Executive Briefings

Multi-Echelon Supply Chains: Multi-Inventory, Multi-Problems

Inventory optimization solutions give companies a new weapon to combat the layers of safety stock that often plague complex, multi-echelon supply chains.

A million in safety stock here, a million in safety stock there, and pretty soon you're talking real money.

That take on the late Sen. Everett Dirksen's famous comment about government spending aptly describes the safety stock "creep" that plagues many of today's supply chains. Globalization, product complexity and high service demands all contribute to the problem, but perhaps the biggest culprit is the multi-echelon nature of supply chains that rely heavily on outsourced manufacturing and layers of suppliers and distribution points. If each layer adds just a little safety stock to protect against increased supply risk, longer global lead times or faster service requirements, you're soon talking real money-non-working capital sitting on a shelf, risking obsolescence and contributing nothing to the bottom line.

Traditional ways of managing inventory fail to address this problem for several reasons. One is that supply chain and enterprise resource planning systems are based on linear programming, says Rajib Roy, vice president of Optiant, Boston, which provides network and inventory optimization. "Therein lies the rub, because all the risks and variables of this world follow non-linear patterns," he says.

Another problem with local optimization is that it focuses only on one piece of the puzzle at a time, says David Simchi-Levi, co-founder and CEO of LogicTools Inc., Chicago. "When each facility tries to optimize its own decisions with little regard to the impact of those decisions on other parts of the supply chain, you end up with the overall supply chain having high inventory levels and low inventory turns," he says. "With the new inventory optimization approach, you look at the whole supply chain and try to identify what is appropriate for the entire network, not for each specific facility."

Most modeling and inventory decisions today are being done by rule of thumb, says Jeffrey Bodenstab, vice president of marketing at ToolsGroup, Cambridge, Mass. "Companies don't really understand how their safety stock is being calculated so they buffer this uncertainty with more inventory. With these new solutions, we are able to offer them mathematical or more scientific models that actually give them control, not just the impression that they are in control," he says.

Inventory optimization, or IO, is a relatively new class of enterprise solution designed to help companies optimize what, how much and where inventory should be placed throughout the supply network. Rather than a linear, rules-based approach, these solutions use a stochastic or probabilistic approach that takes into account multiple factors impacting demand and supply variability across many products and sites.

Various conditions converged around 2000 to enable development of IO tools. Ongoing work at several universities resulted in new, sophisticated algorithms designed to solve the multi-echelon and variability problems; data needed as inputs to these equations became widely available due to implementation of ERP and visibility software; and business computers had become powerful enough to quickly optimize very large problems.

Speed and scale both are critical to these solutions, says Sridhar Tayur, founder and CEO of SmartOps, Pittsburgh, Pa. "First, you have to be able to connect all the dots-and there are lots and lots of dots. That's the science," he says. "Second, you have to be able to optimize this vast amount of information very quickly. That's the software architecture. Once you have a brain that can handle this high level of complexity, you have to be able to spin that brain pretty fast."

Optiant, ToolsGroup, LogicTools and SmartOps pioneered the inventory optimization space with stand-alone solutions. Other supply-chain technology companies followed and now offer their own solutions for the multi-echelon problem. These include i2 Technologies, Dallas; Logility, Atlanta; Manhattan Associates, Atlanta; and Smart Software, Belmont, Mass. MCA Solutions, Philadelphia, specializes in inventory optimization for service parts and Rocky-Soft, Denver, provides inventory solutions geared to mid-sized companies.

 

Proof of Concept

Early adopters are proving the value of IO solutions. While only 20 percent of all companies recently surveyed by Aberdeen Group, Boston, report using multi-echelon optimization, that percentage jumps to 36 percent among best-in-class companies. Best in class in this survey is defined as having achieved a 96 percent or better service level while simultaneously reducing inventory carrying costs. Inventory reductions typically range from 20 percent to 30 percent, the report says.

In a few cases, these high performing companies actually are "leveraging their inventory as a competitive weapon," says Nari Viswanathan, Aberdeen research director and author of the Technology Strategies for Inventory Management Benchmark Report. They are looking beyond the cost benefits, he says, and "viewing inventory as a means to gain market share through superior service and product availability."

This strategy is apparent in what networking leader Cisco is doing to optimize inventory. "If I look at our results to date, it is less about a dramatic inventory reduction and more about smarter inventory," says Karl Braitberg, senior director of demand and management planning at Cisco. "We have inventory in the right places to meet demand surges and that has been the biggest win for us in the short term."

Cisco's manufacturing is mostly outsourced so its supply chain has many echelons. "We have manufacturing and distribution locations that cover the globe and inventory points that span between component suppliers all the way through finished goods," Braitberg says. The company produces high velocity, complex products, many of which are configure-to-order. "We have 20,000 SKUs across about 200 product families, ranging from high-end CRS routers all way down to telephones," he says.

One of the keys for Cisco has been to recognize that different parts of its product portfolio need different inventory policies. Basically, Cisco places its products on a 2 x 2 matrix, with supply uncertainty on one axis and demand uncertainty on the other, each moving from low to high. "Our products fall all over the place," Braitberg says. "We have lots of product where we have very good demand certainty, but we still have to look at other variables and make sure we understand how reliable the supply chain is in total. Then, instead of using an aggregate buffer strategy that may raise all product family buffers equally, we get more pointed and more statistically relevant on individual SKUs inside a family. We basically dial it for service levels to meet our customer expectations."

Optimizing inventory to meet specific service targets is a capability that many users of this technology are embracing. "These models enable different inventory policies and service levels to different customers," says Bodenstab. "By modeling the entire network, you are able to determine how much material at each location is required to meet a service goal."

Starting with a service level and then working backward can give new cost-saving insights, says Mike Matacunas, vice president of product strategy at Manhattan Associates. "A retailer, for example, might start with the question, how do I achieve a 99 percent service level in the store? With these systems, it can take all the variables-the transit time from the distribution center to the store, lead time from the supplier to the DC, minimum purchase quantities, storage capacity in the store, and so on, and do a very sophisticated analysis. The result is that companies often find they don't need to buy a product once a week. They can replenish every 10 days, reduce their carrying costs and still meet that 99 percent service goal."

Often this involves optimizing the product mix at each location. Aberdeen notes that a company can achieve an overall service level of 98 percent for a product group by setting some items at 99 percent and others at 95 percent, which can save 20 percent to 30 percent on inventory. "Mix optimization enables operations to reduce inventory while maintaining stringent customer service level requirements," according to the report.

Customer requests for a better way to calculate safety stock in support of established service levels led SmartSoftware to add inventory optimization to its suite of demand planning and forecasting solutions. "Our clients want us to help them think through their service requirements at all levels of the network," says president Charles Smart. "We may find that they need really high service levels for certain things at the lowest echelon of the network, whereas at a higher echelon, closer to the customer, their lead time may basically be transit time from the DC to a warehouse. Assuming that is relatively short, they won't need to maintain high levels at the warehouse, particularly on expensive items. We work through these trade-offs then calculate their inventory sweet spot, the point where they are able to satisfy service levels with the minimum amount of inventory."

"This is what is so exciting about IO technology," says Kiron Shastry, a consultant with Accenture. "Instead of a myopic view of the world, you now get a global view, which enables you to squeeze out that additional optimization and to do it in a targeted way. If the key thing to you is the service level for your customer, you can focus on that one thing and get an answer that shows you the amount of inventory you need across the network to achieve that goal."

Companies most advanced in inventory optimization are using this type of analysis to direct more supply and higher service to their most profitable demand, according to Aberdeen. As a result, "these companies are able to attain top line revenue increases through inventory management even in situations where the rest of their competitors are facing a flat market."

To make that strategy work, however, companies must be able to forecast at the customer level and be able to understand who are "the significant few and the trivial many" among their customers, says Adeel Najmi, solution executive at i2 Technologies. As part of its IO product, i2 offers a "very rich segmentation function," he says. "Our clients are able to dive into the transaction level details from ERP and automatically categorize customers by profitability, product profile or whatever." The solution also identifies buying behavior patterns by analyzing past sales history. "We build a statistical picture behind that, which lets our users create very different inventory policies based on demand patterns," says Najmi. "Inventory optimization really allows our clients to have different virtual supply chains with the same physical supply chain."

To complete the optimization process, i2 similarly analyzes supply variability, taking metrics that companies already are measuring about their supplier lead times and mapping those to determine supply risk. "We don't look at just the stated lead time but at how often the lead time is missed and by how much," Najmi says.

These factors then are used to calculate where and how to position inventory across the entire supply chain-from raw materials, through manufacturing and to finished goods distribution. "It is really important to marry inventory policy to the life cycle of the product," Najmi says. "Policies in product launch are very different as opposed to when you are ramped up to full volumes and then when you hit end of life."

Such a marriage also enables companies to understand the impact of product design decisions on inventory and life cycle costs, says Simchi-Levi. One of LogicTools's customers recently ran inventory optimization models during the product design process to help decide between product design alternatives so as to reduce lifecycle costs, he says. Similarly, the analysis supports better decisions on postponement strategies. "Many clients use our tools to identify how much postponement they need and where it should occur," he says. "In the chemical industry, for example, where lead times are as long as six to nine months, decisions about how much processing is needed at different buffering points has a huge impact on performance."

Logility also subscribes to this approach and has actually embedded its inventory optimization functions into product lifecycle management. "We see these two as being very tightly integrated," says Karin Bursa, vice president of marketing.

 

Service Parts

Products with intermittent demand also present a unique inventory challenge. "This includes not only slow-moving items, but higher volume products that may have seasonal patterns or promotion-driven demand," says Smart.

His company has a number of clients that deal with the intermittent demand problem in relation to service parts or spare parts, he says. One example is major U.S. theme park that uses Smart's demand forecasting, planning and inventory optimization solutions for its internal MRO operations. "They need to have a lot of spare parts on hand so the rides don't break down," he says. Even though the park covers only a few square miles, it still is a multi-echelon problem. "The park has the equivalent of central DC as well as local sites throughout that function as mini warehouses. They have to decide how much of each part to keep right next to the ride as opposed to how much to keep in a central location," he says.

MCA Solutions specializes in optimizing MRO parts for industries like aerospace, defense and high-tech. "These industries have and use products that are technologically complex and subject to random failure and maintenance requirements," says Morris Cohen, MCA chairman. As products have become more reliable, he notes, it has become more difficult to predict when and where they will fail. "And this is in an environment where response time is the key driver of customer satisfaction. It is not unusual to have a guaranteed response time of minutes when a critical machine goes down," he says.

"You cannot afford to have a copy of every part beside every machine in the field and you can't always afford to have redundant copies of that machine available-the cost of ownership becomes too high," says Cohen. "The answer is to selectively position parts at forward and central locations, with one location backing up another."

Decisions on where to place inventory involve not only a geographic hierarchy but also a product hierarchy, he says. "Do you stock the piece part, the assembly, the sub-assembly or the final field units? You can't solve these questions by breaking the problem into pieces and solving one location or one product at a time. Decisions about what to stock and where are highly interrelated and occur over these two networks."

In all industries, a close relationship exists between inventory optimization and network design, which evaluates where to locate supply chain facilities. "As a company changes its network, it also needs to make decisions about the inventory level at each location," says Simchi-Levi. "The beauty of this technology is that 70 percent of the data that is used for network design is used also for inventory optimization."

LogicTools, which has both solutions, recently introduced an extended version of LogicNet Plus that combines strategic and tactical functionalities. This move is in line with a recommendation from Aberdeen, which also advises companies to review network design at least annually instead of the typical two- to five-year cycle.

"We see more and more companies that understand they need to review their network more frequently," says Simchi-Levi. Unless there has been a major change, such as a merger or acquisition, companies that review the network annually or semi-annually usually do not have a full redesign in mind, he says. "Rather, they want to reconfigure and rebalance material flow across the network to better match supply and demand."

Inventory optimization also is a dynamic process that needs frequent evaluation, says Najmi. "The more we work with customers and see them move up the change management curve, the more we see the need to look at this frequently." Najmi says reviews should occur every one or two weeks, or perhaps once a month, in order to align inventory policies with changes in the supply chain. "You have to remember that inventory optimization is about changing the policies that determine what you will do tomorrow, so it allows you to be very proactive about regulating supply chain levers," he says.

Again, the Aberdeen survey provides confirmation: Results show that above average inventory performers are more than 2.5 times as likely as others to update their inventory strategies and policies multiple times a year.

Some companies are finding synergies between inventory optimization and sales and operations planning. "For a lot of companies, inventory is what is missing in S&OP," says Tayur. I have seen some clients start referring to the process as S&IOP to make sure inventory isn't forgotten."

"Inventory numbers are vital to the S&OP process and critical to understanding whether an organization is achieving the service level objectives and margin objectives that have been set out and where it needs to invest in the business," says Bursa. "It's all tied together."

Logility customer Shaw Industries provides one example. Shaw is a $6bn maker of carpet and importer of hard floor surfaces. "It's important for our operations people to really understand our inventory requirements so they can know if they are looking at a capacity roadblock and the need to add machinery at specific locations,' says Chris Whisenant, manager of logistics systems and forecasting. "The thing is, some of the machines we buy are $1m each, so the important question for us is that we know when we need to buy it. I would love to have that $1m in the bank for a month in order for me to better use it."

 

New CIO?

Renewed interest in S&OP holds the promise of addressing a widespread problem-the lack of single responsibility for inventory within corporations. "There is no one person in any company who is responsible for inventory," says Optiant's Roy. "No one has ever heard of a chief inventory officer." Roy believes that will change over the next five to seven years. "I don't know at what level, but there will be a role where someone will be responsible for all inventory throughout the supply chain," he says.  The lack of this position to date "is the biggest reason in my mind that we are not seeing more adoption of these solutions."

Aberdeen suggests that the issue can be addressed by cross-functional teams. It cites one high-tech company that has created a specific individual responsibility for inventory as part of an S&OP cross-functional team. In this case, however, the organizational change is not yet supported by the technology, which is still siloed.

Another organizational issue-the lack of skilled resources-is the top barrier to inventory technology adoption, according to Aberdeen.

Even some vendors agree that skilled resources are a problem. "What the Aberdeen report says is absolutely true," says Simchi-Levi. "Inventory optimization is a new technology and there is a lack of business professionals who understand this type of capability and who can use it effectively." However, he adds, this shouldn't be overstated. "It's like using an automobile," he says. "You don't have to be a mechanic, but you do have to know how to drive."

This is indicative of the need for more education among both existing and potential workers, says Roy. "To tell the truth, 70 percent of our sales cycle is pure education and training," he says. "There have been customers who have asked us to help them with continuous consulting because they don't believe they can upgrade their existing employees. And colleges aren't producing people with the right knowledge either."

Outsourcing this type of service appears to have potential, Aberdeen says. Two-thirds of respondents in a separate survey on managed services expressed interest in using a managed services approach for network design and strategic inventory optimization.

Some already have taken that step. Panasonic is using managed inventory management services from i2 to improve performance on one of its key product lines. "This frees up their line of business people to focus on core business decisions and leaves the data crunching and software operations to skilled resources from i2," says the Aberdeen report.

Outsourcing eventually may be an especially good answer for mid-sized companies, but at this point the mid-market is not interested in multi-echelon solutions, says Jeffrey Porter, vice president of business development at RockySoft, which specializes in this market. "We offered a multi-echelon inventory optimization solution a few years ago, but the reality was that people in this market were not ready for it and partnerships were not open to sharing the information."

The mid-market wants easy-to-use tools with a two- to six-month return on investment, he says. Moreover, he adds, "These companies typically don't have the stomach for a long implementation. When they sign the check they are ready for the go-live to happen because many are at a point where their resources are stretched to capacity. They need a remedy immediately."

This is still an early adopter market and, among all companies, the use of technology to manage inventory is "disturbingly low," says Aberdeen, which places it somewhere between 10 percent and 35 percent. Consequently, most companies view their current inventory management capability as immature. Only 9 percent say their current technology fully meets their network design needs and only 10 percent say it fully supports inventory optimization.

There are indications that this is changing, however. Two thirds of respondents to the Aberdeen survey say they have made or been asked to provide recommendations to management in the past six months on how to improve inventory management technology. And large companies are committing money: 54 percent plan to spend $100,000 or more in the next 12 months and 32 percent have budgeted $500,000 or more.

Companies that have not made investment plans should closely assess their inventory management technology, says Aberdeen. The right investments could "drive financial or customer service benefit and help the company remain competitive."

A million in safety stock here, a million in safety stock there, and pretty soon you're talking real money.

That take on the late Sen. Everett Dirksen's famous comment about government spending aptly describes the safety stock "creep" that plagues many of today's supply chains. Globalization, product complexity and high service demands all contribute to the problem, but perhaps the biggest culprit is the multi-echelon nature of supply chains that rely heavily on outsourced manufacturing and layers of suppliers and distribution points. If each layer adds just a little safety stock to protect against increased supply risk, longer global lead times or faster service requirements, you're soon talking real money-non-working capital sitting on a shelf, risking obsolescence and contributing nothing to the bottom line.

Traditional ways of managing inventory fail to address this problem for several reasons. One is that supply chain and enterprise resource planning systems are based on linear programming, says Rajib Roy, vice president of Optiant, Boston, which provides network and inventory optimization. "Therein lies the rub, because all the risks and variables of this world follow non-linear patterns," he says.

Another problem with local optimization is that it focuses only on one piece of the puzzle at a time, says David Simchi-Levi, co-founder and CEO of LogicTools Inc., Chicago. "When each facility tries to optimize its own decisions with little regard to the impact of those decisions on other parts of the supply chain, you end up with the overall supply chain having high inventory levels and low inventory turns," he says. "With the new inventory optimization approach, you look at the whole supply chain and try to identify what is appropriate for the entire network, not for each specific facility."

Most modeling and inventory decisions today are being done by rule of thumb, says Jeffrey Bodenstab, vice president of marketing at ToolsGroup, Cambridge, Mass. "Companies don't really understand how their safety stock is being calculated so they buffer this uncertainty with more inventory. With these new solutions, we are able to offer them mathematical or more scientific models that actually give them control, not just the impression that they are in control," he says.

Inventory optimization, or IO, is a relatively new class of enterprise solution designed to help companies optimize what, how much and where inventory should be placed throughout the supply network. Rather than a linear, rules-based approach, these solutions use a stochastic or probabilistic approach that takes into account multiple factors impacting demand and supply variability across many products and sites.

Various conditions converged around 2000 to enable development of IO tools. Ongoing work at several universities resulted in new, sophisticated algorithms designed to solve the multi-echelon and variability problems; data needed as inputs to these equations became widely available due to implementation of ERP and visibility software; and business computers had become powerful enough to quickly optimize very large problems.

Speed and scale both are critical to these solutions, says Sridhar Tayur, founder and CEO of SmartOps, Pittsburgh, Pa. "First, you have to be able to connect all the dots-and there are lots and lots of dots. That's the science," he says. "Second, you have to be able to optimize this vast amount of information very quickly. That's the software architecture. Once you have a brain that can handle this high level of complexity, you have to be able to spin that brain pretty fast."

Optiant, ToolsGroup, LogicTools and SmartOps pioneered the inventory optimization space with stand-alone solutions. Other supply-chain technology companies followed and now offer their own solutions for the multi-echelon problem. These include i2 Technologies, Dallas; Logility, Atlanta; Manhattan Associates, Atlanta; and Smart Software, Belmont, Mass. MCA Solutions, Philadelphia, specializes in inventory optimization for service parts and Rocky-Soft, Denver, provides inventory solutions geared to mid-sized companies.

 

Proof of Concept

Early adopters are proving the value of IO solutions. While only 20 percent of all companies recently surveyed by Aberdeen Group, Boston, report using multi-echelon optimization, that percentage jumps to 36 percent among best-in-class companies. Best in class in this survey is defined as having achieved a 96 percent or better service level while simultaneously reducing inventory carrying costs. Inventory reductions typically range from 20 percent to 30 percent, the report says.

In a few cases, these high performing companies actually are "leveraging their inventory as a competitive weapon," says Nari Viswanathan, Aberdeen research director and author of the Technology Strategies for Inventory Management Benchmark Report. They are looking beyond the cost benefits, he says, and "viewing inventory as a means to gain market share through superior service and product availability."

This strategy is apparent in what networking leader Cisco is doing to optimize inventory. "If I look at our results to date, it is less about a dramatic inventory reduction and more about smarter inventory," says Karl Braitberg, senior director of demand and management planning at Cisco. "We have inventory in the right places to meet demand surges and that has been the biggest win for us in the short term."

Cisco's manufacturing is mostly outsourced so its supply chain has many echelons. "We have manufacturing and distribution locations that cover the globe and inventory points that span between component suppliers all the way through finished goods," Braitberg says. The company produces high velocity, complex products, many of which are configure-to-order. "We have 20,000 SKUs across about 200 product families, ranging from high-end CRS routers all way down to telephones," he says.

One of the keys for Cisco has been to recognize that different parts of its product portfolio need different inventory policies. Basically, Cisco places its products on a 2 x 2 matrix, with supply uncertainty on one axis and demand uncertainty on the other, each moving from low to high. "Our products fall all over the place," Braitberg says. "We have lots of product where we have very good demand certainty, but we still have to look at other variables and make sure we understand how reliable the supply chain is in total. Then, instead of using an aggregate buffer strategy that may raise all product family buffers equally, we get more pointed and more statistically relevant on individual SKUs inside a family. We basically dial it for service levels to meet our customer expectations."

Optimizing inventory to meet specific service targets is a capability that many users of this technology are embracing. "These models enable different inventory policies and service levels to different customers," says Bodenstab. "By modeling the entire network, you are able to determine how much material at each location is required to meet a service goal."

Starting with a service level and then working backward can give new cost-saving insights, says Mike Matacunas, vice president of product strategy at Manhattan Associates. "A retailer, for example, might start with the question, how do I achieve a 99 percent service level in the store? With these systems, it can take all the variables-the transit time from the distribution center to the store, lead time from the supplier to the DC, minimum purchase quantities, storage capacity in the store, and so on, and do a very sophisticated analysis. The result is that companies often find they don't need to buy a product once a week. They can replenish every 10 days, reduce their carrying costs and still meet that 99 percent service goal."

Often this involves optimizing the product mix at each location. Aberdeen notes that a company can achieve an overall service level of 98 percent for a product group by setting some items at 99 percent and others at 95 percent, which can save 20 percent to 30 percent on inventory. "Mix optimization enables operations to reduce inventory while maintaining stringent customer service level requirements," according to the report.

Customer requests for a better way to calculate safety stock in support of established service levels led SmartSoftware to add inventory optimization to its suite of demand planning and forecasting solutions. "Our clients want us to help them think through their service requirements at all levels of the network," says president Charles Smart. "We may find that they need really high service levels for certain things at the lowest echelon of the network, whereas at a higher echelon, closer to the customer, their lead time may basically be transit time from the DC to a warehouse. Assuming that is relatively short, they won't need to maintain high levels at the warehouse, particularly on expensive items. We work through these trade-offs then calculate their inventory sweet spot, the point where they are able to satisfy service levels with the minimum amount of inventory."

"This is what is so exciting about IO technology," says Kiron Shastry, a consultant with Accenture. "Instead of a myopic view of the world, you now get a global view, which enables you to squeeze out that additional optimization and to do it in a targeted way. If the key thing to you is the service level for your customer, you can focus on that one thing and get an answer that shows you the amount of inventory you need across the network to achieve that goal."

Companies most advanced in inventory optimization are using this type of analysis to direct more supply and higher service to their most profitable demand, according to Aberdeen. As a result, "these companies are able to attain top line revenue increases through inventory management even in situations where the rest of their competitors are facing a flat market."

To make that strategy work, however, companies must be able to forecast at the customer level and be able to understand who are "the significant few and the trivial many" among their customers, says Adeel Najmi, solution executive at i2 Technologies. As part of its IO product, i2 offers a "very rich segmentation function," he says. "Our clients are able to dive into the transaction level details from ERP and automatically categorize customers by profitability, product profile or whatever." The solution also identifies buying behavior patterns by analyzing past sales history. "We build a statistical picture behind that, which lets our users create very different inventory policies based on demand patterns," says Najmi. "Inventory optimization really allows our clients to have different virtual supply chains with the same physical supply chain."

To complete the optimization process, i2 similarly analyzes supply variability, taking metrics that companies already are measuring about their supplier lead times and mapping those to determine supply risk. "We don't look at just the stated lead time but at how often the lead time is missed and by how much," Najmi says.

These factors then are used to calculate where and how to position inventory across the entire supply chain-from raw materials, through manufacturing and to finished goods distribution. "It is really important to marry inventory policy to the life cycle of the product," Najmi says. "Policies in product launch are very different as opposed to when you are ramped up to full volumes and then when you hit end of life."

Such a marriage also enables companies to understand the impact of product design decisions on inventory and life cycle costs, says Simchi-Levi. One of LogicTools's customers recently ran inventory optimization models during the product design process to help decide between product design alternatives so as to reduce lifecycle costs, he says. Similarly, the analysis supports better decisions on postponement strategies. "Many clients use our tools to identify how much postponement they need and where it should occur," he says. "In the chemical industry, for example, where lead times are as long as six to nine months, decisions about how much processing is needed at different buffering points has a huge impact on performance."

Logility also subscribes to this approach and has actually embedded its inventory optimization functions into product lifecycle management. "We see these two as being very tightly integrated," says Karin Bursa, vice president of marketing.

 

Service Parts

Products with intermittent demand also present a unique inventory challenge. "This includes not only slow-moving items, but higher volume products that may have seasonal patterns or promotion-driven demand," says Smart.

His company has a number of clients that deal with the intermittent demand problem in relation to service parts or spare parts, he says. One example is major U.S. theme park that uses Smart's demand forecasting, planning and inventory optimization solutions for its internal MRO operations. "They need to have a lot of spare parts on hand so the rides don't break down," he says. Even though the park covers only a few square miles, it still is a multi-echelon problem. "The park has the equivalent of central DC as well as local sites throughout that function as mini warehouses. They have to decide how much of each part to keep right next to the ride as opposed to how much to keep in a central location," he says.

MCA Solutions specializes in optimizing MRO parts for industries like aerospace, defense and high-tech. "These industries have and use products that are technologically complex and subject to random failure and maintenance requirements," says Morris Cohen, MCA chairman. As products have become more reliable, he notes, it has become more difficult to predict when and where they will fail. "And this is in an environment where response time is the key driver of customer satisfaction. It is not unusual to have a guaranteed response time of minutes when a critical machine goes down," he says.

"You cannot afford to have a copy of every part beside every machine in the field and you can't always afford to have redundant copies of that machine available-the cost of ownership becomes too high," says Cohen. "The answer is to selectively position parts at forward and central locations, with one location backing up another."

Decisions on where to place inventory involve not only a geographic hierarchy but also a product hierarchy, he says. "Do you stock the piece part, the assembly, the sub-assembly or the final field units? You can't solve these questions by breaking the problem into pieces and solving one location or one product at a time. Decisions about what to stock and where are highly interrelated and occur over these two networks."

In all industries, a close relationship exists between inventory optimization and network design, which evaluates where to locate supply chain facilities. "As a company changes its network, it also needs to make decisions about the inventory level at each location," says Simchi-Levi. "The beauty of this technology is that 70 percent of the data that is used for network design is used also for inventory optimization."

LogicTools, which has both solutions, recently introduced an extended version of LogicNet Plus that combines strategic and tactical functionalities. This move is in line with a recommendation from Aberdeen, which also advises companies to review network design at least annually instead of the typical two- to five-year cycle.

"We see more and more companies that understand they need to review their network more frequently," says Simchi-Levi. Unless there has been a major change, such as a merger or acquisition, companies that review the network annually or semi-annually usually do not have a full redesign in mind, he says. "Rather, they want to reconfigure and rebalance material flow across the network to better match supply and demand."

Inventory optimization also is a dynamic process that needs frequent evaluation, says Najmi. "The more we work with customers and see them move up the change management curve, the more we see the need to look at this frequently." Najmi says reviews should occur every one or two weeks, or perhaps once a month, in order to align inventory policies with changes in the supply chain. "You have to remember that inventory optimization is about changing the policies that determine what you will do tomorrow, so it allows you to be very proactive about regulating supply chain levers," he says.

Again, the Aberdeen survey provides confirmation: Results show that above average inventory performers are more than 2.5 times as likely as others to update their inventory strategies and policies multiple times a year.

Some companies are finding synergies between inventory optimization and sales and operations planning. "For a lot of companies, inventory is what is missing in S&OP," says Tayur. I have seen some clients start referring to the process as S&IOP to make sure inventory isn't forgotten."

"Inventory numbers are vital to the S&OP process and critical to understanding whether an organization is achieving the service level objectives and margin objectives that have been set out and where it needs to invest in the business," says Bursa. "It's all tied together."

Logility customer Shaw Industries provides one example. Shaw is a $6bn maker of carpet and importer of hard floor surfaces. "It's important for our operations people to really understand our inventory requirements so they can know if they are looking at a capacity roadblock and the need to add machinery at specific locations,' says Chris Whisenant, manager of logistics systems and forecasting. "The thing is, some of the machines we buy are $1m each, so the important question for us is that we know when we need to buy it. I would love to have that $1m in the bank for a month in order for me to better use it."

 

New CIO?

Renewed interest in S&OP holds the promise of addressing a widespread problem-the lack of single responsibility for inventory within corporations. "There is no one person in any company who is responsible for inventory," says Optiant's Roy. "No one has ever heard of a chief inventory officer." Roy believes that will change over the next five to seven years. "I don't know at what level, but there will be a role where someone will be responsible for all inventory throughout the supply chain," he says.  The lack of this position to date "is the biggest reason in my mind that we are not seeing more adoption of these solutions."

Aberdeen suggests that the issue can be addressed by cross-functional teams. It cites one high-tech company that has created a specific individual responsibility for inventory as part of an S&OP cross-functional team. In this case, however, the organizational change is not yet supported by the technology, which is still siloed.

Another organizational issue-the lack of skilled resources-is the top barrier to inventory technology adoption, according to Aberdeen.

Even some vendors agree that skilled resources are a problem. "What the Aberdeen report says is absolutely true," says Simchi-Levi. "Inventory optimization is a new technology and there is a lack of business professionals who understand this type of capability and who can use it effectively." However, he adds, this shouldn't be overstated. "It's like using an automobile," he says. "You don't have to be a mechanic, but you do have to know how to drive."

This is indicative of the need for more education among both existing and potential workers, says Roy. "To tell the truth, 70 percent of our sales cycle is pure education and training," he says. "There have been customers who have asked us to help them with continuous consulting because they don't believe they can upgrade their existing employees. And colleges aren't producing people with the right knowledge either."

Outsourcing this type of service appears to have potential, Aberdeen says. Two-thirds of respondents in a separate survey on managed services expressed interest in using a managed services approach for network design and strategic inventory optimization.

Some already have taken that step. Panasonic is using managed inventory management services from i2 to improve performance on one of its key product lines. "This frees up their line of business people to focus on core business decisions and leaves the data crunching and software operations to skilled resources from i2," says the Aberdeen report.

Outsourcing eventually may be an especially good answer for mid-sized companies, but at this point the mid-market is not interested in multi-echelon solutions, says Jeffrey Porter, vice president of business development at RockySoft, which specializes in this market. "We offered a multi-echelon inventory optimization solution a few years ago, but the reality was that people in this market were not ready for it and partnerships were not open to sharing the information."

The mid-market wants easy-to-use tools with a two- to six-month return on investment, he says. Moreover, he adds, "These companies typically don't have the stomach for a long implementation. When they sign the check they are ready for the go-live to happen because many are at a point where their resources are stretched to capacity. They need a remedy immediately."

This is still an early adopter market and, among all companies, the use of technology to manage inventory is "disturbingly low," says Aberdeen, which places it somewhere between 10 percent and 35 percent. Consequently, most companies view their current inventory management capability as immature. Only 9 percent say their current technology fully meets their network design needs and only 10 percent say it fully supports inventory optimization.

There are indications that this is changing, however. Two thirds of respondents to the Aberdeen survey say they have made or been asked to provide recommendations to management in the past six months on how to improve inventory management technology. And large companies are committing money: 54 percent plan to spend $100,000 or more in the next 12 months and 32 percent have budgeted $500,000 or more.

Companies that have not made investment plans should closely assess their inventory management technology, says Aberdeen. The right investments could "drive financial or customer service benefit and help the company remain competitive."