Printer Friendly

How SUPPLY CHAIN ANALYSIS Enhances Product Design.

Product design can have a huge impact on supply chain costs. The most effective way to gauge those costs -- while at the same time identifying opportunities for revenue enhancement -- is through supply chain analysis during the product development process. Described here are the analytical tools and techniques that Hewlett-Packard used to enhance the design and rollout of a key product.

Striking the right balance between breadth of product offering and cost constraints is a challenge for most businesses. Hewlett-Packard Company (HP) is no exception. Founded in 1939 by William Hewlett and David Packard, HP employs 88,500 people and generates more than $49 billion a year in revenues from 25,000-plus different products and services.

The supply chain analysis described in this article helped to strike that balance. The effort focused on optimizing product fan-out in HP's CD-RW (Compact Disc Re-Writeable) business. (Fan-out is the process whereby a generic product becomes a finished item for customer purchase.) The objective was to enhance the overall profitability of the CD-RW business while creating an accessible analytical tool that business managers could use in their daily decision making.

Before examining the details of how HP applied supply chain analysis to product fan-out, it's important to understand the key phases and organizational dynamics of product development. It's also important to know the severe cost and lifecycle constraints associated with inventory in the high-tech sector.

Phases of Product Development

Supply chain analysis can play a critical role in the three key phases of product development--investigation, design, and manufacturing or rollout.

The investigation phase begins when a marketing group sees an opportunity. For example: "We think we can sell 100,000 widgets with this feature set for $110 each." Within a few months, the marketers and engineers meet to present their conclusions on the product's feasibility. These conclusions include the price at which they think they can sell the product, how much they think it will cost to produce (raw material cost) and distribute, and how long they think it will take to develop (time to market).

If the conclusions from the investigation are favorable, the design phase begins. During this phase, each functional group does the detail work to assess exactly how the product will be designed, manufactured; distributed, and marketed in a way that meets the system cost and time-to-market goals defined during the investigation phase. Research and development (R&D) engineers must design the product to meet the raw material cost goals with the required functionality; manufacturing must design the appropriate system to manufacture and distribute the product; and marketing must devise a marketing plan to promote and sell the product. Most of the supply chain analysis should take place in the design phase because this is where the greatest efficiencies can be gained.

During the manufacturing phase, the product is manufactured and sold, and the company reaps the benefits of (or pays the price for) decisions made during the design phase. It is in this phase that the supply chain decisions come to fruition.

The Organizational Metrics

For supply chain analysis to be effective, it must address the metrics used by each functional group to monitor its own performance and drive improvement for teams and individuals. These metrics typically have a functional focus; in other words, people are measured on things that are directly under their control. Each functional organization has its own set of metrics, which tend to encourage certain behaviors or methods of achieving the accompanying goals.

* For marketing, these metrics are revenue growth and market share. Marketing teams promote their goals by shortening the time to market, diversifying product offerings, being as responsive as possible to changes in customer demand, and ensuring that all versions of the product are available to fulfill potential sales.

* For R&D, the metrics are quality, development time, and material cost. R&D's method of achieving this goal is to create the best possible product at the lowest material cost.

* For manufacturing/distribution, the metrics are supply reliability and low manufacturing costs. To promote this goal, it is in manufacturing's interest to have fewer products and maintain steady volumes of production.

For the most part, the typical metrics are good ones. Yet supply chain analysis can add value beyond these metrics in at least two ways. First, it can mitigate inherent conflicts of interest between the functional groups to find the best overall solution. Second, supply chain analysis can identify opportunities such as overall profitability that may be masked by the traditional metrics.

Mitigating Conflicts of Interest

Throughout product design, a natural tension exists between the three functional teams. The reason is that their respective objectives and measurements are sometimes at cross-purposes with each other. Decisions made to improve the product along one team's dimension can adversely affect performance along a vector important to another team. For example, marketing's focus on revenue growth motivates this team to expand product functionality and product variety--including adding more SKUs (stock-keeping units)--to broaden the attractiveness of the product. In contrast, manufacturing and distribution want to constrain product variety to enable the supply chain to be as cost-effective and responsive as possible. (Exhibit 1 depicts the drivers for each functional area.)


Effective product design analysis should consider all of these areas and make the appropriate trade-offs. Because it measures the impacts of design decisions on the overall system, supply chain analysis can arbitrate between these conflicting interests during the design phase.

By providing thorough, rigorous, data-driven quantitative analysis of costs and benefits, supply chain analysis can help in the negotiations between the functional areas. This enables explicit trade-offs to be made that will improve the product's overall value proposition. By using supply chain analysis, the functional teams can more clearly recognize that the greater goal is a product that enhances the company's top and bottom lines (revenue and profit).

Identifying New Opportunities

Although the metrics may be consistent from one product to the next, the goals for each functional group are usually set individually for each product. These goals typically relate to specific dollar amounts, calendar dates, or some other clearly quantifiable objective.

The R&D team's goals are set when a product moves from the investigation phase into the later stages of development. These goals would include an introduction date for the product (development time) and a specific material cost. During the design process, R&D engineers work closely with procurement engineers to design components that can be fabricated at the projected sales volumes to meet the material cost targets. Note that the primary cost goals for R&D are direct material costs and rarely overall system costs, over which this function may have little control.

The marketing team is responsible for driving sales. Accordingly, its metrics are revenue and market share, and its specific goals center on projected sales volumes. Marketing teams are usually measured and rewarded on a quota basis; the overriding goal is to meet the year-over-year financial growth or market share goals. The more product moved, the more marketing is rewarded. As with R&D, this organization is rarely held accountable for system-related costs or the resulting product profitability because most of those cost structures are beyond marketing's control.

The manufacturing/distribution team is responsible for the systems that build and deliver the product. Through the design phase, this team provides input to R&D to encourage product designs that facilitate a cost-effective manufacturing and distribution system. The goal of the manufacturing/distribution system is to provide product reliably to the customer base as cost-effectively as possible. The goals for this team may include low manufacturing and distribution costs, sufficient inventory, high product quality, and delivery reliability.

Although all of these goals are important, they are function specific. Further, they tend to focus on specific activity costs rather than on systemwide costs or on profitability. Supply chain analysis can help companies look beyond these functional metrics to identify opportunities for reducing costs and generating revenues that will benefit the organization as a whole.

The Cost of Inventory in High-Tech Markets

To fully appreciate the role that supply chain analysis can play in an HP product fan-out, you need to understand the nature of inventory in high, tech consumer products. Products in this sector often are characterized by a high degree of product innovation, short product life cycles, strong price competition, high expectations from customers regarding availability, and steep production ramps. It is not uncommon to see products whose sales go from zero to 200,000 in a month and back to zero a few months later.

This combination makes for a challenging business. Because of the high volumes required and the short product life, the supply chain structure and supporting processes must be complete before the product is officially introduced to the market. A sudden change in market conditions, such as a competitive product introduction, a component price drop, or an economic downturn, can result in significant financial exposure through devaluation of excess system inventory. Regardless of who actually holds the inventory at the time, the cost most often is born by the manufacturer--that is, HP for our products. Within the highly uncertain world of electronic products, inventory is a major system cost. In fact, it often represents the most critical category in deciding on a particular supply chain structure. Effective supply chain management can help companies manage inventory levels to achieve maximum sell-through with minimum excess.

A Case Example: CD-RW Product Fan-Out

HP's CD-RW drives are a good example of a high-tech consumer product with the characteristics stated above. Although HP is a market leader in CD-RW products, it faces fierce competition from other large competitors with established brand names. Competition in the market focuses on pricing and on development of new technologies, which must be brought to market as quickly as possible. Today's hot new products become yesterday's news every nine months.

Because the profit margin is low and the product life is so short, inventory management is critical. These are consumer products; availability is the key to sales and to maintaining good relationships with retailers. The retailers want higher levels of inventory to make sure that they always have units on hand to sell. HP, for its part, wants to maintain low levels of inventory, just enough to meet demand. Getting the right units to the right place at the right time can make the difference between profit and loss.

Short product life cycles mean that excess inventory must be discounted when competitors release new versions. However, the low margin also means that any significant price reduction will result in selling the product for less than the cost of its parts. Complicating matters, the core components have long leadtimes. Requirements for high availability, coupled with long leadtimes and retail sales, translate to large amounts of inventory. At the same time, inventory devalues rapidly and profit margins are low. Even a little excess inventory makes the business unprofitable. All in all, this makes for a tough proposition!

To manage the business successfully in this kind of environment, our managers focused on a number of supply chain initiatives. These included inventory optimization, collaboration with suppliers, product postponement, and joint planning with the retailers. In addition, the supply chain manager for the CD-RW business put together a project team to determine if the product fan-out could be better managed--and, if so, what the impact would be on business profitability. The group included supply chain engineers as well as members of HP's Strategic Planning and Modeling (SPaM) team. (For more on SPaM's activities, see the "An Overview of SPaM's Activities".)

The Fan-Out Process

Product fan-out describes the process by which a generic product (in this case, a CD-RW drive) becomes a finished product for purchase by a customer. To make a finished product, manufacturing inserts the base mechanism into a case, adds a power supply (110V or 220V), adds I/O cables, boxes the unit, and inserts localized manuals and software. When we started working with the CD-RW business, the generic product "fanned out" to 20 SKUs. The SKUs were differentiated by power supply, language (such as French, German, or English), and special promotions.

Though we often forecast the number of core products needed from the suppliers with fair accuracy, we were less accurate forecasting the number of each SKU we could sell. As a result, we ended up having enough core mechanisms, but our mix was often wrong for the end products. This is to be expected because it is easier to forecast demand for one worldwide product than to forecast product sales in 18 different countries. Forecasting one generic product realizes benefits from risk pooling, as lower-than-expected sales in one country are offset by higher-than-expected sales in another.

Product postponement helped us manage the mix problem by delaying product differentiation to as late in the process as possible. However, because inventory was held in warehouses close to the retail channel (where the product was in final form), reducing product fan-out could also prove beneficial. If we made one worldwide SKU instead of 20 local SKUs, we could move excess inventory to where it was demanded. Furthermore, we could forecast the one SKU's demand with greater accuracy.

At the same time that the supply chain manager was investigating SKU reduction, marketing was putting plans in place to quadruple SKU count. Our marketing colleagues felt that by increasing SKU count they could sell more units. Additional SKUs allow the creation of products that meet specific customer needs better and therefore,can be differentiated from the competition. However, adding SKUs also increases costs and increases the risk of excess inventory for one or more models. When costs increase, expectations for volume also increase to cover the costs, and marketing sometimes asks for more SKUs to allow it to meet the new goals. This can create a reinforcing loop as shown below.(1)


Promising Initial Results

The Strategic Planning and Modeling team agreed to help the marketing and manufacturing/distribution managers complete a quantitative analysis on the relationship between SKU count and inventory for the current CD-RW product. Because there could be design implications in our analysis (for example, we might recommend a product with a universal power supply), we reviewed our plan with the R&D team as well. Managers from all three functions asked the SPaM team not only to provide analysis and recommendations but also to train their staffs on the process.

We began by investigating the cost of inventory in the system. Often, inventory costs are hidden and therefore not considered correctly. We wanted to make these costs as explicit as profit or sales revenue. We looked at historical financial reports and measured the costs from the following sources: devaluation, price protection, opportunity cost of capital tied up in inventory, warehousing, and shrinkage. We turned this number into a percentage per unit and created an annual "holding cost" percentage that could be applied to the product cost.

For example, consider a $200 product with a 35-percent annual holding cost of inventory. If we make the product in January of 2000 and don't sell it until January 2001, the holding cost for that unit of inventory is $70. To turn this into a simple measure for inventory, we created Equation 1 shown below. It calculates the annual cost per unit of holding one week of inventory supply. A week of supply is equal to one week's worth of demand.

Equation 1

Inventory cost in $/unit/WOS

[I.sub.$/unit/WOS] = [h x c] / 52


h = annual holding cost %

c = material cost

I = cost per unit per week of supply (WOS)

For our CD-RW product example, reducing inventory by one week of supply saves $1.35/unit (all else being equal).(2) This is the same as reducing material cost by $1.35 -- quite significant for a high-volume product with low margins.

Armed with this measure, we set out to calculate the reductions in inventory expected in the current product from reducing SKU count. We started by looking at the current levels of inventory by modeling two types of inventory at each stocking location: cycle stock and safety stock. This is an established technique; first described in Tom Davis's article "Effective Supply Chain Management" in Sloan Management Review.(3) Our team has applied the approach with great success over the past decade.(4)

Cycle stock accounts for the inventory required assuming no uncertainty and is determined by the delivery frequency to the stocking location. If deliveries are received once per week, on average there will be half a week of cycle stock at the receiving location. Similarly, if deliveries are received once per month on average, there will be half a month of cycle stock on hand.

The second type of inventory, safety stock, exists to buffer against uncertainty of supply and demand. The reduction in SKUs would pool demand uncertainty because product demands are easier to forecast with less mix. We reviewed the ordering patterns for each SKU, measuring the mean demand and demand variability. We used this information to lead a discussion with our marketing colleagues on how the SKUs could be consolidated. For example, U.S., Canadian, and Mexican products could become one North American product. We calculated the consolidated mean demands and reduced demand variabilities for these hypothetical "combined SKUs."

Because we knew the current inventory levels and delivery frequencies, we could simply subtract the cycle stock from the current inventory ]eve] to determine the safety stock currently kept on hand. That amount might not be the optimal level of safety stock, but it was historically what was required to run the business. All other aspects of the system being constant (delivery frequencies, target service levels, and so forth), we could use this technique to predict the reduction of inventory resulting from changes in product fan-out. By using this approach across regions, we were able to predict the reductions in inventory associated with different SKU counts. We translated this inventory reduction into a per-unit cost differential and created the graph in Exhibit 2. The blue line in the exhibit indicates the cost increase or reduction from inventory associated with different SKU counts.


The Beginning of the "Rough-Cut" Method

Although the ability to predict reductions in inventory was exciting, it only met one of our deliverables: that of quantifying the effect of SKU count on inventory holding costs. We still needed to create a method whereby members of the business team could perform the analysis themselves. Although this technique would not replace exhaustive supply chain analysis, it would allow the people making day-to-day decisions to identify which areas offered the greatest promise for improvement. By examining the behavior of variability and quantifying the reduction in safety stock across different scenarios, we were able to develop simplified techniques to approximate the benefit from SKU reduction. In effect, we tried to duplicate our more exhaustive analysis with a simple approach that we called the "rough-cut" method.

In conducting a rough-cut analysis, we make some simplifying assumptions. We assume that our suppliers always deliver on time and that demand for each SKU is identical and uncorrelated. This means that individual SKUs have the same volume and the same profile of demand uncertainty and that demand for one SKU is unaffected by the demand patterns of other SKUs. These assumptions mean that rough-cut techniques can predict more savings from SKU reduction than do detailed product-by-product analyses.

Suppose we have one worldwide 220V SKU that is sold everywhere except Portugal, where we have a different SKU of which we sell very few units. We decide to stop selling the Portuguese SKU and to modify the worldwide 220V SKU slightly so that we can sell it to Portuguese customers. Will we see reduced demand variability for the new 220V unit that we also sell in Portugal? Yes, but very little. This reduction depends on the demand variability in Portugal and the ability to combine that variability with the larger 220V SKU.

We examined the changes in safety stock associated with different risk-pooling techniques: creating a universal SKU, spreading the volume, and consolidating the volume. The figure in Exhibit 3 illustrates these different methods and outlines the rough-cut calculations to determine the reduction in safety stock expected with each alternative.


The following example illustrates the "universal SKU" technique. If the business currently has six weeks of supply of finished product and receives deliveries once per week, it has half a week of cycle stock and 5.5 weeks of safety stock. Suppose there are 10 SKUs and management wants to determine the benefit of consolidating these into one worldwide SKU. The reduction rate (RR) to apply to historical safety stock is 1 - 1/sq. root (10), or 68 percent. This predicts that the business can be run with 1.8 weeks of "universal SKU" safety stock instead of 5.5 weeks of "local SKU" safety stock, while providing the same level of product availability to our retail channels and our customers. A figure of $1.35/unit/WOS (weeks of supply) means we save $5/unit. Therefore, we should be willing to spend up to $5/unit to create a worldwide SKU, as long as we do not expect other things in the system to change (such as overall sales or delivery frequency).

This rough-cut method is not as accurate as a more robust, detailed analysis. However, the estimates are directionally correct and save significant time in analysis. Although the proposed SKU consolidation will never save $10/unit, it might save as much as $5/unit and would probably save more than $4/unit. A savings of $4/unit might not be exciting for some companies. But for a business where margins are low and volume is high, $4/unit can affect profitability significantly.

Extending the "Rough-Cut" Tool

We can extend the rough-cut approach to other product design decisions that affect required system inventory levels. In particular, we have used this technique in two applications: (1) determining the trade-off between material cost and leadtime and (2) assessing the value of an increase in supplier delivery frequency.

Trade-Off of Material Cost vs. Leadtime

This application is extremely useful because it enables R&D and manufacturing/distribution engineers to jointly determine the overall system cost implications of choosing a low-cost supplier over another who is more responsive. Because the R&D team is measured on direct material costs, the R&D engineer normally would default to the cheaper supplier. The creation of this rough-cut tool enables the two engineers to quickly identify the best decision for the company as a whole, incorporating both the direct material costs and the inventory holding costs. The manufacturing/distribution team can also use this technique to determine the value of reducing leadtime with an existing supplier.

For a particular component, suppose that a business is deciding between a new supplier and an existing one. The new supplier is offering a much shorter leadtime with a small increase in price, as follows:

Existing supplier: Core drive cost = ~$100, leadtime = 8 weeks

New supplier: Core drive cost = ~$102, leadtime = 2 weeks

Should the business switch suppliers to minimize its overall system cost? This analysis identifies the reduction rate in the required safety stock and then compares the inventory savings against the increase in material cost. (See Equation 2.) Using the earlier example of 5.5 weeks of safety stock and assuming a review period of one week, the safety stock reduction resulting from the supplier switch equates to a reduction rate (RR) = 1 - sq. root (3/9), or 42 percent. This predicts that the business can be run with approximately three weeks of safety stock, which translates into more than $3/unit inventory cost savings.(5) Given that the material cost increase is only $2/unit, the business should move to the new supplier, assuming that the switching costs are not significant.

Equation 2

Safety stock reduction rate due to a leadtime reduction



RR = reduction ratio for safety stock (SS)

[L.sup.old] = old leadtime

[] = new leadtime

R = review period

The above equation assumes deterministic leadtimes. This analysis assumes no sustained forecast bias, calculating only the benefit of intra-platform mix changes due to leadtime reduction.

Value of Increased Delivery Frequency

The second extended application of the rough-cut technique is assessing the value of increased delivery frequency. Supplier delivery frequency is often overlooked in the product design process. However, it can have a significant impact on the system cost through the resulting cycle stock if a supplier delivers infrequently. In addition to assessing the impact on inventory, any analysis of delivery frequency must also include freight. Freight rates vary as a function of the average volume/weight being shipped. Thus, a change in the delivery frequency is going to change the freight cost per unit, assuming all other things remain constant.

The rough-cut approach enables the design and manufacturing engineers to determine how a particular product design decision will affect freight considerations such as the packing density of the part or product. With the design change and the material cost to make that change both identified, the team can quickly rough out the other cost elements of freight and inventory and decide which configuration provides the lowest system cost. The technique can also be used to compare two different suppliers where one offers a higher delivery frequency than the other.

Making Better Design Decisions

The SPaM team delivered our rough-cut analysis through a series of workshops. It will come as no surprise that our colleagues in R&D and technical marketing were not afraid of the math. Along with their colleagues in manufacturing/distribution, they embraced the new approaches and were delighted to have an analytical tool that would add speed and confidence to their decision making. They decided to reduce the SKU count for new products, rather than to go ahead with plans to quadruple the number. Management estimates that this change increased profitability of the new product by 16 percent.

These rough-cut approaches now reside on our internal Web site as a "product design for supply chain" calculator. Designers across HP visit the Web site and perform a few calculations to test the inventory-driven costs associated with different product design decisions. If their own analysis is inconclusive, they are encouraged to contact our team for assistance.

As this article shows, product designs have a big impact on supply chain costs. In our experience, supply chain engineers and managers are in an excellent position to serve as arbitrators between marketing and R&D to help make better overall design decisions. Creating a clear, visible measure of the impact of inventory on business success is a key enabler to this approach. Exhaustive analysis can be done to support these decisions, but often the organization cannot wait for the analysis to be performed, nor does it have the data necessary to support rigorous review of the alternatives. Simple, rough-cut analysis is often better--speed is more important than detail. In this case, rough-cut analysis allows management to make important decisions quickly and creates substantial cost savings while maintaining high levels of customer service and customer satisfaction.

But whether formal or rough-cut, supply chain analysis needs to play a prominent role in any product fan-out initiative.


(1) Presented at Hewlett-Packard by Prof. Dr. Ulrich W. Thonemann of Universitat Munster.

(2) This example is for illustrative purposes only. It does not reflect actual material cost or inventory holding costs for HP products.

(3) Tom Davis, "Effective Supply Chain Management," Sloan Management Review, Summer 1993.

(4) Brian Cargille, Steve Kakouros, and Robert Hall, "Part Tool, Part Process; Inventory Optimization at Hewlett-Packard," ORMS Today, October 1999.

(5) This calculation was done using an inventory holding cost of $1.35/unit/WOS.

Authors' note: The authors wish to thank Jason Amaral and Professor Hau Lee of Stanford University for their invaluable contributions to the development of these rough-cut analysis techniques. We also acknowledge Tom Phelps for his implementation of the methods at HP's CD ReWriteable business.

RELATED ARTICLE: An Overview of SPaM's Activities

HP's Strategic Planning and Modeling (SPaM) team was formed more than 10 years ago to develop practical supply chain management solutions and disseminate them broadly across HP's businesses. We have completed many analyses for HP businesses, addressing such strategic supply chain questions as "Where should we put the next factory?" and "Should we manufacture regionally or at one worldwide location?" Our models incorporate all major relevant supply chain costs -- freight, duties, production and distribution costs, and an array of inventory-driven costs. These models allow management to choose among a set of alternative manufacturing and distribution strategies. Through our experiences in modeling alternative supply chains, we were exposed to the impact of design decisions on inventory-driven costs as well as the relative importance of inventory costs within the system.

To help HPX manage its systems most cost effectively within this environment, our team has led several initiatives that underscore the value of designing products to facilitate the most inventory-efficient supply chains. Examples of these efforts include the following:

* Work with the DeskJet printer division in Vancouver, Wash., identified savings in implementing late point differentiation of localized material (manuals, power cables, and so forth) for our DeskJet printers. This initial effort yielded hundreds of millions of dollars in reduced inventory exposure while enabling HP to provide a higher level of product availability to our customers. Since that experience, design for postponement has become an accepted practice within the company.(1)

* Projects with the Laser Jet printer division in Boise, Idaho, identified the potential savings of using a universal power supply over the existing design, which incorporated voltage-specific ones (110V and 220V). This effort yielded significant inventory-cost-related savings for HP.(2)


(1) Hau L. Lee, Corey Billington, and Brent Carter, "Hewlett-Packard Gains Control of Inventory and Service through Design for Localization," Interfaces, July-August 1993.

(2) Hau L. Lee, and M. Sasser, "Product Universality and Design for Supply Chain Management," Production Planning & Control, Vol. 6, No. 3, 1995.

Brian Cargille and Robert Bliss are process technology managers on Hewlett-Packard's Strategic Planning and Modeling (SPaM) team.
COPYRIGHT 2001 Peerless Media, LLC
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2001 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Title Annotation:Hewlett-Packard Co.
Author:Cargille, Brian; Bliss, Robert
Publication:Supply Chain Management Review
Article Type:Company Profile
Geographic Code:1USA
Date:Sep 1, 2001
Next Article:Integrating Demand and Supply Chain Management.

Terms of use | Privacy policy | Copyright © 2019 Farlex, Inc. | Feedback | For webmasters