Executive Briefings

Advanced Order Fulfillment Requires Warehouses With 'On Demand' Capability

Increased offshore manufacturing coupled with growing demands for more customized products and rapid delivery require a more flexible order-fulfillment model. Supply-chain strategists and solution providers are responding to this challenge.

The old order-fulfillment model of pick, pack and ship - even when done flawlessly and at top speed - is no longer sufficient for most companies. Holding the necessary amounts of finished-goods inventory to maintain superior customer service levels with this model is simply too costly. And it's too clunky. Today's dynamic supply chains, with their complex mix of global sourcing, multiple channels, customized orders and rapid delivery, require more adaptive solutions.

AMR Research, Boston, calls the new model Advanced Order Fulfillment. "Much as warehouses in the 1980s and 1990s evolved from traditional stocking warehouses into distribution centers that supported cross-docking, drop-shipping and more complex packaging and labeling, customer-facing warehouses are increasingly used to support more complex fulfillment activities, including postponement activities," says Larry Lapide, vice president of research at AMR.

Another take on this trend comes from Forte Industries, Cleveland, which provides consulting and technology solutions for warehouse management. Forte recently launched a web-based initiative called Distribution on Demand (www.distributionondemand.com), which it defines as "the order fulfillment state an organization achieves when it can respond close to real time to changes in demand, while shipping 100 percent customer-compliant orders at the least cost."

"There is no question that distribution has entered an on-demand environment," says Louie Hollmeyer, director of marketing improvement at Forte. At the same time, he says, manufacturing is moving offshore, which puts even more pressure on the distribution and fulfillment side. "Basically, the distribution-on-demand initiative is a call to arms for the distribution segment to seize this opportunity and promote distribution as a frontline business strategy," he says. "Companies can make a great product, manufacture it at the least cost and sell it extremely well, but if they can't deliver it to the right place, at the right time for the right price, then that effort is all for naught."

By whatever name, this more demanding fulfillment model is enabled by robust and well integrated order management, warehouse management and transportation management systems, as well as by inventory optimization technology. And since the goal is to keep the customer happy, Advanced Order Fulfillment also includes metrics designed to look beyond silo performance to overall customer satisfaction.

Moving to this model involves many parts of an organization and requires considerable systems integration, says Lapide. "Ultimately, the warehouse management system will be the key component in a larger, make-to-order fulfillment application ecosystem."

Leading WMS vendors that will help companies make this transition recognize that it is about more than new functionality. "Applications have to move from being data-centric to being process-centric and from being based on rigid business rules to being highly flexible," says Rob Sweeney, vice president of product management at Yantra, an order management and supply-chain execution software provider based in Tewksbury, Mass. "The idea is to define what business process is required to satisfy a customer's demand and then adapt the software or technology component to that business process, rather than trying to mold the business process to how the technology works." Typically, he says, this is where companies hit a brick wall because "infrastructure tends to harden in place over time."

Yantra meets this challenge by building its applications on a component-based architecture. "Each little piece of business logic in the software is like a service," says Sweeney. "You link together different services to complete a process."

"We have built the same kind of algorithms used in manufacturing into our warehouse management system."
- Paul Crist of Provia

Processes change with different service and product mixes, he explains, and applications that are process driven are able to adapt to these changes. For example, he notes that a warehouse behaves very differently when a product is fulfilled through a build-to-order or configure-to-order postponement strategy as opposed to being picked from stock. With the former, the warehouse becomes a value-added service operation, he says, "so you have to have strong work-order capability and the ability to manage unique items through your supply chain."

Enablement of postponement or delayed differentiation is an important part of Advanced Order Fulfillment because companies increasingly are moving toward this strategy. The demand for greater customization of products and services is one driver. Another is the challenge posed by extended supply chains that must synchronize the long lead times inherent in offshore manufacturing with highly uncertain demand.

"Managing supply chains when you have more and more outsourcing and a high degree of customization is a very complex problem to solve, particularly as customers are unwilling to pay more or wait longer," says Nadeem Syed, vice president of advanced planning applications development at ERP vendor Oracle, Redwood Shores, Calif. "Postponement is one way to do this."

A recent survey conducted by Oracle and Cap, Gemini, Ernst & Young found that 75 percent of responding companies that had implemented postponement considered their implementation a success. Ninety-one percent noted significant improvements in customer satisfaction and inventory costs, and a large majority said they achieved significantly improved order-fill rates with decreased lead times.

The primary reason given by surveyed companies for not considering a postponement strategy was a general lack of understanding of the concept. Additional key inhibitors, according to the report, are: an inability to recognize where postponement is most effective; an inability to quantify benefits; and a belief that technology does not support implementation.

These issues largely have been resolved, Syed says. "What has been lacking in the environment until recently, but what is available now, are systems and processes that help companies approach a postponement strategy in a systemic way," he says. "In the last five years, technology has advanced significantly and we have lots of optimization capabilities and a level of computing power that allows us to solve these complex problems in a reasonable amount of time."

Blurring the Line
Postponement essentially requires turning a DC into a light manufacturing or assembly operation. "Our customers are asking for more and more features that blur the line between distribution and manufacturing," says Chris Heim, president of HighJump Software, a 3M company based in Eden Prairie, Minn. "DCs can be asked to do anything from tagging to final assembly to personalizing a product for a particular company, so in some cases they are almost a mini-factory."

"A key capability to support the transformation or conversion of product, whether raw materials or components, into a new SKU, is work-order processing," says Jim Stollberg, vice president and general manager at Irista, a supply-chain execution provider based in Milwaukee. Conversions also could involve taking a finished goods product that is stored in bulk and packaging it into different case or carton configurations, he says. "Again, it is all about being able to differentiate and customize that product in the last leg before it goes out the door. This is what gives you the flexibility to adapt to changes in demand."

Postponement has been used by the high-tech industry for some time, "but now we are seeing it in all the markets we serve, including our third-party logistics clients who want to offer this type of service to their customers," says Paul Crist, vice president of sales and marketing at Provia Software, Grand Rapids, Mich. "People want unique products, configured to their needs, right now," he says, "and that is the strength of our kitting solution."

As an example, Crist notes that cellular phone companies sell the same products to different mass retailers, but they customize these products by configuring and packaging them differently. "So, rather than building up inventory specific to all those retailers, they store components and build to order," he says. When the real orders come in, the fulfillment center takes a generic phone and sends it down a kitting line in the DC for that particular retailer, each with unique paperwork and unique handling requirements and unique configurations and packaging."

RedPrairie, an execution software provider based in Waukesha, Wis., has enhanced its postponement capabilities by moving the kits themselves to an on-demand model. "Manufacturers historically have tried to meet order demand by creating kits to complete generic products, but then they put the kits in a warehouse and wait for the orders to arrive," says Tom Kozenski, value delivery leader.

RedPrairie's solution allows companies to store information about the components needed for each kit, and then automatically list the components to be picked as part of the order-fulfillment process, he says. The kit is created on demand and not by forecast. "This helps because order processing doesn't have to see that kit in inventory in order for it to release an order," he says. "They can release the components and let the warehouse management system figure out how to put it together."

Stollberg notes that fulfillment customization also extends to delivery and transportation services, a fact driven, in part, by customer segmentation. A mass merchandiser like Wal-Mart may have very specific instructions about how it wants things shipped and delivered, he says, while other customers simply want the lowest cost mode that meets their service need. "To support the wide difference in customer requirements, warehousing and transportation need to be tightly coupled," he says. "Just interfacing WMS to TMS doesn't cut it in today's world. The two systems have to behave as one."

Short Planning Windows
Value-added delivery services also play a role in Advanced Order Fulfillment. "A lot of our customers are no long selling products, but products bundled with a high level of services," says Sweeney. This adds another level of complexity. "In addition to having inventory you want to move rapidly, you also want to make sure you are effectively using all your service capacity," he says. "If you are paying someone to install a big-screen TV and there is no work for them to do, that is the same as inventory sitting on the shelf."

These challenges are exacerbated by very short planning windows in today's supply chains. "If you think back five or 10 years ago, distribution centers had a nice window of time to plan their daily operations," says Stollberg. "Managers could take a day or two to consolidate orders and group them into nice waves. They don't have that luxury anymore. Now, if an order is in by noon, it usually is expected to ship the same day, so we have become much more dependent on real-time systems. We have to react and respond a lot faster."

Fingerhut, a general merchandise mail-order retailer, provides a good example of this time compression. Before installing new fulfillment software from HighJump, Fingerhut operated on a batch process bringing orders into the warehouse once a day. "Now we bring orders in once an hour," says Loren Eggert, vice president of distribution and fulfillment. "Typically orders go out in 24 to 36 hours." This timeframe gives the company the flexibility of shipping most orders the same day, while managing the workload so that there is enough in the bucket to start the following morning efficiently. "Overall, we have cut two days out of our delivery time," he says. This not only has taken Fingerhut's service "to a new level," he says, but has had a "very positive impact" on the rate of returns.

Lamps Plus is another retailer that accelerated its order turns after implementing fulfillment software from Escalate Fulfillment, Redwood Shores, Calif. This retailer's internet store has become its largest, with more than 500 orders per day. Orders that used to ship in five to seven days now are out the door in 24 to 48 hours, according to Bill Gratke, director of planning and supply-chain management. One key part of the Escalate solution was a packing optimization engine that enables Lamps Plus to easily consolidate multi-line orders into a smaller number of packages. "The strongest thing to say about what Escalate did for us is that they listened to us and molded their software into what we could use," Gratke says.

The length of supply chains when products are coming from other parts of the world also challenge fulfillment managers. "There is a whole new move to make sure, on the front end, that orders are filled accurately," says Mark Holmes, a consultant with Avicon, Waltham, Mass. "I call this advanced order fulfillment accuracy. This concept uses event management to red flag orders when the purchase order and loading manifest do not match, he explains, so that resolution can occur before the shipment ever leaves port. "It is very costly to wait until something gets to the DC to discover it is not as ordered," he says.

Optimization
Improved optimization technology also is helping companies enable Advanced Order Fulfillment. For example, says Kozenski, knowing which warehouse to ship from is an important consideration when stock is held or produced in multiple locations. "Some warehouses may not have an item in stock, some may be too far away and transportation costs too high, or some may ship direct from one of several vendors, so which is the optimal location to ship that order or line item from?" RedPrairie's SourceRight solution provides that calculation.

Being able to allocate inventory in transit also is crucial, says Lars-Goran Olsson, vice president of product management of Industri-Matematik International, Mt. Laurel, N.J. "One of the trends we see is that a large number of products are not stocked in quantity in the warehouse, so companies need to be able to act on the inventory that is in motion. This goes back to needing a closer relationship between different processes."

Execution optimization is "the natural extension of the warehouse becoming an on-demand operation," says Crist. Provia borrowed its approach to optimization from the manufacturing sector and its use of constraint-based modeling. "Warehouses also have constraints - certain processes take so long to do, equipment constraints, labor constraints, inventory constraints, transportation constraints, value-added service constraints, and so on. We have built the same kind of algorithms used in manufacturing into our warehouse management system, so we can model those constraints. You run a group of orders through - for the next hour or the next day, whatever time period - and it will churn through all the information and release a sequence of orders for processing that results in the maximum throughput. As things change, the optimization can be re-run in the background."

For example, he says, if a truck that was expected to come in at 11:00 in the morning with parts needed to perform a kitting function doesn't arrive, "We can rerun the optimization and move those orders to later in the day and fill that time quickly with orders that can be processed."

Jay Baitler, executive vice president of Staples Contract Division, recently installed the latest upgrade to its PkMS warehouse management solution from Manhattan Associates, Atlanta. "This software has a unique predictive capability that really intrigues me," he says. "It used to be that you would walk around a warehouse and if you saw a bottleneck building up you would throw labor at it. The first line of defense was a good, smart supervisor who could see that bottleneck developing in receiving or sortation or wherever. This installation of PkMS we are now doing has predictive intelligence so that it can tell you, based on the activity it is seeing, that a backlog will develop in this area by 2 p.m. today and you can afford to transition some labor from this place to that place to smooth it out over the balance of the day.

"I am really excited about that," he says. "It shows how far we have come in realizing this quest for operational excellence. Of course, we have to keep kicking that can to make it even better tomorrow, but this is pretty exciting stuff."

How Not to Deceive Yourself About the Perfect Order
How to best measure the success of fulfillment efforts is the subject of continuing debate. Kate Vitasek, managing partner at Supply Chain Visions, Bellevue, Wash., does a lot of consulting and training work in this area.

"It is amazing how many companies will tell you they have a fill rate of 99 percent," she says, "but then when you really start drilling down, you find that the way they got to 99 percent was by expediting a lot of shipments." Another common tactic companies use is to drop an order to the warehouse only when they know the product is in stock. "I go into companies all the time that measure their fill rate based on when an order is dropped in the DC, not when they received the order," she says. "The order could have set around for three days waiting for the inventory to come in or it could have been on credit hold or whatever. The customer wanted it yesterday and it hasn't even shipped yet, but that is being counted as a success in terms or their fulfillment metrics."

To avoid this and other "order fill traps" Vitasek recommends that companies use the Perfect Order Index (POI) to understand how they are performing and where they need to improve. "A perfect order has typically been defined as on time, complete, damage free and having the correct invoice," she says, though there are variations on this theme. Using POI, performance is measured by multiplying the metrics for each of these to each other. "For example," she says, "if a firm is experiencing 95 percent on-time delivery, fill rate, correct invoice and damage-free shipments, the resulting perfect order index would be 81.4 percent. This explains why many companies think they have good performance when they actually are not meeting customer expectations - or are cultivating behaviors that drive costs up."

When Jay Baitler, executive vice president of Staples Contract Division, understood the impact of the multiplier effect, he decided to no longer rate success based on individual components of Staples' version of the perfect order. "We rate ourselves very, very, very hard," he says. For example, if a product is out of stock but can be obtained from a supplier and still delivered on time and to the customer's expectation, "we still ping ourselves for that as a failure because we didn't have it in inventory," says Baitler. "We work very hard to make sure that our recovery in such situations is excellent, but we measure ourselves so as to make sure that those performance numbers get better every single day. The quest for that perfect order is never over."

To effectively implement a perfect order approach, companies need to change their incentives, says Vitasek. "Instead of measuring people just on what is in their control, they need to also be measured on the bigger picture, the overall outcome."

Steve Banker of ARC Advisory Group, Dedham, Mass., agrees. To really address the cross-functional nature of the perfect order, incentives have to be designed so that at least part of the bonus for people in purchasing depends on how well the people do in manufacturing or in the warehouse, and vice versa, he says. Otherwise, "the natural tendency of people is to define metrics from their own internal point of view," he says. "So you end up with a series of departmental metrics that, when you add them up, don't really mean much to the customer."

The old order-fulfillment model of pick, pack and ship - even when done flawlessly and at top speed - is no longer sufficient for most companies. Holding the necessary amounts of finished-goods inventory to maintain superior customer service levels with this model is simply too costly. And it's too clunky. Today's dynamic supply chains, with their complex mix of global sourcing, multiple channels, customized orders and rapid delivery, require more adaptive solutions.

AMR Research, Boston, calls the new model Advanced Order Fulfillment. "Much as warehouses in the 1980s and 1990s evolved from traditional stocking warehouses into distribution centers that supported cross-docking, drop-shipping and more complex packaging and labeling, customer-facing warehouses are increasingly used to support more complex fulfillment activities, including postponement activities," says Larry Lapide, vice president of research at AMR.

Another take on this trend comes from Forte Industries, Cleveland, which provides consulting and technology solutions for warehouse management. Forte recently launched a web-based initiative called Distribution on Demand (www.distributionondemand.com), which it defines as "the order fulfillment state an organization achieves when it can respond close to real time to changes in demand, while shipping 100 percent customer-compliant orders at the least cost."

"There is no question that distribution has entered an on-demand environment," says Louie Hollmeyer, director of marketing improvement at Forte. At the same time, he says, manufacturing is moving offshore, which puts even more pressure on the distribution and fulfillment side. "Basically, the distribution-on-demand initiative is a call to arms for the distribution segment to seize this opportunity and promote distribution as a frontline business strategy," he says. "Companies can make a great product, manufacture it at the least cost and sell it extremely well, but if they can't deliver it to the right place, at the right time for the right price, then that effort is all for naught."

By whatever name, this more demanding fulfillment model is enabled by robust and well integrated order management, warehouse management and transportation management systems, as well as by inventory optimization technology. And since the goal is to keep the customer happy, Advanced Order Fulfillment also includes metrics designed to look beyond silo performance to overall customer satisfaction.

Moving to this model involves many parts of an organization and requires considerable systems integration, says Lapide. "Ultimately, the warehouse management system will be the key component in a larger, make-to-order fulfillment application ecosystem."

Leading WMS vendors that will help companies make this transition recognize that it is about more than new functionality. "Applications have to move from being data-centric to being process-centric and from being based on rigid business rules to being highly flexible," says Rob Sweeney, vice president of product management at Yantra, an order management and supply-chain execution software provider based in Tewksbury, Mass. "The idea is to define what business process is required to satisfy a customer's demand and then adapt the software or technology component to that business process, rather than trying to mold the business process to how the technology works." Typically, he says, this is where companies hit a brick wall because "infrastructure tends to harden in place over time."

Yantra meets this challenge by building its applications on a component-based architecture. "Each little piece of business logic in the software is like a service," says Sweeney. "You link together different services to complete a process."

"We have built the same kind of algorithms used in manufacturing into our warehouse management system."
- Paul Crist of Provia

Processes change with different service and product mixes, he explains, and applications that are process driven are able to adapt to these changes. For example, he notes that a warehouse behaves very differently when a product is fulfilled through a build-to-order or configure-to-order postponement strategy as opposed to being picked from stock. With the former, the warehouse becomes a value-added service operation, he says, "so you have to have strong work-order capability and the ability to manage unique items through your supply chain."

Enablement of postponement or delayed differentiation is an important part of Advanced Order Fulfillment because companies increasingly are moving toward this strategy. The demand for greater customization of products and services is one driver. Another is the challenge posed by extended supply chains that must synchronize the long lead times inherent in offshore manufacturing with highly uncertain demand.

"Managing supply chains when you have more and more outsourcing and a high degree of customization is a very complex problem to solve, particularly as customers are unwilling to pay more or wait longer," says Nadeem Syed, vice president of advanced planning applications development at ERP vendor Oracle, Redwood Shores, Calif. "Postponement is one way to do this."

A recent survey conducted by Oracle and Cap, Gemini, Ernst & Young found that 75 percent of responding companies that had implemented postponement considered their implementation a success. Ninety-one percent noted significant improvements in customer satisfaction and inventory costs, and a large majority said they achieved significantly improved order-fill rates with decreased lead times.

The primary reason given by surveyed companies for not considering a postponement strategy was a general lack of understanding of the concept. Additional key inhibitors, according to the report, are: an inability to recognize where postponement is most effective; an inability to quantify benefits; and a belief that technology does not support implementation.

These issues largely have been resolved, Syed says. "What has been lacking in the environment until recently, but what is available now, are systems and processes that help companies approach a postponement strategy in a systemic way," he says. "In the last five years, technology has advanced significantly and we have lots of optimization capabilities and a level of computing power that allows us to solve these complex problems in a reasonable amount of time."

Blurring the Line
Postponement essentially requires turning a DC into a light manufacturing or assembly operation. "Our customers are asking for more and more features that blur the line between distribution and manufacturing," says Chris Heim, president of HighJump Software, a 3M company based in Eden Prairie, Minn. "DCs can be asked to do anything from tagging to final assembly to personalizing a product for a particular company, so in some cases they are almost a mini-factory."

"A key capability to support the transformation or conversion of product, whether raw materials or components, into a new SKU, is work-order processing," says Jim Stollberg, vice president and general manager at Irista, a supply-chain execution provider based in Milwaukee. Conversions also could involve taking a finished goods product that is stored in bulk and packaging it into different case or carton configurations, he says. "Again, it is all about being able to differentiate and customize that product in the last leg before it goes out the door. This is what gives you the flexibility to adapt to changes in demand."

Postponement has been used by the high-tech industry for some time, "but now we are seeing it in all the markets we serve, including our third-party logistics clients who want to offer this type of service to their customers," says Paul Crist, vice president of sales and marketing at Provia Software, Grand Rapids, Mich. "People want unique products, configured to their needs, right now," he says, "and that is the strength of our kitting solution."

As an example, Crist notes that cellular phone companies sell the same products to different mass retailers, but they customize these products by configuring and packaging them differently. "So, rather than building up inventory specific to all those retailers, they store components and build to order," he says. When the real orders come in, the fulfillment center takes a generic phone and sends it down a kitting line in the DC for that particular retailer, each with unique paperwork and unique handling requirements and unique configurations and packaging."

RedPrairie, an execution software provider based in Waukesha, Wis., has enhanced its postponement capabilities by moving the kits themselves to an on-demand model. "Manufacturers historically have tried to meet order demand by creating kits to complete generic products, but then they put the kits in a warehouse and wait for the orders to arrive," says Tom Kozenski, value delivery leader.

RedPrairie's solution allows companies to store information about the components needed for each kit, and then automatically list the components to be picked as part of the order-fulfillment process, he says. The kit is created on demand and not by forecast. "This helps because order processing doesn't have to see that kit in inventory in order for it to release an order," he says. "They can release the components and let the warehouse management system figure out how to put it together."

Stollberg notes that fulfillment customization also extends to delivery and transportation services, a fact driven, in part, by customer segmentation. A mass merchandiser like Wal-Mart may have very specific instructions about how it wants things shipped and delivered, he says, while other customers simply want the lowest cost mode that meets their service need. "To support the wide difference in customer requirements, warehousing and transportation need to be tightly coupled," he says. "Just interfacing WMS to TMS doesn't cut it in today's world. The two systems have to behave as one."

Short Planning Windows
Value-added delivery services also play a role in Advanced Order Fulfillment. "A lot of our customers are no long selling products, but products bundled with a high level of services," says Sweeney. This adds another level of complexity. "In addition to having inventory you want to move rapidly, you also want to make sure you are effectively using all your service capacity," he says. "If you are paying someone to install a big-screen TV and there is no work for them to do, that is the same as inventory sitting on the shelf."

These challenges are exacerbated by very short planning windows in today's supply chains. "If you think back five or 10 years ago, distribution centers had a nice window of time to plan their daily operations," says Stollberg. "Managers could take a day or two to consolidate orders and group them into nice waves. They don't have that luxury anymore. Now, if an order is in by noon, it usually is expected to ship the same day, so we have become much more dependent on real-time systems. We have to react and respond a lot faster."

Fingerhut, a general merchandise mail-order retailer, provides a good example of this time compression. Before installing new fulfillment software from HighJump, Fingerhut operated on a batch process bringing orders into the warehouse once a day. "Now we bring orders in once an hour," says Loren Eggert, vice president of distribution and fulfillment. "Typically orders go out in 24 to 36 hours." This timeframe gives the company the flexibility of shipping most orders the same day, while managing the workload so that there is enough in the bucket to start the following morning efficiently. "Overall, we have cut two days out of our delivery time," he says. This not only has taken Fingerhut's service "to a new level," he says, but has had a "very positive impact" on the rate of returns.

Lamps Plus is another retailer that accelerated its order turns after implementing fulfillment software from Escalate Fulfillment, Redwood Shores, Calif. This retailer's internet store has become its largest, with more than 500 orders per day. Orders that used to ship in five to seven days now are out the door in 24 to 48 hours, according to Bill Gratke, director of planning and supply-chain management. One key part of the Escalate solution was a packing optimization engine that enables Lamps Plus to easily consolidate multi-line orders into a smaller number of packages. "The strongest thing to say about what Escalate did for us is that they listened to us and molded their software into what we could use," Gratke says.

The length of supply chains when products are coming from other parts of the world also challenge fulfillment managers. "There is a whole new move to make sure, on the front end, that orders are filled accurately," says Mark Holmes, a consultant with Avicon, Waltham, Mass. "I call this advanced order fulfillment accuracy. This concept uses event management to red flag orders when the purchase order and loading manifest do not match, he explains, so that resolution can occur before the shipment ever leaves port. "It is very costly to wait until something gets to the DC to discover it is not as ordered," he says.

Optimization
Improved optimization technology also is helping companies enable Advanced Order Fulfillment. For example, says Kozenski, knowing which warehouse to ship from is an important consideration when stock is held or produced in multiple locations. "Some warehouses may not have an item in stock, some may be too far away and transportation costs too high, or some may ship direct from one of several vendors, so which is the optimal location to ship that order or line item from?" RedPrairie's SourceRight solution provides that calculation.

Being able to allocate inventory in transit also is crucial, says Lars-Goran Olsson, vice president of product management of Industri-Matematik International, Mt. Laurel, N.J. "One of the trends we see is that a large number of products are not stocked in quantity in the warehouse, so companies need to be able to act on the inventory that is in motion. This goes back to needing a closer relationship between different processes."

Execution optimization is "the natural extension of the warehouse becoming an on-demand operation," says Crist. Provia borrowed its approach to optimization from the manufacturing sector and its use of constraint-based modeling. "Warehouses also have constraints - certain processes take so long to do, equipment constraints, labor constraints, inventory constraints, transportation constraints, value-added service constraints, and so on. We have built the same kind of algorithms used in manufacturing into our warehouse management system, so we can model those constraints. You run a group of orders through - for the next hour or the next day, whatever time period - and it will churn through all the information and release a sequence of orders for processing that results in the maximum throughput. As things change, the optimization can be re-run in the background."

For example, he says, if a truck that was expected to come in at 11:00 in the morning with parts needed to perform a kitting function doesn't arrive, "We can rerun the optimization and move those orders to later in the day and fill that time quickly with orders that can be processed."

Jay Baitler, executive vice president of Staples Contract Division, recently installed the latest upgrade to its PkMS warehouse management solution from Manhattan Associates, Atlanta. "This software has a unique predictive capability that really intrigues me," he says. "It used to be that you would walk around a warehouse and if you saw a bottleneck building up you would throw labor at it. The first line of defense was a good, smart supervisor who could see that bottleneck developing in receiving or sortation or wherever. This installation of PkMS we are now doing has predictive intelligence so that it can tell you, based on the activity it is seeing, that a backlog will develop in this area by 2 p.m. today and you can afford to transition some labor from this place to that place to smooth it out over the balance of the day.

"I am really excited about that," he says. "It shows how far we have come in realizing this quest for operational excellence. Of course, we have to keep kicking that can to make it even better tomorrow, but this is pretty exciting stuff."

How Not to Deceive Yourself About the Perfect Order
How to best measure the success of fulfillment efforts is the subject of continuing debate. Kate Vitasek, managing partner at Supply Chain Visions, Bellevue, Wash., does a lot of consulting and training work in this area.

"It is amazing how many companies will tell you they have a fill rate of 99 percent," she says, "but then when you really start drilling down, you find that the way they got to 99 percent was by expediting a lot of shipments." Another common tactic companies use is to drop an order to the warehouse only when they know the product is in stock. "I go into companies all the time that measure their fill rate based on when an order is dropped in the DC, not when they received the order," she says. "The order could have set around for three days waiting for the inventory to come in or it could have been on credit hold or whatever. The customer wanted it yesterday and it hasn't even shipped yet, but that is being counted as a success in terms or their fulfillment metrics."

To avoid this and other "order fill traps" Vitasek recommends that companies use the Perfect Order Index (POI) to understand how they are performing and where they need to improve. "A perfect order has typically been defined as on time, complete, damage free and having the correct invoice," she says, though there are variations on this theme. Using POI, performance is measured by multiplying the metrics for each of these to each other. "For example," she says, "if a firm is experiencing 95 percent on-time delivery, fill rate, correct invoice and damage-free shipments, the resulting perfect order index would be 81.4 percent. This explains why many companies think they have good performance when they actually are not meeting customer expectations - or are cultivating behaviors that drive costs up."

When Jay Baitler, executive vice president of Staples Contract Division, understood the impact of the multiplier effect, he decided to no longer rate success based on individual components of Staples' version of the perfect order. "We rate ourselves very, very, very hard," he says. For example, if a product is out of stock but can be obtained from a supplier and still delivered on time and to the customer's expectation, "we still ping ourselves for that as a failure because we didn't have it in inventory," says Baitler. "We work very hard to make sure that our recovery in such situations is excellent, but we measure ourselves so as to make sure that those performance numbers get better every single day. The quest for that perfect order is never over."

To effectively implement a perfect order approach, companies need to change their incentives, says Vitasek. "Instead of measuring people just on what is in their control, they need to also be measured on the bigger picture, the overall outcome."

Steve Banker of ARC Advisory Group, Dedham, Mass., agrees. To really address the cross-functional nature of the perfect order, incentives have to be designed so that at least part of the bonus for people in purchasing depends on how well the people do in manufacturing or in the warehouse, and vice versa, he says. Otherwise, "the natural tendency of people is to define metrics from their own internal point of view," he says. "So you end up with a series of departmental metrics that, when you add them up, don't really mean much to the customer."