Executive Briefings

Inventory Optimization Is a Delicate Balancing Act to Cut Costs Yet Keep Service Levels Up

Afraid to let safety stocks get too low, companies often get overwhelmed by their inventory. New ideas, processes and technologies in inventory management can solve that.

Business for Storage Technology Corporation is about helping an extremely diverse client base keep on top of their business-critical data storage needs. It often means making sure that needed parts are delivered for service to a client on time; more often than not within two hours of the customer placing a service request. That isn't so easy to do if you aren't on top of your own storage-your inventory of printed circuit boards, disk drive assemblies and myriad other service parts. But Storage Technology, or StorageTek as it is more commonly known, has that part down now.

For 20 years or more StorageTek has had some kind of automated software to help manage its service parts inventory. Initially, the software was an internal creation. Ten years ago, StorageTek went to an outside developer for a system, but by 2003 it felt its needs had outstripped the capabilities of that product. What the Louisville, Colo.-based company felt it needed was a system that could not only disseminate order entries immediately but determine how much inventory it needed to stock based on customers' contractual service requirements and how many distribution centers it needed to maintain overall. Like any manufacturer and retailer worldwide, regardless of vertical or product line, it needed an optimized inventory approach.

How much to keep, and where to keep it are basic questions to address in implementing any inventory optimization policy. But if any part of a company can ill afford to act on its own, it's those tasked with managing inventory. They often are stuck in between production and demand planners on one side and the transportation department on the other. Depending on the nature of the business, inventory managers must juggle inbound and outbound, balance safety stocks and service-level objectives, make sense of forecasts that often are more exercises in wishful thinking than anything else, and, oh yes, keep upper management happy by keeping a lid on costs. Seldom can this be done without some outside consultation and expertise.

StorageTek was acutely aware of what it wanted. It had customers in 63 countries, but its systems weren't keeping order information around the world as timely as desired. "The existing software wasn't following the sun," says James Matney, the company's director of global service logistics.

StorageTek also wanted to import contract database intelligence into the service parts planning system so response-time guarantees that customers were paying for could be honored. But it also needed to know how many parts to stock for those clients. Previous generations of software weren't very smart; rather, they just looked at past usage history.

Finally, StorageTek needed an application that would "graphically display where are the best places to put stocking locations and how many of these stocking locations you should have," Matney says.

The information storage solutions provider turned to Servigistics of Atlanta to provide the storage parts planning and distribution software.

Whether they call it modeling, network design or by some other name, companies like Servigistics make inventory optimization suggestions based on an analysis of the location of the client's customers and its current DCs, and the nature of the obligations mandated by service-level agreements the client has signed. The assessment attempts to determine not only how many parts should be kept on hand but where.

"Not all of them have to be at field service locations," says Cliff Isaacson, director of product marketing for Servigistics. Pointing to Avaya, the communications systems giant, he says the cost of a given part may be a factor in the analysis. "Some of its parts are worth a hundred or two hundred thousand dollars, because they are very high-end network and telecommunication equipment. You obviously don't want to stock that part at every single service location. If there's an opportunity to pool a part, meaning having a single part in one place that supports multiple service locations, we [look] for that."

While Matney runs the logistics business at StorageTek, the Servigistics software "runs where all the spare parts need to be, how many parts are being used, calculating desired stocking amounts, and then every night it's looking for changes," he says. "If a part's been used, and it drops below a certain level, it generates an order and sends it up to our ERP system to either move a part from a distribution center, move a part from another depot, or for us to go procure a part. It's doing all the logic and creating the transactional orders."

Follow the Money
In Isaacson's view, the "glamour" part of the supply chain has always been in enterprise software. The service supply chain was an afterthought, he says, until management realized the huge gross margins that parts management and sales could bring. Every time a new primary good was introduced-a car or a cellphone, for example-a whole new set of service parts was introduced as well. That brought added opportunities for aftermarket sales, but it also caused a huge proliferation in SKUs of both new and now-obsolete parts.

"In many cases companies know the problem is there," Isaacson says. "It's just that their focus and attention on technology has been on the front end of the supply chain. And it's only been in the last few years that there have been technologies and computing power strong enough to consider all the different service parts. One thing that's characteristic of service parts: If you think of the limited set of primary goods that companies are selling, there are millions of service parts that are associated with repairing them or replacing them. So the volume of parts and parts locations is a really big problem to solve."

Though there is a lot of money to be saved in DC consolidation, customers are generally cautious because of the risk in negatively affecting customer service. "The bottom line is you consistently want to hit those service levels, especially if you believe you drive greater customer growth by having better service than your competition," Isaacson says.

Trading partner separations or disconnects can occur anywhere in the supply chain, and the more extended it is, perhaps the greater the risk. "Inventory typically is the result of poor information between planners and the warehouse," says Jim Erdman, Penske Logistics' senior vice president for the Asia Pacific region. "Here in China, it's probably the biggest single problem. There's just not a lot of good information so everybody tends to have a lot of stock."

Creating Obsolescence
It's particularly acute in the electronics industry, which has a large presence in the country, but occurs in other sectors as well, Erdman says. Information gaps create obsolescence. "Electronics manufacturers still build products that have become obsolete. They don't want to have too much inventory, but they need enough to meet the demand."

Erdman said one industrial machine manufacturer, who sourced half or more of his product's components from outside of China, had a terrible overstocking problem due to information gaps. Data transfer was so poor, even from a parts supplier in Taiwan, that the manufacturer kept six to nine months' of inventory on hand.

Compounding the problem is the length of the supply chain, Erdman says. "Somebody once wrote that the 'tyranny of China is its locationn' because it's so far away from everything. It's true. And with the growth in Asia, the congestion at the ports, the capacity restraints in North America and even in Europe, it's just getting further away. If you work on something with, say, a 36-day cycle, there's no guarantee that you're going to get 36 days. It may be 40 days because there's a tie-up in Long Beach. So everybody overcompensates for those schedules by carrying more inventory than they need. If it's supposed to be a 36-day cycle, but a lot of times it's been 40, they'll probably keep 50 days' or 60 days' worth just to make sure."

The Chinese have made tremendous strides, Erdman says, "but there is still a awful lot done on gut feeling and not on information. They haven't invested enough in the technology side to get that information."

It's not a situation at all alien to Charles Smart, president of Belmont, Mass.-based Smart Software. His company's SmartForecasts product essentially analyzes demand or sales data for a product item, considers any intermittent demand patterns or seasonality factors, then determines the optimal forecasting method based on the appropriate lead time for that product.

"But in that process, we also have to analyze and project what the expected deviation from the forecast might be," says Smart. "And that information is the basis that we use to create estimates of safety stock parameters to add to the demand forecast-to get the inventory levels to correspond to the service levels that a client might desire."

Often this analysis is missing when customers forecast on their own, Smart says. "When they get down to taking the demand forecast and adding something to it, they don't use much analysis. They use a lot of-almost folklore-in creating the safety stock. 'Well, we've always doubled the demand forecast to get the buffer stock. Or if the lead time is always 3 weeks, we've always added 6 weeks of expected demand to get the safety stock.'

"We still find a lot of companies locked into that sort of ad hoc approach to inventory planning and stocking." It's a huge waste of money and generally fails to achieve service-level objects, Smart says.

Demand is what drives inventory, Erdman says. "Good demand forecasting is the most important thing in controlling inventory."

Jonathan Colehower, CEO of Optiant of Burlington, Mass., might say that the most important thing in keeping a lid on inventory is staying on top of variability in a supply chain that's becoming "unbundled."

The typically vertically integrated company that sources, manufactures and distributes on its own is no more for the most part. "You may have one organization do the sourcing for you, you have a contract manufacturer doing some of the base manufacturing, another person doing the subassembly, then you may have a 3PL doing the warehousing and distribution. Through this unbundling, we're seeing two things: risk and the need for data synchronization."

Of course, inventory is impacted because variability is introduced with every step in the supply chain, Colehower says. "'Am I going to get it in two days? Or am I going to get it in a week?'

"We need inventory for two reasons: one is demand variability. How often am I going to be demanding of it? Second is supply variability, or lead time variability.

"Any time you introduce variability into the supply chain, the natural response from any company is to buffer it with inventory."

Whose Inventory?
The solution should not be to push inventory around the supply base like billiard balls on a table, Colehower says. "I've witnessed companies write into their suppliers' contracts, 'You must keep 25 percent of annual demand on hand for me.' They are so naïve that they don't think that that cost is being passed on to them. They're not showing inventory on the balance sheet, but they guarantee that it's being booked into the cost of the product."

Inventory optimization is a core piece of Optiant's PowerChain 5.0 solutions suite. With it, Colehower says, Optiant can give clients what it calls "Efficient Frontier Analysis"-a representation of the tradeoff between business costs and service levels. "After evaluating a customer's supply chain based on optimizing inventory and service levels, we say, 'Here's where you could be; let's figure out the way to get you there.'

"Can they increase service without increasing costs? Can they maintain current service levels and take cost out of the business? Our goal is to get customers to that efficient frontier."

Demand planners often don't understand that every time they revise a forecast, there are upstream implications on the supply chain, says Sridhar Tayur, CEO of Pittsburgh-based SmartOps. Consequences include continued production of items no longer needed; mounting buffer stock because there is a history of forecasts being trashed; and costly shuttling of inventory from one location to another to meet urgent but unexpected demand.

Tayur says his company's Adaptive Structural Forecasting solution permits a much more detailed level of granularity in forecasts. Second, ASF helps pinpoint the reason a forecast has moved up or down. "In the past, there was no reason [given] for changing a forecast. ASF allows you to keep track of the structure and adapts by saying, 'You know, the last time you gave me a revised forecast because of this or that reason, you uplifted the demand by about 23 percent, but the actual was about 30 percent. I know you are uplifting it again by 23 percent, but I recommend you do it at about 30.'"

Holistic Supply Chains
Today's supply chains are growing longer and more complex all the time. SKU proliferation, offshore manufacturing, increased lead times, new-product introductions, and shorter lifecycles each contribute to the increasing complexity of today's supply chain, says Tarun Goyal, a manager in the Manugistics Solution Management Team. "Carrying more inventory is not an option to counter this variability," he says. Yet that's precisely what occurs absent a holistic view of the supply chain.

"Production planners, for example, want to get the maximum throughput from their machines," Goyal says. "They sometimes are very myopic and they don't understand that things they do to reduce production costs can cause inventory costs to go through the roof. The same thing with the demand forecasters. They sometimes are incented to make sure that all demand is met and they tend toward higher forecasts, not realizing how that's driving things."

Rockville, Md.-based Manugistics Group recently took a large step toward augmenting its holistic supply chain approach by partnering its transportation management system with the warehouse management system of MARC Global of Atlanta.

"From the perspective of a transportation manager, the ideal scenario would be somebody in the warehouse who accumulates everything, packs and holds it, and knows exactly what inventory they are going to ship and when," says Bob Jabjiniak, MARC Global senior vice president of product management. "They can do the best job of minimizing the transportation cost."

Jabjiniak sees trading partners deriving mutual benefit from another area-RFID-if only in the future, and not just in the warehouse. Once you look past the hype and hysteria over tags and readers, he says, one can see the potential in information sharing.

"We're starting to see some customers take advantage of the level of information sharing made possible with RFID, and it is what I would consider revolutionary."

Historically, there was feedback between retailers and suppliers (often a shortclaim) if an order was missing a case. "But you didn't know which case they didn't receive," he says. "You didn't know if you hadn't shipped that extra case or if it got lost at a distribution center point or if it was somewhere in their warehouse or ended up being shipped to another customer."

Having the level of detailed and "serialized" information that RFID promises is good, he says. It provides visibility and analytics, and it reduces claims in historically adversarial relationships, like that between retailers and suppliers.

"It's not about replacing a scan with an RFID reader," says Jabjiniak. "It's about information sharing and the analytics."

Total Costs
Sometimes it helps to get the perspective of someone outside your own specialty. Transplace might seem to fit that bill. The Plano, Tex.-based company was formed from the merger of the 3PL businesses of the six largest publicly traded truckload carriers. It specializes in transportation services, not inventory management. But Tom Sanderson, Transplace president and COO, feels no part of the supply can afford not to know about the expertise of its partners.

"We're very involved with inventory management, and when we work with a customer we like to go in upfront examining the total supply chain costs," says Sanderson. "We do a lot of supply chain network design, and we're examining those tradeoffs between inventory cost, transportation costs, and facility costs. We think that we can help our customers most if we focus on total supply chain costs consistent with their service objectives rather than approaching things just from the transportation component in isolation."

He feels inventory optimization issues are essentially divided into two parts: design/planning, then execution. The first deals with the optimum number of distribution centers and their locations. The task there is to align suppliers with DCs and then the DCs with the stores in order to minimize the aggregate transportation/inventory/facilities cost. "And that's a combination of science and art in coming up with the right strategy in laying that out," says Sanderson.

Then there's the execution area, and one of its biggest components is how much control a retailer has over managing inbound supply from its vendors. "Inbound is a significant challenge on the execution side for a lot of retailers, and it definitely has an impact on inventory levels at the DCs as well as on transportation costs."

Every supply chain is different, and so might be every distribution center strategy. When Transplace first began working with AutoZone, a major auto parts and accessories retailer, it already had the capital expenditure approved to double its distribution centers to accommodate growth. "Our first step was to do a network design," Sanderson says. "We showed that even with the projected volume growth and store growth, their then-current distribution network was already optimal and that adding facilities would just drive up inventory and facility costs. In fact, since that time, they've doubled the number of stores, but they've only added one DC."

Mobile Inventory
In a time of tightly constrained transportation capacity, Sanderson says, companies need to be aware that any inefficiencies in the warehouse or DC that hobbles the transportation leg is costing them a lot of money.

"Carriers are very focused on velocity, keeping the tractors, the drivers and the trailers moving," says Sanderson. "We do a lot of work with our shippers in terms of making sure the product isn't tying up trailer yards and making sure the docks are efficient for when the driver gets there for loading or unloading. There's still a lot of inventory out there that's not in a warehouse or in the back room of the store. It's on trailers, and that's only getting more expensive in a tight-capacity situation."

David W. Morgan would call that "fifth inventory," those assets that are not found within the warehouse or field service location, yet are inventory in every sense of the term, including carrying cost and potential liability.

Morgan, president and CEO of D.W. Morgan of Pleasanton, Calif., says that four things generally come to mind when inventory is mentioned: raw materials, work in progress, finished goods, and service inventory. Morgan says that basically all in-transit or pipeline goods form the fifth. Because title usually has passed to the receiving party, it's critical to treat in-transit items as inventory. The problem, he says, is that warehouse management and enterprise resource planning systems were designed to recognize only inventory within the four walls of a DC or other stocking location.

"So things are coming into to you, even though it's not yet on your ERP," Morgan says. "Technically, it's on your books because you purchased this thing. Title has passed from one entity to another. And Sarbanes-Oxley is requiring that companies begin looking at this." The reference is to the federal law on corporate governance that has mandated new accounting procedures and documentation.

Morgan says his company's ChainLinq solutions set addresses the issue. "It brings all the various databases of the fifth inventory into a single user interface for complete visibility."
Morgan says other examples of fifth inventory include returns of any kind, and items or substances that manufacturers are obliged to dispose of under any number of environmental regulations.

"This is important once you accept that just because it's not within your four walls, you still own it and have to comply with certain mandates and global requirements."

But why should inventory ever be within the confines of one's facility? Wouldn't the best inventory be ... no inventory? It isn't entirely a fantasy, though clearly it doesn't work for every vertical or for every product line. Ross Elliott, vice president of product strategy for Infor Distribution Essentials, Alpharetta, Ga., says some of his customers have managed to take a mixed approach: keeping some inventory on hand, while taking orders for other products and shipping them without ever touching them.

An Infor customer in Toronto has more than 500,000 products in its system, Elliott says. "They ship 40 percent of their orders with direct shipments from manufacturers. The products never touch their logistical capacity. They simply take an order from a customer and ship it directly from the original manufacturer of the component to final destination."

But as with any other inventory holder, Elliott says, the company occasionally has problems within the 60 percent that it does stock.

"Our approach with all clients has been to build inventory management and warehouse management systems to work together, to try to optimize the purchasing side of the house, the logistics side of the house and the location side of the house."

That requires identifying the logical places to store inventory based on demand, logistics and, in some cases, customers' preferences. Customers, of course, fall into either the forecast-driven or historical-usage categories, so those patterns must be analyzed.

How effective is vendor-managed inventory? It depends on the vertical, and it certainly requires a reliable supplier, Elliott says. He says some customers in the electrical supplies industry have attained the goal of balancing inventory costs with service objectives with VMI. But the industry at large traditionally is prey to unpredictable demand and has been notoriously overstocked. Pity the poor inventory manager caught between unreliable suppliers on the one side and salesmen on the other who heedlessly load up the warehouse. That, however, seems to be changing, says Elliott. In the last two years, the industry as a whole has seen its service levels rise while average inventory has gone down.
Exactly what StorageTek sought in its vertical. Exactly what any company wants from an inventory optimization initiative.<//strong>

 

Business for Storage Technology Corporation is about helping an extremely diverse client base keep on top of their business-critical data storage needs. It often means making sure that needed parts are delivered for service to a client on time; more often than not within two hours of the customer placing a service request. That isn't so easy to do if you aren't on top of your own storage-your inventory of printed circuit boards, disk drive assemblies and myriad other service parts. But Storage Technology, or StorageTek as it is more commonly known, has that part down now.

For 20 years or more StorageTek has had some kind of automated software to help manage its service parts inventory. Initially, the software was an internal creation. Ten years ago, StorageTek went to an outside developer for a system, but by 2003 it felt its needs had outstripped the capabilities of that product. What the Louisville, Colo.-based company felt it needed was a system that could not only disseminate order entries immediately but determine how much inventory it needed to stock based on customers' contractual service requirements and how many distribution centers it needed to maintain overall. Like any manufacturer and retailer worldwide, regardless of vertical or product line, it needed an optimized inventory approach.

How much to keep, and where to keep it are basic questions to address in implementing any inventory optimization policy. But if any part of a company can ill afford to act on its own, it's those tasked with managing inventory. They often are stuck in between production and demand planners on one side and the transportation department on the other. Depending on the nature of the business, inventory managers must juggle inbound and outbound, balance safety stocks and service-level objectives, make sense of forecasts that often are more exercises in wishful thinking than anything else, and, oh yes, keep upper management happy by keeping a lid on costs. Seldom can this be done without some outside consultation and expertise.

StorageTek was acutely aware of what it wanted. It had customers in 63 countries, but its systems weren't keeping order information around the world as timely as desired. "The existing software wasn't following the sun," says James Matney, the company's director of global service logistics.

StorageTek also wanted to import contract database intelligence into the service parts planning system so response-time guarantees that customers were paying for could be honored. But it also needed to know how many parts to stock for those clients. Previous generations of software weren't very smart; rather, they just looked at past usage history.

Finally, StorageTek needed an application that would "graphically display where are the best places to put stocking locations and how many of these stocking locations you should have," Matney says.

The information storage solutions provider turned to Servigistics of Atlanta to provide the storage parts planning and distribution software.

Whether they call it modeling, network design or by some other name, companies like Servigistics make inventory optimization suggestions based on an analysis of the location of the client's customers and its current DCs, and the nature of the obligations mandated by service-level agreements the client has signed. The assessment attempts to determine not only how many parts should be kept on hand but where.

"Not all of them have to be at field service locations," says Cliff Isaacson, director of product marketing for Servigistics. Pointing to Avaya, the communications systems giant, he says the cost of a given part may be a factor in the analysis. "Some of its parts are worth a hundred or two hundred thousand dollars, because they are very high-end network and telecommunication equipment. You obviously don't want to stock that part at every single service location. If there's an opportunity to pool a part, meaning having a single part in one place that supports multiple service locations, we [look] for that."

While Matney runs the logistics business at StorageTek, the Servigistics software "runs where all the spare parts need to be, how many parts are being used, calculating desired stocking amounts, and then every night it's looking for changes," he says. "If a part's been used, and it drops below a certain level, it generates an order and sends it up to our ERP system to either move a part from a distribution center, move a part from another depot, or for us to go procure a part. It's doing all the logic and creating the transactional orders."

Follow the Money
In Isaacson's view, the "glamour" part of the supply chain has always been in enterprise software. The service supply chain was an afterthought, he says, until management realized the huge gross margins that parts management and sales could bring. Every time a new primary good was introduced-a car or a cellphone, for example-a whole new set of service parts was introduced as well. That brought added opportunities for aftermarket sales, but it also caused a huge proliferation in SKUs of both new and now-obsolete parts.

"In many cases companies know the problem is there," Isaacson says. "It's just that their focus and attention on technology has been on the front end of the supply chain. And it's only been in the last few years that there have been technologies and computing power strong enough to consider all the different service parts. One thing that's characteristic of service parts: If you think of the limited set of primary goods that companies are selling, there are millions of service parts that are associated with repairing them or replacing them. So the volume of parts and parts locations is a really big problem to solve."

Though there is a lot of money to be saved in DC consolidation, customers are generally cautious because of the risk in negatively affecting customer service. "The bottom line is you consistently want to hit those service levels, especially if you believe you drive greater customer growth by having better service than your competition," Isaacson says.

Trading partner separations or disconnects can occur anywhere in the supply chain, and the more extended it is, perhaps the greater the risk. "Inventory typically is the result of poor information between planners and the warehouse," says Jim Erdman, Penske Logistics' senior vice president for the Asia Pacific region. "Here in China, it's probably the biggest single problem. There's just not a lot of good information so everybody tends to have a lot of stock."

Creating Obsolescence
It's particularly acute in the electronics industry, which has a large presence in the country, but occurs in other sectors as well, Erdman says. Information gaps create obsolescence. "Electronics manufacturers still build products that have become obsolete. They don't want to have too much inventory, but they need enough to meet the demand."

Erdman said one industrial machine manufacturer, who sourced half or more of his product's components from outside of China, had a terrible overstocking problem due to information gaps. Data transfer was so poor, even from a parts supplier in Taiwan, that the manufacturer kept six to nine months' of inventory on hand.

Compounding the problem is the length of the supply chain, Erdman says. "Somebody once wrote that the 'tyranny of China is its locationn' because it's so far away from everything. It's true. And with the growth in Asia, the congestion at the ports, the capacity restraints in North America and even in Europe, it's just getting further away. If you work on something with, say, a 36-day cycle, there's no guarantee that you're going to get 36 days. It may be 40 days because there's a tie-up in Long Beach. So everybody overcompensates for those schedules by carrying more inventory than they need. If it's supposed to be a 36-day cycle, but a lot of times it's been 40, they'll probably keep 50 days' or 60 days' worth just to make sure."

The Chinese have made tremendous strides, Erdman says, "but there is still a awful lot done on gut feeling and not on information. They haven't invested enough in the technology side to get that information."

It's not a situation at all alien to Charles Smart, president of Belmont, Mass.-based Smart Software. His company's SmartForecasts product essentially analyzes demand or sales data for a product item, considers any intermittent demand patterns or seasonality factors, then determines the optimal forecasting method based on the appropriate lead time for that product.

"But in that process, we also have to analyze and project what the expected deviation from the forecast might be," says Smart. "And that information is the basis that we use to create estimates of safety stock parameters to add to the demand forecast-to get the inventory levels to correspond to the service levels that a client might desire."

Often this analysis is missing when customers forecast on their own, Smart says. "When they get down to taking the demand forecast and adding something to it, they don't use much analysis. They use a lot of-almost folklore-in creating the safety stock. 'Well, we've always doubled the demand forecast to get the buffer stock. Or if the lead time is always 3 weeks, we've always added 6 weeks of expected demand to get the safety stock.'

"We still find a lot of companies locked into that sort of ad hoc approach to inventory planning and stocking." It's a huge waste of money and generally fails to achieve service-level objects, Smart says.

Demand is what drives inventory, Erdman says. "Good demand forecasting is the most important thing in controlling inventory."

Jonathan Colehower, CEO of Optiant of Burlington, Mass., might say that the most important thing in keeping a lid on inventory is staying on top of variability in a supply chain that's becoming "unbundled."

The typically vertically integrated company that sources, manufactures and distributes on its own is no more for the most part. "You may have one organization do the sourcing for you, you have a contract manufacturer doing some of the base manufacturing, another person doing the subassembly, then you may have a 3PL doing the warehousing and distribution. Through this unbundling, we're seeing two things: risk and the need for data synchronization."

Of course, inventory is impacted because variability is introduced with every step in the supply chain, Colehower says. "'Am I going to get it in two days? Or am I going to get it in a week?'

"We need inventory for two reasons: one is demand variability. How often am I going to be demanding of it? Second is supply variability, or lead time variability.

"Any time you introduce variability into the supply chain, the natural response from any company is to buffer it with inventory."

Whose Inventory?
The solution should not be to push inventory around the supply base like billiard balls on a table, Colehower says. "I've witnessed companies write into their suppliers' contracts, 'You must keep 25 percent of annual demand on hand for me.' They are so naïve that they don't think that that cost is being passed on to them. They're not showing inventory on the balance sheet, but they guarantee that it's being booked into the cost of the product."

Inventory optimization is a core piece of Optiant's PowerChain 5.0 solutions suite. With it, Colehower says, Optiant can give clients what it calls "Efficient Frontier Analysis"-a representation of the tradeoff between business costs and service levels. "After evaluating a customer's supply chain based on optimizing inventory and service levels, we say, 'Here's where you could be; let's figure out the way to get you there.'

"Can they increase service without increasing costs? Can they maintain current service levels and take cost out of the business? Our goal is to get customers to that efficient frontier."

Demand planners often don't understand that every time they revise a forecast, there are upstream implications on the supply chain, says Sridhar Tayur, CEO of Pittsburgh-based SmartOps. Consequences include continued production of items no longer needed; mounting buffer stock because there is a history of forecasts being trashed; and costly shuttling of inventory from one location to another to meet urgent but unexpected demand.

Tayur says his company's Adaptive Structural Forecasting solution permits a much more detailed level of granularity in forecasts. Second, ASF helps pinpoint the reason a forecast has moved up or down. "In the past, there was no reason [given] for changing a forecast. ASF allows you to keep track of the structure and adapts by saying, 'You know, the last time you gave me a revised forecast because of this or that reason, you uplifted the demand by about 23 percent, but the actual was about 30 percent. I know you are uplifting it again by 23 percent, but I recommend you do it at about 30.'"

Holistic Supply Chains
Today's supply chains are growing longer and more complex all the time. SKU proliferation, offshore manufacturing, increased lead times, new-product introductions, and shorter lifecycles each contribute to the increasing complexity of today's supply chain, says Tarun Goyal, a manager in the Manugistics Solution Management Team. "Carrying more inventory is not an option to counter this variability," he says. Yet that's precisely what occurs absent a holistic view of the supply chain.

"Production planners, for example, want to get the maximum throughput from their machines," Goyal says. "They sometimes are very myopic and they don't understand that things they do to reduce production costs can cause inventory costs to go through the roof. The same thing with the demand forecasters. They sometimes are incented to make sure that all demand is met and they tend toward higher forecasts, not realizing how that's driving things."

Rockville, Md.-based Manugistics Group recently took a large step toward augmenting its holistic supply chain approach by partnering its transportation management system with the warehouse management system of MARC Global of Atlanta.

"From the perspective of a transportation manager, the ideal scenario would be somebody in the warehouse who accumulates everything, packs and holds it, and knows exactly what inventory they are going to ship and when," says Bob Jabjiniak, MARC Global senior vice president of product management. "They can do the best job of minimizing the transportation cost."

Jabjiniak sees trading partners deriving mutual benefit from another area-RFID-if only in the future, and not just in the warehouse. Once you look past the hype and hysteria over tags and readers, he says, one can see the potential in information sharing.

"We're starting to see some customers take advantage of the level of information sharing made possible with RFID, and it is what I would consider revolutionary."

Historically, there was feedback between retailers and suppliers (often a shortclaim) if an order was missing a case. "But you didn't know which case they didn't receive," he says. "You didn't know if you hadn't shipped that extra case or if it got lost at a distribution center point or if it was somewhere in their warehouse or ended up being shipped to another customer."

Having the level of detailed and "serialized" information that RFID promises is good, he says. It provides visibility and analytics, and it reduces claims in historically adversarial relationships, like that between retailers and suppliers.

"It's not about replacing a scan with an RFID reader," says Jabjiniak. "It's about information sharing and the analytics."

Total Costs
Sometimes it helps to get the perspective of someone outside your own specialty. Transplace might seem to fit that bill. The Plano, Tex.-based company was formed from the merger of the 3PL businesses of the six largest publicly traded truckload carriers. It specializes in transportation services, not inventory management. But Tom Sanderson, Transplace president and COO, feels no part of the supply can afford not to know about the expertise of its partners.

"We're very involved with inventory management, and when we work with a customer we like to go in upfront examining the total supply chain costs," says Sanderson. "We do a lot of supply chain network design, and we're examining those tradeoffs between inventory cost, transportation costs, and facility costs. We think that we can help our customers most if we focus on total supply chain costs consistent with their service objectives rather than approaching things just from the transportation component in isolation."

He feels inventory optimization issues are essentially divided into two parts: design/planning, then execution. The first deals with the optimum number of distribution centers and their locations. The task there is to align suppliers with DCs and then the DCs with the stores in order to minimize the aggregate transportation/inventory/facilities cost. "And that's a combination of science and art in coming up with the right strategy in laying that out," says Sanderson.

Then there's the execution area, and one of its biggest components is how much control a retailer has over managing inbound supply from its vendors. "Inbound is a significant challenge on the execution side for a lot of retailers, and it definitely has an impact on inventory levels at the DCs as well as on transportation costs."

Every supply chain is different, and so might be every distribution center strategy. When Transplace first began working with AutoZone, a major auto parts and accessories retailer, it already had the capital expenditure approved to double its distribution centers to accommodate growth. "Our first step was to do a network design," Sanderson says. "We showed that even with the projected volume growth and store growth, their then-current distribution network was already optimal and that adding facilities would just drive up inventory and facility costs. In fact, since that time, they've doubled the number of stores, but they've only added one DC."

Mobile Inventory
In a time of tightly constrained transportation capacity, Sanderson says, companies need to be aware that any inefficiencies in the warehouse or DC that hobbles the transportation leg is costing them a lot of money.

"Carriers are very focused on velocity, keeping the tractors, the drivers and the trailers moving," says Sanderson. "We do a lot of work with our shippers in terms of making sure the product isn't tying up trailer yards and making sure the docks are efficient for when the driver gets there for loading or unloading. There's still a lot of inventory out there that's not in a warehouse or in the back room of the store. It's on trailers, and that's only getting more expensive in a tight-capacity situation."

David W. Morgan would call that "fifth inventory," those assets that are not found within the warehouse or field service location, yet are inventory in every sense of the term, including carrying cost and potential liability.

Morgan, president and CEO of D.W. Morgan of Pleasanton, Calif., says that four things generally come to mind when inventory is mentioned: raw materials, work in progress, finished goods, and service inventory. Morgan says that basically all in-transit or pipeline goods form the fifth. Because title usually has passed to the receiving party, it's critical to treat in-transit items as inventory. The problem, he says, is that warehouse management and enterprise resource planning systems were designed to recognize only inventory within the four walls of a DC or other stocking location.

"So things are coming into to you, even though it's not yet on your ERP," Morgan says. "Technically, it's on your books because you purchased this thing. Title has passed from one entity to another. And Sarbanes-Oxley is requiring that companies begin looking at this." The reference is to the federal law on corporate governance that has mandated new accounting procedures and documentation.

Morgan says his company's ChainLinq solutions set addresses the issue. "It brings all the various databases of the fifth inventory into a single user interface for complete visibility."
Morgan says other examples of fifth inventory include returns of any kind, and items or substances that manufacturers are obliged to dispose of under any number of environmental regulations.

"This is important once you accept that just because it's not within your four walls, you still own it and have to comply with certain mandates and global requirements."

But why should inventory ever be within the confines of one's facility? Wouldn't the best inventory be ... no inventory? It isn't entirely a fantasy, though clearly it doesn't work for every vertical or for every product line. Ross Elliott, vice president of product strategy for Infor Distribution Essentials, Alpharetta, Ga., says some of his customers have managed to take a mixed approach: keeping some inventory on hand, while taking orders for other products and shipping them without ever touching them.

An Infor customer in Toronto has more than 500,000 products in its system, Elliott says. "They ship 40 percent of their orders with direct shipments from manufacturers. The products never touch their logistical capacity. They simply take an order from a customer and ship it directly from the original manufacturer of the component to final destination."

But as with any other inventory holder, Elliott says, the company occasionally has problems within the 60 percent that it does stock.

"Our approach with all clients has been to build inventory management and warehouse management systems to work together, to try to optimize the purchasing side of the house, the logistics side of the house and the location side of the house."

That requires identifying the logical places to store inventory based on demand, logistics and, in some cases, customers' preferences. Customers, of course, fall into either the forecast-driven or historical-usage categories, so those patterns must be analyzed.

How effective is vendor-managed inventory? It depends on the vertical, and it certainly requires a reliable supplier, Elliott says. He says some customers in the electrical supplies industry have attained the goal of balancing inventory costs with service objectives with VMI. But the industry at large traditionally is prey to unpredictable demand and has been notoriously overstocked. Pity the poor inventory manager caught between unreliable suppliers on the one side and salesmen on the other who heedlessly load up the warehouse. That, however, seems to be changing, says Elliott. In the last two years, the industry as a whole has seen its service levels rise while average inventory has gone down.
Exactly what StorageTek sought in its vertical. Exactly what any company wants from an inventory optimization initiative.<//strong>