Super-Charging Your Supply Chain

Companies use different options to fight the never-ending battle of optimizing their supply chains.


Logistics optimization is neither easy nor cheap, but it is the biggest opportunity for most companies to significantly reduce costs,” notes Don Ratliff, Ph.D, president and CEO of Velant Inc., an Atlanta-based transportation solution company. He is also executive director of the Logistics Institute of the Georgia Institute of Technology, Atlanta.

Optimization means different things to different people. “Most people use the word ‘optimization’ in terms of making something better or good,” says Richard F. Powers, co-founder, president, and CEO of Insight, a supply chain optimization company based in Manassas, Va.

That’s not really what it means, at least not to operations researchers such as Powers. To them, “optimization has a specific meaning: a model that guarantees you mathematically have the best possible solution considering the variables you put into it,” Powers says.

He links optimization with planning, “looking into the future and, given certain variables and events that might happen, determining the best course of action to achieve your objective.”

While this approach is great for optimizing the planning process, it overlooks other key processes, notes Kate Vitasek, managing partner of Supply Chain Visions, a supply chain strategy company in Bellevue, Wash. “You need to look across the four to six key processes that run your business.”

Vitasek’s approach to optimization ties back to performance management. “Does the company focus on getting results relative to its strategic goals?”

Most companies optimize at the functional level and not necessarily at the process level, she says, which can result in functional work being optimized while the overall supply chain is suboptimized.

The Supply-Chain Council’s Supply-Chain Operations Reference (SCOR) model provides a framework that helps companies optimize their supply chains, says Scott Stephens, chief technology officer for the Council.

SCOR is “the standard process reference model for communicating supply chain management practices across companies,” according to the Council. It enables companies to capture the “as-is” or current state of a practice and to derive the desired “to-be” or future state.

Then there’s optimizing logistics execution. “We normally differentiate between the optimization of design and strategy issues—such as where to put your facilities, or what size quantities to deliver to customers—and the optimization of logistics execution,” notes Don Ratliff.

Every day, companies can optimize logistics decisions such as how they will get product delivered to customers, and how often to reorder inventory. Whatever level of optimization a company chooses, the payoff can be significant.

“If you do optimization right, you can save an incredible amount of money,” Powers says. “The benefits could be in the millions of dollars.”

Companies could save 10 to 15 percent of total logistics costs from properly aligning their logistics networks, he estimates. Optimizing production planning could save another 10 to 15 percent, and optimizing production scheduling could potentially cut an additional 10 percent in logistics costs. Optimizing transportation can save a company 10 to 40 percent in transportation costs, Ratliff projects.

One company that had no previous experience with SCOR achieved a $4.3-million return on its $50,000 investment in implementing the model, Scott Stephens reports. He also cites the case of an electronics company that invested $3 to $5 million (which included software) in implementing the model. The company completed implementation in about five months and is on track to achieve a $230-million return in the multi-year project.

The SCOR model includes a cross-functional framework, standard terminology, common metrics, and best practices. The model is organized around five management processes—Plan, Source, Make, Deliver, and Return. Supply-Chain Council members developed the model, which is continuously enhanced, in 1996 and 1997.

“We use the model to describe the supply chain, using relatively high-level process modeling,” Stephens notes. “The difference is, traditional modeling techniques take months or years, while the SCOR model allows you to do it in hours or days.”

While not an intricate model, it is sufficiently detailed to allow supply chain partners to communicate effectively. It also allows individuals and organizations to go to the level of detail they require.

“Everybody uses the SCOR model differently,” Stephens says. It has been used successfully at both the corporate and local level, and by companies ranging from the Fortune 50 to medium and small businesses.

What the model has in common is that “it moves you out of the silos of purchasing, raw materials, work-in-process and finished goods, transportation, warehousing, and order management into a process flow,” he explains.

From As-Is to To-Be

Dan Peck, now director of planning and logistics at American Coin Merchandising Inc., was first exposed to the SCOR model when he was with Procter & Gamble, shortly after the Supply-Chain Council started up.

“I liked how the model uses metrics. It is simple, easy to follow, and doesn’t take a large investment to get going,” Peck recalls.

When Peck recently joined American Coin Merchandising (ACMI), he gravitated to the SCOR model right away. “It was a good process for me to learn ACMI’s supply chain,” he says.

ACMI, based in Boulder, Colo., supplies amusement vending equipment to restaurants, bowling alleys, and retailers such as Wal-Mart. The company, about 15 years old, has been growing fast.

“ACMI has been throwing together bits and pieces of the supply chain as it needed them,” Peck explains. “My job was to come in and rationalize everything.”

It took Peck about one month to document the as-is model for ACMI’s toy supply chain. The toys, used as prizes in the company’s games, are procured from suppliers in China.

“Then I started working on different options for what we wanted the supply chain to be,” Peck says. “I did some to-be modeling and tried different options to see how they looked.”

After deciding the general direction in which the company wanted to go, the next step was to design a new logistics network using network optimization software.

“Right now we have one distribution center in Kent, Wash. We bring all our toys from China to that DC, then distribute around the United States from there,” Peck explains. In the to-be model, which is currently being implemented, the large DC will move to China, “located centrally with all the vendors, so that we can pull toys from the vendor as we need them,” he says.

The DC in China will ship directly to six new regional or crossdock centers, which will be set up in the United States. The new network will cut overall cycle time, reduce inventory by one- third, and generate cost savings of 15 to 20 percent, Peck projects.

After the toy supply chain is optimized, ACMI will use the SCOR model to optimize its supply chain for machines.

“The SCOR model is a great way to get a standardized look at your supply chain, and break it down into easy bite-sized chunks,” Peck says. It’s also a practical way to start supply chain optimization “particularly with a complex supply chain or if you haven’t done supply chain optimization before,” he notes.

SCOR Exceeds All Targets

William Frank Quiett, C.P.M., A.P.P., agrees. He’s a project lead of strategic sourcing and supply chain management for United Space Alliance, LLC, at Kennedy Space Center.

“All of us want an optimized supply chain. But how do you start?” he asks. The United Space Alliance, like numerous other organizations in a variety of industries, found the answer and process it needed in the Supply-Chain Council’s SCOR model.

“We use the SCOR model as a framework to lead us through identifying our as-is supply chain, and comparing attributes of different segments of the supply chain to best practices,” he says.

The organization began using the model two years ago, completing its pilot program late last year. “We exceeded all our targets significantly,” Quiett reports.

Today, United Space Alliance is analyzing the feasibility of rolling out the optimization program to the rest of the enterprise. “We’ve proven that the SCOR model works in our environment, and are now analyzing the next step,” he says.

Quiett recommends the SCOR model as an optimization tool. “It helps you identify, analyze, and benchmark your current supply chain,” he explains. These are critical steps in the optimization process. “You can’t optimize unless you know what you’re doing and how well you’re doing it,” Quiett says.

Intuit Quickens its Supply Chain

Intuit Inc., Mountain View, Calif., is a leading provider of business and financial management solutions—including the best-selling Quicken and QuickBooks—for small businesses, consumers, and accounting professionals.

Intuit takes a different approach to supply chain optimization. Until recently, Intuit served retailers through distributors. Now, however, Intuit has direct relationships with major retailers, such as Staples.

In addition, Intuit has a direct channel, through which product is shipped directly to end-users, value-added resellers, and small businesses. Changing to direct store fulfillment means that Intuit has gone from supporting 600 storefronts to 11,500 storefronts in the past three years.

“That creates a much different fulfillment infrastructure requirement,” according to Brett Aumack, Intuit’s director of supply chain strategy. “We had to scale the infrastructure and add partners.” The company currently outsources its manufacturing and fulfillment.

Intuit is tackling a number of optimization initiatives, including:

SKU stratification. Intuit is steadily broadening its product line, for example, developing business products for specific industries to create “right for my business” solutions. “This creates wild and crazy SKU proliferation,” Aumack says.

So the company did an ABC analysis of SKUs, classifying them based on volume and variability, and determining the optimum fulfillment mode. Intuit now targets “C” SKUs—those with low volume and velocity—for a make-to-order model, burning individual CDs and printing manuals using print-on-demand technology when an order is received. “B” items are built to order, using a kanban approach. “A” SKUs, which are high volume with low variability and very predictable, are made to stock.

Using different fulfillment models enables Intuit to control inventory levels and effectively manage an expanding number of SKUs.

“We expect to see more flavors of products,” Aumack says. “We need to make sure we don’t scale the infrastructure linearly against the SKU proliferation. We want to make sure we focus on the core SKUs that make up 80 percent of our volume, and probably 80 percent of the dollars.”

Expected Delivery Date management. “Every customer has a different critical date,” Aumack says, “the date on which they measure our success. We decided to go to each customer to determine what date in their system they draw our performance metrics from.”

After gathering the information from its key retailers, fine-tuning and confirming it with their customers, Intuit began tracking its ability to deliver on time and complete by the customer’s critical date.

“We hoped to see an increase in our on-time and complete performance, and we did,” Aumack says. Performance went from 83 percent to 98 percent in less than one year, and the amount of expedited freight was reduced significantly.

The first phase of the Expected Delivery Date initiative was at the corporate retail level. Intuit is now going to the next level, customizing delivery goals by individual store.

For example, Intuit may build an extra day into the time allotted for shipping to a specific store because of special circumstances at that location.

Improve internal demand planning. Intuit has taken several steps to improve the accuracy and timeliness of its internal demand planning, including building an early warning system that triggers an exception alert.

“By putting in more structured processes and reports, and improving the timing, we’ve upgraded our internal demand planning process,” Aumack notes. “It used to take two or three days to digest the necessary information to produce a master schedule. Now it’s down to 20 minutes,” and the plan is more accurate.

Intuit is building the capability for its system to ping the forecast owner with alerts about trends, and the need to make forecast modifications before stockouts occur.

Intuit is also working on improving supply/demand alignment, shifting from defining demand as consumption to defining demand as point of sale.

Optimize logistics network design. Intuit analyzed its logistics network, looking at how many locations it needed, and where its manufacturing locations and suppliers should be.

“We’ve tried co-location models with particular outsourcers. Then we started to default to more near point geographies,” Aumack explains. “We are trying to get suppliers, assembly, and manufacturing partners regionalized, and to limit the number of regional cells we’ll use across the United States.”

Intuit’s business is highly seasonal, with a busy peak period that runs from November to January, and a slow period running from April to July. As a result, the company is building a dynamic network model that will flex the number of sites depending upon the season.

For example, “we’d have a cell of suppliers on the West Coast for a period of time, then not use that cell during the slow period,” Aumack says. Moving to this flexible, scaleable model will enable Intuit to boost product availability, improve cycle time, cut costs, and improve service.

A key part of Intuit’s optimization efforts has been revamping its performance measurement process. After interviewing customers and analyzing performance, a crossfunctional team recommended moving from measuring fill rate to measuring on-time and complete, which tracks the performance of the delivery process from order receipt to product receipt. The metric was designed by Intuit colleague Patrick Maiorano.

Process-focused metrics such as this “are aimed at optimizing customer satisfaction, rather than optimizing at the individual functional level,” notes Kate Vitasek.

Fill rate, for example, optimizes at the functional level. Most companies today use functionally-organized metrics. “Metrics should be aligned to corporate goals. Your goals should drive your strategic objectives, which should drive your metrics,” Vitasek says. “Your individual metrics should support the functional work that supports the strategic objectives.”

Take fill rate. “If you have a 99.9-percent fill rate, it’s probably not optimizing for the company. In order to have that high a fill rate, you probably have too much inventory.

So, Vitasek asks, “does the company care more about better profits or the 99.9-percent fill rate?” If having happy customers at a profitable base is the goal, “you’ll probably want a 97- to 98-percent fill rate,” she says.

For Good Measure

To develop process-focused metrics, Vitasek recommends putting together a crossfunctional team to analyze existing functional metrics. Using fill rates as an example again, Vitasek recommends bringing together “the distribution people, the order management people who work with EDI and customer service, and the transportation management people who order the trucks.”

The goal: to discuss the tradeoffs involved, and to understand how fill rate fits in the delivery process metric—”Did customers get the product when they wanted it, on time, and complete?”

Companies should take the same type of approach for all key processes, Vitasek says. “Optimization needs to be looked at crossfunctionally, at the corporate goal level, not just at the functional level,” she says.

10 Rules of Logistics Optimization

Optimized planning can be done at both the strategic and day-to-day execution levels, notes Don Ratliff, president and CEO of Velant Inc., and executive director of the Logistics Institute of the Georgia Institute of Technology, Atlanta.

An example of the strategic level is optimizing location of facilities, while an example of the execution level is optimizing loads, routes, and schedules for trucks that deliver products each day.

Ratliff suggests the following 10 rules of logistics optimization that apply to any part of logistics, but particularly pertain to optimizing logistics and supply chain execution.

1. Objectives must be quantified and measurable. “In a lot of cases, people still do not quantify their objectives,” Ratliff says. “If you’re going to optimize something, you have to decide how you’ll know it’s optimized.” Suppose you’re considering two ways to load, route, and schedule trucks. “You can’t tell which way is better unless you have some means of putting numbers to it, and can measure it,” Ratliff explains.

2. Models must faithfully represent required logistics processes. “Models enable us to translate what we’re trying to do in terms of objectives and constraints on the system into something the computer can understand,” says Ratliff. “You have to have models that represent the real world well enough so that when the computer identifies a solution, it will actually work.”

For example, a model used to optimize the way trucks are loaded may need to consider volume, product characteristics, and weight instead of just weight to develop an effective solution.

3. Data must be accurate, timely, and comprehensive. “Data is what drives logistics optimization,” Ratliff notes. Having lots of data is not enough—it has to be accurate, and kept up to date.

However, “there’s a tendency to set up software and let it run, and pretend that you don’t have to do anything further,” he observes.

It’s not uncommon for product dimensions that have changed slightly to not be updated in the model, which means the model won’t give optimal results. So it’s important to keep data cleansed and up to date. “You have to stay on top of this all the time,” Ratliff warns.

4. Integration must supply fully automated data transfer. Logistics optimization requires reams of data. Companies without integrated systems have to enter data manually, which hampers getting the accurate, timely, and comprehensive data that’s required.

“The only way I know to get the data you need is through automated data transfer, with translation and checking done in an automated way,” says Ratliff.

5. Optimized plans must be delivered in a form that facilitates execution, management, and control. “You have to get the solutions to the people that will implement them in a practical way, in a form they can use,” he says. “You can’t just dump a bunch of data on people—you have to provide solutions in a way they can take advantage of,” so they can execute the optimized plan.

6. Algorithms must intelligently exploit individual problem structure. Algorithms are the computer-based processing strategies used to find the best logistics plan, Ratliff explains. They enable the model to consider huge numbers of possible combinations, such as the trillion possible load combinations that exist for 40 shipments.

Because of the complexities of logistics operations, “you end up with an astronomically large number of possibilities,” so much so that it’s simply not possible for people to run such calculations.

“You need a lot of computing power, and algorithms that let you take advantage of a problem’s special structure,” he says, pointing out that each problem’s structure is unique.

To be effective, algorithms must be tunable, or capable of being fine-tuned to meet the structure of a particular problem. “A lot of algorithms are structured to be one-size-fits-all,” Ratliff says.

In addition to the technology being flexible enough to accommodate tuning, it has to be operated by an algorithm specialist who truly understands how to tune it to get optimal results.

7. Computing platforms must have sufficient power to produce optimum plans in the time required. “You have to have enough power to get the good plans you’re looking for in the timeframe in which you need to get them,” Ratliff says. Unfortunately, many companies have not invested in the computing power required to do so.

8. People responsible for the technology must have the domain and technology expertise required to support the models, data, and optimization engines. Optimization technology really is rocket science, Ratliff explains, so you have to have rocket scientists to run it.

“Having the technology is not enough—you have to have people with the domain and technical expertise to support all of this. Optimization is very complex,” Ratliff says. “It doesn’t have to be complex to the people who take advantage of the results, but it is complex to the people who are inside the engine.”

9. Business processes must support optimization and have the ability to continuously improve. “If you’re going to get into optimization, you have to build a process that supports it,” he warns. A continuous improvement process in which the data, models, and optimization engine are continually improved and kept up to date is a critical success factor for effective optimization.

10. Return on investment must be provable, considering the total cost of technology, people, and operations. Logistics optimization can require significant expenditures in technology and people.

“Because of the complexity and effort required to do optimization, you have to believe that you’ll get a good ROI from doing it,” according to Ratliff.

Developing an ROI requires “having good ways to determine what your baseline is, the cost of technology and the people involved, and how much improvement you expect,” then continually monitoring performance. Few companies today know how effective their logistics optimization efforts actually are, he says.

Optimization 101

Companies often use one of three different approaches to supply chain planning, according to Richard F. Powers, Ph.D., co-founder, president, and CEO of INSIGHT, Bend, Ore.

The three approaches are:

1. Optimization, which involves using a mathematical model to identify the best possible solution. Optimization is prescriptive. It will identify a mathematically pure and algorithmically provable solution, Powers explains.

2. Heuristics, or rules of thumb. “Essentially, these are rules or rules-based procedures that we’ve learned from experience generally seem to work.” Because they’re based on previous experience and common sense, “heuristics perpetuate history,” Powers notes.

3. Simulation. “This isn’t prescriptive, it’s descriptive,” Powers points out. “You take a process and reproduce it on a computer, then introduce probabilities. Then you set up a simulation model and run it thousands of times, based on the random distribution of how things happen.”

Simulation models can effectively be used to simulate such things as demand and out-of-stock occurrences. “Simulation doesn’t give you the answer,” Powers notes. “Instead, it lets you try things and see what happens.

“Optimization is prescriptive, providing you with the best answer given all the data that you’ve provided,” he says. “Simulation tells you what’s going to happen if you do things in a certain way.”