What Is an Optimal Distribution Network Strategy?

By Tom Tiede and Kay Ree Lee -- Supply Chain Management Review, 11/1/2005

The questions "How many distribution centers should we have?" and "Where should they be located?" reflect one of the most complex challenges that businesses face today: choosing a distribution network design. That challenge is intensified when the company is confronted with pressures to reduce costs while increasing service levels to customers. Oftentimes it falls to the organization's supply chain managers to meet this challenge. So they diligently march forward, seeking to optimally balance service and costs.

Supply chain managers need to consider a number of factors over and above the basic cost-vs.-service tradeoff analysis before they can complete the final network design. Rarely is an effective network strategy derived from a mathematical equation alone. Instead, traditional optimization analysis is only the beginning of a series of compromise decisions that must be made to ultimately determine the "optimal" distribution network strategy for an organization. This article presents a short list of traditional and other decision variables that corporations should address in designing a flexible, cost-effective, and implementable distribution network strategy.

Generally, there are two basic reasons to redesign a distribution network: (1) to realign the network to support shifts in the business and (2) to move the "supply chain exchange curve." The supply chain exchange curve illustrates the tradeoff between service and supply chain costs. To improve service, a company typically will need to increase costs. Examples include expediting freight, adding personnel, and increasing inventory levels. When redesigning a distribution network, the objective is not to move along the exchange curve. Instead, the goal is to actually move the curve, thereby enabling higher levels of service at a lower overall cost to the business. Exhibit 1, on page 34, depicts the two main reasons for realignment, listing the typical business shifts that can occur in the first case and the business objectives that drive the need to move the supply chain exchange curve in the second.

Shifts continually occur in any business. Volumes grow or erode, customers come and go, and service requirements change. A distribution network that was optimal just a few years ago can become inadequate in meeting the current and future needs of the business. Consequently, most companies have a genuine need to periodically re-evaluate their network strategy. In doing so, the main goal is to find an optimal balance between costs and service. But the analysis also needs to include a host of other decision variables such as taxes, growth or acquisition plans, local labor conditions, and more. Careful evaluation of all the factors will result in a new strategy that shifts the supply chain exchange curve, reducing costs to the business while also improving service to customers.

Accurately identifying all costs is important for forming a solid baseline for comparing alternative network scenarios. But identifying costs is by no means an end in itself. The final design will be derived as much through art as through science. The process will involve evaluating a range of alternatives and reaching consensus among stakeholders on a number of tradeoff decisions, both quantitative and qualitative in nature. Leaders responsible for sales, customer service, inventory management, manufacturing, and distribution must be involved in this decision-making process. In any case, building an effective network strategy begins with a thorough understanding and evaluation of costs.

The primary cost components traditionally considered in a network analysis relate to facility, personnel, transportation, inventory, and system costs. Ongoing costs and one-time investments associated with each of these components should be evaluated when comparing the business case for any alternative network scenario. Typical ongoing and one-time expenditures are shown by category in Exhibit 2.

Generally, there is a tradeoff between providing exceptional service to customers and the costs associated with that level of service. However, improving service does not always mean increasing costs. For example, opening an additional distribution center to upgrade service in a particular region of the country may yield freight, tax, or other savings that offset or surpass the additional cost for warehousing.

Service requirements can be met by an array of alternative network designs. The illustrative matrix shown in Exhibit 3 is an effective way of visualizing the cost-service tradeoff associated with different what-if scenarios. The graphic depicts the sum of costs and taxes for each alternative vs. the service level provided by the overall network. The matrix, in effect, puts a "stake in the ground" from which further analysis can be performed and decisions made. So in the example shown in the exhibit, the "two-facility" design alternatives can be eliminated from future consideration if high service levels are a priority.

State tax implications are often overlooked in the distribution network equation. Unfortunately, this oversight can easily lead to a suboptimal network. Tax laws vary greatly from state to state. Ignoring the differences when deciding where to locate facilities and what customers to service from those facilities can needlessly add millions of dollars to a corporation's annual tax burden. In fact in some high-margin businesses, it is not unusual for a high state tax burden to exceed annual logistics costs.

For corporations that ship products from one state to another, income must be allocated to the state to which to the product is shipped—the destination rule. In some cases, the destination state does not levy an income tax because that state does not have a corporate income tax or the corporation does not have a sufficient level of payroll or assets in that state (also known as "nexus"). This results in "nowhere income"—sales that are not taxable in destination states. This does not necessarily mean that the income on these sales goes untaxed. Many states have a throwback rule that results in "nowhere income" being passed back to the state of origin for tax purposes.

Even more burdensome is a unitary reporting requirement imposed in a few states. It requires corporations to combine all related business income for purposes of apportioning taxable profit back to the origin state. Unitary reporting specifically targets corporate tax-avoidance strategies set up to artificially shift costs and profits from state to state.

From a tax standpoint, the ideal distribution network would avoid the throwback rule, unitary tax, and any state income tax obligations whatsoever. In the real world, however, this scenario is rarely feasible. And while tax considerations certainly merit careful attention, they should not be the sole criterion for facility location decisions. As with basing a decision solely on cost, this approach would lead to suboptimization of the distribution network. A more balanced approach is required, one that weighs service requirements against a complete picture of total cost to serve—supply chain costs, sales and administrative costs, and taxes. Exhibit 4 shows the difference between suboptimal strategies and an optimal distribution network strategy.

Unfortunately, the traditional approach to analyzing network alternatives rarely considers taxation. Instead companies rely on complex computer models to find the lowest combination of freight and warehousing cost to serve customers within a predetermined mileage radius. This can be a worthwhile exercise, providing keen insight into the relationship between delivery lead times, alternative location scenarios, and logistics costs. And often, this exercise can help companies identify considerable potential savings through geographic realignment of facilities, customer segmentation of delivery service requirements, alterations to the inventory deployment strategy, changes to the transportation mode mix, and so forth. However, the total savings potential will likely be suboptimized if taxes are not considered.

Consider the following when designing a tax-efficient distribution network:

• In what states do you have a nexus?
• Can you effectively move payroll and assets elsewhere to change nexus status?
• In what states are you the most vulnerable to taxation?
• What are the service requirements to customers in these states? Distribution from what other states could service these requirements?
• What are the tax laws in each of the states under consideration?
• What legislation is being considered in these states that may alter current tax laws?
• Can you shift distribution of higher-profit products to a facility in a lower-tax state?
• Does the benefit of relocating to a less tax-burdensome state outweigh the associated costs and impact on service?

The following hypothetical example speaks to that last point. ABC Corp. has three distribution centers (DCs) located in California, Georgia, and Illinois. These are the only states in which ABC has a nexus. All of the distribution centers ship products to other states around the country. In addition, all three states in which the DCs are located have a corporate income tax, and California and Illinois both employ the throwback rule. What would happen if ABC decided to close the distribution facilities in Illinois and California and open a facility in Nevada, which does not have a corporate income tax? Exhibit 5, on page 36, shows the current distribution network on the left and the alternative network on the right.

In the alternative, two-facility network, demand fulfilled from the DCs in Georgia and Nevada will lower the overall tax burden for ABC Corp. Throwback rules on nowhere income are avoided under this option. In addition, the Nevada facility avoids all corporate income tax on product distribution. But is this the optimal strategy? It may be. But again, the network decision cannot be based on any single factor like taxation alone. Other factors, such as the impact on service and the other decision variables enumerated below, need to be considered as well.

In addition to costs, service levels, and taxes, other, less quantifiable decision variables play an equally important role in designing a distribution network strategy. The following are among the most important.

Growth Plans. Are you designing for the past or the future? And if it's for the future (as it should be), how far into the future? Traditional network modeling is dependent on historical data—primarily sales-order history and freight history. After much effort to collect, cleanse, organize, and analyze historical data, a skilled modeler with a capable optimization tool can design the ideal network for yesterday's needs. But, how well does the past represent the future? The optimal network strategy must focus on future needs, not past. The network design must support planned growth, generally for the next five to seven years. In simple terms, distribution facilities must be positioned to accommodate new customers and new service requirements in each geographic market, and each distribution facility must have the flexibility to accommodate this growth. This may require locating facilities in areas other than what is suggested by the model and sizing them to accommodate increased storage and throughput volume through expansion or other means.

Acquisition Plans. Acquisitions are often a major part of an organization's growth strategy. Rarely does an acquisition fail to create a need to realign the distribution network. Redundant geographic coverage, undersized facilities, and potential transportation synergies are among the reasons for realignment. Further, the facility network of the acquired company may provide a better foundation for the combined distribution needs of the merged companies, essentially antiquating the acquiring company's existing network. Therefore, to limit their risk, companies need to carefully consider long-term financial commitments in buildings and equipment. With an uncertain future, flexibility is key. Shorter-term facility leases, contract warehousing, transferable equipment, and expansion rights are among the ways of creating flexibility and mitigating risk in the design and implementation of a distribution network.

Local Government Incentives. Distribution facilities are expensive—and it's not just the cost of the physical plant. Local property and payroll taxes, for example, can make up a sizeable proportion of year-end costs. Incentives on both a state and local level can easily sway the decision on where to locate a facility. In most cases, companies are well advised to engage a state agency responsible for business development and local Chambers of Commerce in the facility search. Alternatively, a knowledgeable broker can be a valuable resource and a worthwhile investment. This individual can save you time and effort in identifying facilities in areas with the most advantageous incentives.

Availability of Warehousing. Your network optimization model may suggest a facility location within a 100-mile radius of your major customers. Your best customer may suggest a facility location within a 10-minute drive. Both may be excellent suggestions. But if facilities are unavailable, if they do not meet your specific needs, or if they are simply too costly to build or lease within the "suggested" areas, you need to make a tradeoff between cost and service distance. This may mean choosing to service your customers from a less costly, existing warehouse in an area that may be good—though not necessarily ideal—from a network optimization standpoint.

Local Labor Pool. A great deal on a great building has little value without a good workforce to put in the facility. Stability, skill, dedication, and motivation form the foundation for distribution excellence. If you have difficultly finding people in the local workforce who possess these qualities, you run the risk of having a substandard, or at best mediocre, distribution operation. Without the solid foundation of a good workforce, operations will suffer and customer goodwill will erode over time. In the end, this will cost an organization far more in lost profit than they saved on a facility located in a market with an inadequate labor pool.

Access to Carriers. It's 5 pm, and you just received 10 new orders to six destinations that will fill three trucks. This is in addition to the 12 trucks you've already planned, and all of your orders need to be delivered tomorrow. It's a typical day, only each destination is different than yesterday's orders. The only way to get the job done is to rely on a pool of "go-to" carriers willing to take your loads on short notice. Or better yet, you need a dedicated carrier with the flexibility to take all of your shipments. Most network optimization models don't address carrier accessibility to this level of detail. Yet it's an important consideration for businesses with tight turnaround times between order cut-off and order delivery.

Proximity to Strategic Customers. The reality is that some customers are more important than others, and, as such, they need a higher degree of service. Often, this means locating a facility close to a customer for next-day or same-day service. Traditional optimization analysis has difficulty weighing customer importance when modeling a network. Human intelligence and judgment, rather than math, is generally the most effective means of aligning the distribution network to serve strategic customers.

Bandwidth for Change. A "green field" network optimization study can be very enlightening, often suggesting large savings from a network design that may be far different from the existing one. However, can the organization handle a change that is sweeping in nature? Implementing a network redesign can be a "bet the business" proposition and the bigger the change, the higher the bet. Success here requires exceptional executive leadership and a full complement of dedicated resources that will tightly control implementation. This is a rare combination. More often than not, compromises are made to control risk and manage resource constraints. Compromise, resulting in incremental changes over time rather than a complete one-time overhaul, may indeed be the more optimal strategy for an organization.

Executive Preference. The final variable in the network-design strategy resides within the organization's top executives. They most likely will consider all of the modeling results and subjective analyses presented to them. But in the end, the decision of what distribution network to adopt is often a "bet the business" decision. And this typically translates to a "bet the career" decision. Therefore, top executives tend to gravitate toward changes that offer lower overall risk to the business.

Designing an optimal distribution network strategy is a delicate combination of science, art, and compromise. There are numerous factors to consider in approaching the task. Traditionally, the process has focused on evaluating the tradeoff between total logistics cost and service-level capabilities among alternative network scenarios. As we discussed above, however, this limited focus is short-sighted. For one thing, it fails to take into account state and local taxes, which in some cases can easily exceed the logistics costs.

But even when taxes are part of the analysis, there is more involved in developing the optimal distribution network strategy. Some of these other less quantifiable factors that need to be considered include planned growth and acquisitions; availability of labor, warehousing, and carriers; and organizational appetite for change and risk. The optimal design will result from careful evaluation of these and other variables in the context of the organization's overall business strategy. We emphasized earlier that redesigning a distribution network can be a "bet the business" endeavor because of the critical customer linkage involved and the considerable investment required, both financially and in the time commitment of your most qualified individuals. It's neither trivial nor inexpensive, and it certainly cannot be ignored. Periodic redesign of the distribution network—whether it's in response to shifts within the business or a desire to move the cost vs. service exchange curve—is vital to the ongoing success of almost every business that distributes products.

So, what is "optimal" when it comes to a distribution network strategy? Ultimately, it's a commitment among key stakeholders to move forward with a strategy that balances the broad and future needs of the business. It goes beyond cost vs. service. It goes beyond algorithmic analysis. Ideally, it's a cost effective, service-friendly, and implementable plan that minimizes tax burden, mitigates risk, allows flexibility, and accommodates growth.

Of course, an "ideal" plan is an elusive target. Service and business requirements will continue to change over time, making it nearly impossible to implement the ideal. So in this sense "optimal" will never equal "ideal." But optimal is certainly forward-looking, broad in view, practical, and implementable. Further, an optimal distribution network strategy has the steadfast commitment of a broad array of stakeholders to agreed-upon goals. Developing an optimal distribution network strategy may not be easy, but it will have a hugely positive impact on your business success.