legends. Established by two Stanford University graduates, William Hewlett and David Packard, in 1939, the company initially prided itself on supplying superior engineering tools, designed for engineers by engi- neers. As the company grew and diversified, the strong belief in technological innovation as the key to competitive advantage persisted.

Innovation was the key to HP’s strategy. In 1957, Packard expressed his belief in the importance of this capability:

Improvement is accomplished by better methods, better techniques, better machinery and equipment and by people continually finding better ways to do their jobs and to work together as a team. I will never see the day when there is not yet room for improvement.

C A S E

sim8239X_ch11_331-364.qxd 5/9/07 12:01 PM Page 358

CHAPTER 11: COORDINATED PRODUCT AND SUPPLY CHAIN DESIGN 359

Through time, HP’s focus on innovation had brought the world products such as the hand-held calculator and the ink-jet printer. In 1992, the com- pany continued to invest heavily in technology, spending $1.6 billion, or 10 percent of revenue, on research and development. The high levels of invest- ment have paid off: For three straight years, over half of HP’s orders had been for products introduced within the last two years.

CHANGING MARKET CONDITIONS

In the early 1990s, while technological innovation continued to drive the company’s success, many busi- ness units were being forced to compete on other dimensions. In consumer product lines, low prices, broad availability, and ease of use had become com- petitive requirements. Lew Platt, HP’s current presi- dent and chief executive officer, once acknowledged the importance of improving customer service and responsiveness:

We’re not doing as good a job in order fulfillment as we need to. In fact it’s where we get our lowest marks from customers. We have to be a lot easier to do business with. Improvement in order fulfillment will strengthen HP’s competitiveness, increase customer satisfaction, and reduce expenses, so this is an area of great urgency. Along with improving profitability, it’s our top priority.

In addition, product life cycles were continually shrinking, making time to market the difference between maximizing market opportunities and miss- ing them. Nowhere were these demands more important than in the laser printer division. HP held a dominant 57 percent of the worldwide laser printer market, but several formidable competitors, includ- ing Apple, Fuji-Xerox, Kyocera, Oki, and Compaq, had recently entered the market; life cycles had fallen to under three years; and the quality of com- petitive products made consumers willing to switch brands if HP’s price was too far above the market average or if the product was not easily available.

To meet these challenges, HP had aggressively worked to improve its product development process. Cross-functional teams that brought specialists from all functional areas together to create a new product were becoming standard. The primary benefit of such teams was their ability to identify and eliminate potential problems early in the design cycle while the financial and time-to-market costs of changing

the product design were low. As intended, the differ- ent perspectives of the team members often gave rise to heated debates over design decisions.

THE NETWORK PRINTER DIVISION SUPPLY CHAIN

The laser products as a group constitute a major and rapidly rising portion of HP’s revenue. In 1992 the revenue of laser products was $3 billion, but was pro- jected to reach close to $8 billion by 1998. The net- work printer is a high-end laser printer that has networking capabilities and special functionalities. Rainbow, the network printer under development, is a product with much more configurable options and features for the printer, such as memory, stapling abil- ity, firmware, system software, fax modems, paper handling, linkage to print server, scanner, and printer stand. It will be priced between $5,000 to $6,000.

The network printer division at HP currently out- sources the procurement and assembly of the prod- uct’s main engine to a Japanese partner. The components, including the power supply and fuser unit, were fully integrated with a printed circuit board from HP’s Boise factory into the printer engine at the partner’s factory. Monopoly control of one of the key components allowed this partner to require a 14-week lead time from HP.

The design team of Rainbow recognized that the multiple thousands of configurable options for the new product would be a nightmare for forecasting and production planning. Consequently, special efforts were spent in the design of the products so that most of the customization of the products, like the installation of paper input units, cabinet stands, fax modems, paper output units, stapler upgrade package, memory, and print server linkage, can all be carried out at the distribution centers (DCs). Hence, all these options can be installed as accessories at the DCs. In addition, the localization of the product through the inclusion of driver software disks, manu- als, power cords, and front panels (with the correct mix of languages) are also done at the DCs.

Hence, the supply chain process involves the trans- portation of the base printer, almost exclusively by boat, from the partner’s facility to HP’s DCs in either North America or Europe. The shipment process lasted one month. The demand for a network printer in Asia and Latin America was still minimal compared with the demand in North America or Europe.

sim8239X_ch11_331-364.qxd 5/9/07 12:01 PM Page 359

360 DESIGNING AND MANAGING THE SUPPLY CHAIN

Similarly, all necessary accessories and localization materials are also shipped to the DCs from the respec- tive suppliers. Both the printers and other materials are stocked at the DCs. When customer orders from resellers arrive, the printers are customized and local- ized, followed by appropriate labeling and packaging. Final transportation time, typically via truck, to the resellers in each region, the United States or Europe, ranges from a few days to approximately a week.

THE UNIVERSAL POWER SUPPLY DECISION

The Marketing Perspective Jane Schushinski, marketing manager:

I think changing to a universal power supply is a fantastic idea if it does not add cost to the product. Customers will not pay for features that they don’t need, and universal power supply is irrelevant to them—the network printer is not like a portable hair dryer that they would carry with them to travel around the world.

The biggest difficulty we have in marketing is not will there be demand for our product, but how much and where. HP makes great printers. We have always been the leader in innovation, reliability, and service. Rainbow is just the first of our series of new network printer line, and we expect to sell 25,000 per month of the product worldwide, with North America having about 60 percent of the market.

What hurts us is our inability to accurately forecast the mix of demands in geographical regions. We may think that Europe will need 10,000 units and North America 2,000 when the numbers may turn out to be 15 and 15,000 respectively. The problem lies in market conditions where increased competition and constantly changing technical innovations can drastically change the demand for a prod- uct in a few weeks. In addition, there are a lot of firms try- ing to compete on price. This too changes demand. Predicting these changes is quite difficult.

Finally, the long lead time from Japan causes my mar- keting staff to pull their hair out. We have to specify the market for the printer four and a half months ahead of delivery. We estimate that the entire life cycle of the prod- uct is at most 18 months. Four-and-a-half-months lead time in an 18-month market—it’s ridiculous! The last thing that we want is a repeat of the VIPER debacle. That episode has my hair turning prematurely gray. We had so much of that product laying around we started calling our factory the “snake pit!”

It is easy to see why we love the universal power sup- ply. With the universal power supply, we only need to esti- mate worldwide product demand four months ahead of time instead of numbers for each market. We can make the determination of individual market demands much later,

and this postponement will help us create more accurate forecasts and help prevent expensive localization errors.

The VIPER was an earlier-generation HP laser printer. While the printer itself was very successful, the VIPER’s story illustrates the difficulties with demand uncertainties. The VIPER was developed in the same manner as the new printer being consid- ered. The main components of the VIPER were sourced from Japan and resulted in the same three- and-a-half-month lead time to the factories. The product required a dedicated power supply and fuser, 110V or 220V, and these were not interchangeable. Specification of the dedicated power supply, at the beginning of the three and a half months, committed the product either to the North American or the European market.

HP had not forecasted the correct mix of European and North American VIPER demand. The printer was sold out in Europe while demand in the United States was less than anticipated—HP filled a warehouse with unwanted North American printers that could not be used to satisfy demands for the European market without incurring heavy costs of disassembling the printer and reconfiguring the power supply and fuser in the engine. Eventually, heavy discounting, or “fire sales,” was needed to rid the excess inventory, incurring very high cost. Buyers in the North American market now expected HP to reduce printer prices over time. Inadvertently, HP had undercut its ability to command premium prices in the market.

The Product Development Perspective The product life cycle of printers can be divided into three stages: ramp-up, maturity, and end of life. The ramp-up period is the time from the initial introduc- tion of the product until HP’s production volume lev- els off. During this stage the product is usually the only printer on the market providing its distinctive features. The maturity stage reflects a period of increasing competition. Comparable printers will be introduced and price will become a more influential aspect of the product market. In the last stage, end of life, there is fierce competition on all fronts. Retail profits at this stage reach their lowest point as mar- gins are squeezed. It is here that HP aims to intro- duce its next generation product.

When there is an imbalance of demand in North America and Europe, the division can live with the

sim8239X_ch11_331-364.qxd 5/9/07 12:01 PM Page 360

CHAPTER 11: COORDINATED PRODUCT AND SUPPLY CHAIN DESIGN 361

consequence of having excess inventory in one con- tinent and shortages in another, or ship the excess from one continent to another (an operation known as “transshipment”), where the printer is reconfig- ured and sold.

In the end-of-life-stage, in addition to transship- ping the products across the continent to correct for some of the imbalances, the division can also dis- count the product to create demand, dismantle the product and sell the parts to HP’s service division in Roseville, or just write the product off.

Leo Linbeck’s office was stacked high with what must have represented every available trade journal related to printing technology. From behind his HP workstation, he explained his point of view regarding the universal power supply:

While Jane gains “responsiveness,” I’m staring at a $30- per-unit cost increase. With the pressure to lower material costs, the design team would find it hard to justify this seemingly unnecessary increase in material cost. Although the printer engine costs about $1,000 each, so that $30 may not seem that much, every single dollar increase in material cost is a decrease of a dollar in our profit. That is why our design group is getting so much heat to get the material cost down. My concern is that we have no way to reliably predict how much value the so-called benefits of universal power supply truly represent.

Now, I’m the first to admit that I’m no marketing expert, but it’s pretty clear to me that if we could just learn to forecast demand better, this universal supply would literally be a worthless idea. Maybe pumping $30 per unit into improving the forecasting process makes more sense than sending it out the door in a cardboard box. At least in the first case we have some hope of recov- ering it again.

I do agree with Jane’s point regarding the benefits late in the product life cycle. Currently, reconfiguring the prod- uct with a different power supply is a real pain. We have to purchase new power supplies rated at the correct voltage, ship the printers across the Atlantic from the undersold region, swap the power supply, change the fuser electronic circuit and the fuser bulb, and, finally, distribute the prod- uct to retailers. The old power supplies have to be dis- posed of. To make matters worse, there are all kinds of regulatory issues that surface. A universal power supply eliminates all rework that is now required, but whether the gains it provides outweigh the increase in materials cost remains unclear.

Whichever way we end up going, one thing is certain. We cannot delay our development schedule in order to make this decision. We need to decide on a strategy quickly and GO!

As early as 1991, in order to improve their cost position and speed up time to market, the printer divisions in Boise had implemented two new product development metrics. First, they had instituted cost reduction goals for each new generation of printer. The costs captured in this measure included labor, material, and manufacturing overhead. The second metric, called break-even time (BET), had been man- dated by upper management. It measured the time from project initiation to break even, defined as the point where total discounted cash outflow equaled total discounted cash inflow.

The Finance Perspective Neatly arranged on David Hooper’s desk were the lat- est sets of pro-forma income statements and balance sheets for the new project. Pointing out the effect of the universal power supply on income, he noted,

If we incorporate the universal power supply and sell 450,000 units of Rainbow, it will cost us approximately $13.5 million in additional material costs. If we are not able to pass this increase along to the customer, or at least our retailers, that comes straight out of our bottom line.

I sure agree that there will be benefits from universal power supply. May be we should take a hard look at the costs of stockouts and inventory.

Demand fluctuates during each of the three life-cycle periods and so do the costs of making or missing a sale. We typically estimate that for each lost sale we actually forgo multiple times our profit margin. The reason for this is that if a customer buys a competitive brand due to our inability to keep the resellers on stock, there is a chance that he will stay with that brand when he purchases a printer in the future. This effect might cover three or four generations of printers. Moreover, we may lose the profits from the sales of consumables such as toner cartridges and perhaps even other HP peripheral products.

The cost of stockouts when the product is first intro- duced into the market is even higher, as the potential word-of-mouth and publicity effects can damage the future sales and ultimate success of the product. On the other hand, the cost of stockouts at the end-of-life stage is probably considerably lower, as there is less fear of adverse effect on future sales, and the resellers might in some cases steer the customer to wait for the new, incom- ing, replacement product.

Although the cost of stockouts in the ramp-up stage is the highest, it is also this stage when we know the least about the market response to our new product, and our forecast errors are usually much greater. I understand that Sara’s material planning people had done some homework and found that the standard deviation of our

sim8239X_ch11_331-364.qxd 5/9/07 12:01 PM Page 361

362 DESIGNING AND MANAGING THE SUPPLY CHAIN

monthly forecast error (a new measure of forecast accu- racy that the group has started to measure) was close to 40 percent of the average monthly demand in both mar- kets in the mature and end-of-life stages. Their percep- tion is that the corresponding percentage is 80–90 percent in the ramp-up stage.

The other major cost that I have to monitor is inven- tory. My financial analysts have estimated that our annual holding cost rate is approximately 30 percent, which covers warehousing, insurance, cost of capital, and shrinkage.

The Manufacturing Perspective Sara Donohoe, manufacturing engineering manager, commented,

I think the universal power supply is a great idea. This innovation will improve our flexibility to respond to orders in two key ways. The first is the obvious gain of delaying the regional allocation decision by two and a half months. I’m sure marketing has expounded on this ad infinitum. The second gain is more subtle. You see, while transship- ment has always been possible in theory, we have avoided it whenever possible. Let me explain.

At the ramp-up stage, we always try to stockpile our DC’s with loads of printers so that we don’t ever run out of stock, and, given the high cost of shortage at this stage, this seems reasonable. There is not much of a need for transshipment. In the mature phase, if we keep doing what everyone at HP does and keep enough safety stock to meet the standard service target of around 98 percent then again, the chance of our needing transshipment is still small. However, I am not sure if we want to keep having 98 percent service goals at the end-of-life stage, and indeed that is when transshipments will be most needed.

The whole idea behind transshipment is to adjust inventories in response to market demand. To do this effectively, you need to move the product quickly. Unfortunately, to send a printer by air across the Atlantic costs us $75. Sea shipment reduces costs significantly to approximately $15/unit, but a month out on the ocean does not do much for responsiveness, which is exactly what you’re trying to achieve! In addition to the transportation cost, we know how tedious and complex it is to reconfig- ure the power supply and fuser. I would put my conserva- tive estimate of the activity-based cost for reconfiguration to be at least $250 per printer.

As you might imagine, the quality people go nuts when they find out we’re doing this. How can you establish a controlled process if you only do something once a year? Even worse, since the rework involves electrical compo- nents, safety standards require the reconfiguration process to be certified by Underwriter’s Laboratory. If you’ve ever

dealt with UL, you’ll realize how much trouble you’d have getting a process like this approved.

The universal power supply would allow us to avoid this mess, making transshipment a distinct possibility. The cost of reconfiguration is almost zero. It is at least a possi- bility, although I’m not sure who would coordinate it or decide when to ship . . . our friends in distribution, I guess.

My only real concern with developing the universal supply is the potential power play that could emerge at the time of allocation of the production build to the two regions. Again, I would like some visibility and control over how many units I can count on receiving.

The Distribution Perspective Rob Seigel runs the North America distribution cen- ter. Rob worked in a variety of positions before he moved to management and his present position.

Given a universal product, transshipment won’t present a big problem for the DC; it’s just another shipment to us and we can easily “localize” the product by adding manuals and plug adapters at the DC. Personally, however, I feel like it’s a great way to chew up company profits. I can just see us send- ing 1,000 units to Germany in February only to have them ship another 1,000 back to me in March. Both actions may seem to make sense at the time the decision was made, but in the end the company’s out hundreds of thousands of dollars!

Who is going to make the decision to shift inventory from DC to DC? I can see a real firefight if one DC wants more but the other is unwilling to give up its excess. We all have pressures for high customer service and even if I have some excesses now, that does not mean that I may not need it next month. Sending the product to Europe helps their performance, but what about mine? I hope I do not have to do it! One thing I don’t have time for is spending half of my day on the phone to Germany trying to negotiate a transfer.

I guess, though, if we can avoid what happened with the VIPER we have got to be better off. That was an interesting time. See that warehouse, pretty big. It was so full we stopped leaving the aisles clear and just stacked printers solid, from floor to ceiling, all the way from front to back. I would pay money to prevent that from happening again. All other work grinds to a halt when a crisis like that emerges.

THE DECISION

The team had decision-making authority, but they would have to defend their decision to upper man- agement. From past experience, they knew that if they decided to adopt the universal power supply, management would want to ensure they had performed adequate analyses of all the costs and benefits of such

sim8239X_ch11_331-364.qxd 5/9/07 12:01 PM Page 362

CHAPTER 11: COORDINATED PRODUCT AND SUPPLY CHAIN DESIGN 363

a decision, as well as some estimate of the risks involved. In addition, some consideration of how the decision would impact future generations of products will have to be made.

CASE DISCUSSION QUESTIONS

1. In what way is a universal power supply a post- ponement strategy?

2. What are the costs and benefits of a universal power supply (feel free to make assumptions)?

3. How would such costs and benefits be different over the product life cycle?

4. Besides deciding on a universal power supply, what other operational improvements can you suggest to HP Boise?

5. What would be your recommendations about the adoption of a universal power supply?

sim8239X_ch11_331-364.qxd 5/9/07 12:01 PM Page 363

sim8239X_ch11_331-364.qxd 5/9/07 12:01 PM Page 364