Executive Briefings

How to Build A Supply Chain For All Seasons

Companies use smart logistics strategies and the latest in technology to help predict and meet seasonal demand, but getting it right is still an art.

For many supply chains, particularly those tied to retail, the calendar rules. With a majority of annual sales occurring during short, seasonal windows, having the right products, in the right amounts, on the right shelves during peak selling weeks is a critical test of both supply chain planning and execution. That nearly every season has shortages of "hot" items and early markdowns of "cold" ones attests to the continuing difficulty of this challenge.

"Seasonality, in itself, is neither good nor bad, but it is a magnifier," says Scott Szwast, sector marketing manager for UPS Supply Chain Solutions, Atlanta. "It magnifies the differences between an exceptional supply chain and a mediocre supply chain, and it is absolutely unforgiving of a poor supply chain."

Whether seasonal peaks are around the December holidays, back-to-school, summer outdoor living or some other milestone, companies typically determine when they need product in stock to meet seasonal demand and work back from there to manage supply. This requires a solid understanding of upstream supply chain constraints.

Ed Dupee, director of demand chain management at Teradata, a forecasting software company based in Dayton, Ohio, cites three basic constraints that impact seasonal planning. These are: throughput capacity or the volume that can be moved through a company's logistics and transportation network; manufacturing capacity; and financial and procurement issues, which may require some product to be purchased in advance or some manufacturing to be moved offshore. Once this latter decision is made, Dupee says, a company often loses the ability to respond to demand as it happens.

This is particularly true of private brands, notes Ivan Reed, senior director of campaigns at retailer Office Depot, Del Ray Beach, Fla. "A big push in our industry as well as every other is not only to globally source products but move to private brands," he says, adding that lead times on such products is a lot longer because they require due diligence in terms of ensuring that quality standards are met and that factories comply with social requirements. "If it is coming under the Office Depot name, we want it to be equivalent to a national brand or better and we want to be certain that there are no children working in the factories and that the health issues are up to code," he says. "The point is that these lead times are greatly increased because we don't have a third-party factory that owns the due diligence-we own it ourselves."

In addition, he says, once the company buys from Asia, "it's ours. There is no factory or domestic partner we can return product to." Better margins on these products make the gamble worthwhile, he says, but although Office Depot has been developing and importing its own product for about five years, "we still have issues around seasonal merchandise."

This is partly due to the fact that Office Depot does not bring in safety stock and hold it, but flows inventory through a cross-dock in Los Angeles and on to stores. "It is very difficult to time those direct import products and make sure we have full shelves without loading up our stores with inventory," he says.

Nevertheless, for this year's back-to-school season-one of Office Depot's four key selling seasons-the company is front-loading its stock. "Last year, Reed explains, "we used our standard replenishment algorithms for back to school, which is more of a phased-time replenishment model-you get down to a minimum number and the software kicks out an order for an additional standard pack. We thought we had our algorithms right, but we got hit pretty hard with out-of-stocks and, of course, we live and breathe on our in-stock positions. So this year we have ordered 75 percent of our back-to-school inventory up front. This has put a little bit of a strain on our logistics and on our flow-through and also a bit of strain on our stores because they don't have a lot of back room storage. But by loading that product up front, we should be able to replenish and protect our in-stock position. Since we took a bit of a hit last year, we just don't want take any chances."

Buying Time
Rubie's Costume Co., Richmond Hill, N.Y., provides another example of a highly seasonal company with significant offshore sourcing. The peak season for Rubie's is Halloween and "our typical year starts Nov. 1," says Joel Reisman, general manager. "Right after Halloween we are off to the Orient making purchases and making customer presentations for the next year."

Customers make their commitments to Rubie's in February, March and April, says Reisman, and they want everything in their distribution centers in July and August. "That doesn't give us a great amount of time to manufacture and produce merchandise."

To try and "buy some time," Rubie's has adopted new methods of distribution. Rather than bringing everything into its East Coast distribution center, as had been its practice until a couple of years ago, Rubie's now uses Regal Logistics, a 3PL based near Seattle, Wash., to receive and ship goods imported from Asia. "This saves us about two and a half valuable weeks as well as saving us money," says Reisman. "Instead of us shipping merchandise across the country to the East Coast and then back the other way, where many of our customers are located, we now ship from the West Coast facility out to DCs around the country." An added bonus to this plan is that it frees more resources at Rubie's own warehouses to react to additional orders.

In another enhancement, Rubie's partnership with Maersk Logistics, Madison, N.J., has increased the company's visibility "to what is ready to ship and what is on the water," says Reisman. Maersk handles the movement of Rubie's goods from Asia to the West Coast and "is linked in via EDI directly to our computer system," he says. "This allows us to know the status of all of our shipments."

Ritz Camera, Beltsville, Md., relies on UPS Supply Chain Solutions to arrange ocean freight and handle customs clearance for its seasonal surge in imports. "We try to be proactive by bringing imported December holiday goods in early to our distribution centers before our domestic suppliers send their goods," says Vice President Bob Elton. The company's goal is to get imported goods in as early as July and August for distribution to retail stores in September. Domestic shipments arrive at the DCs in August and September for shipment to retail stores in October and early November. "We like to schedule domestic shipments in three waves, with a third coming in September, a third in October and a third in November," says Elton. "This keeps product flowing through to the point where our DCs can handle the volume and it doesn't load up our stores, which don't have a lot of storage room, with inventory."

UPS also is Ritz's key carrier from its distribution centers to the stores. "Our biggest peak-season nightmare is getting the maximum pick for the day in our warehouse without running over the UPS cut-off time," says Elton. UPS makes this task easier by increasing the number of pickups from one to three. "UPS will have a truck pickup, say, at 1 p.m. and again at 4 p.m., and then they give us a last-minute shot with a 7 p.m. pickup. That is really critical to moving product out of our facility during busy season."

The ability to enable fast scale-ups, with added space and transport capacity, are key reasons that companies turn to 3PLs to help manage seasonal surges. "Often the seasonal volume that must go through a company's distribution system is simply greater than what these companies can handle on their own," says Brian Piern, director of product development at Schneider Logistics, Green Bay, Wis. "We have the ability to help them surge with those situations, and we can do it without disrupting their normal distribution channels."

Where transportation is concerned, even companies that work with several core carriers can have trouble obtaining sufficient capacity to deploy product in a timely fashion during peak seasons, says Piern. That's when Schneider turns to its base of 10,000 small carriers and brokers. "A lot of companies have routing guides that demand major transportation companies so they may not have exposure to the large portion of service providers that are within our reach," he says.

Another solution in Schneider's arsenal is to use rail to forward deploy product to a market and provide temporary storage and delivery as needed.

Technology also can help with this conundrum, says Dan Sobbott, general manager at SLIM Technologies, Boston, which specializes in network optimization software. "Most of our clients design their network for the normal part of their season because this ensures that their asset base gets heavy utilization," he says. "They then use our tool to analyze capacity needs for peak periods." The SLIM tool helps companies determine what third parties to use, how much space they need at which locations, and where different products should be held, he says.

Often 3PLs can help meet seasonal surges and lower costs for customers by sharing assets between companies with opposite seasonal cycles.

DHL Solutions, Plantation, Fla., does this for numerous clients, says Chris Davy, head of sales and marketing. One example matches a large B2C e-tailer that peaks in fourth quarter with a smaller B2B retailer with seasonal demand in the summer and back-to-school. The clients share space at DHL's hub in Wilmington, Ohio. Since all workers are DHL employees they can easily flex between the two clients, says Davy. "We also can draw on thousands of employees that work in the sort operation if we need to," he adds. Space also flexes between the two. In the fourth quarter, the first client may occupy 120,000 to 140,000 square feet of space, but that shrinks by half in the summer months, when the second client begins its busy period, he says.

Yet another way that third parties can help companies with seasonal demand is to provide delayed differentiation or postponement services. "With ports on the West Coast becoming more and more blocked and lead times increasing for off-shore suppliers, there is a higher value on postponement capabilities and on being a short lead-time supplier," says Jared Schreiber, senior supply-chain consultant for Teradata. Companies should look at the trade-off between a 90-day supply chain at a certain cost point on one hand, and increased labor hours with a substantially shortened cycle time on the other, he says. "As capacity constraints grow this will become even more important," says Schreiber.

Paul Dittman, a supply-chain faculty member at the University of Tennessee and former supply-chain executive at Whirlpool, reiterates this point. "The whole issue of manufacturing flexibility is a big deal in terms of seasonality, particularly with the mass exodus to outsourcing that we are seeing," he says. "Trying to match supply and demand is pretty impossible if you are dealing with seasonal spikes and also outsourcing at the same time. Delaying differentiation until later in the supply chain, maybe even in the warehouse, would allow companies to better respond to the variability they can't predict in advance."

Stockpiling Strategy
For now, however, building stock in advance of seasonal demand is the strategy most frequently used by companies. Godiva Chocolate, for example, has 60 percent of its sales between October and Valentine's Day. To meet that demand, it begins manufacturing and freezing its fine chocolates as early as June. "After many years of not having capacity issues, we have finally gotten to the point where our seasonal requirements outstrip our ability to produce from, say, August to December," says Karen Jensen, forecasting manager. "Our sales curve has this huge spike, but you can't have a manufacturing curve quite like that. It needs to be a lot more leveled out."

Godiva uses SmartForecasts from Smart Software, Belmont, Mass., to project demand for 11 customer groups, a method that has helped improve forecast accuracy from 50 percent to 80 percent-a level that Jensen says "is probably as good as we can get." In addition to the software's statistical capability, Jensen says it has imposed a process that is perhaps even more important. "What was happening before is that our marketing group was doing the forecasting," she says. "They would put some numbers out there and never look at them again. The whole process of implementing SmartForecasts has served to establish a procedure. I'm not saying we have the best forecasts in world, but what we do have is people who are paying attention to it, spending 100 percent of their time on forecasting and really looking at the numbers. I think that is the key."

Solid Forecasting
Iowa-based Wells Dairy, maker of Blue Bunny ice cream, Bomb Pop and other ice-cream novelties, starts putting products away as early as January to meet its summer demand surge. "We have everything in place pretty much by May," says Scott Roy, director of demand planning. "After that we are replenishing, shipping out as much as we can make."

Stockpiling products that far in advance requires a solid forecasting process, he says, "because you are putting dollars into inventory before you are even in the season." Wells uses a forecasting solution from John Galt, Chicago. "We really have to have a good forecast in place in September and October for the coming year," says Roy. "We will make adjustments as we get into the year, but we have to make a lot of commitments to our suppliers and we need numbers to back that up."

Roy likes the flexibility of the John Galt solution, which allows users to define the forecast period. "We forecast in four 13-week periods," he says. Because Wells sells primarily to warehouses, he explains, a full truckload order may come in week one, followed by no orders in weeks two and three, and then another truckload in week four. "A lot of applications would see this as random spikes in demand, he says, but because Wells is able to look at 13-week buckets, it can see the patterns. "The application gives you a lot of flexibility to expand or contract how you are forecasting," he says. "With a lot of systems, once you set them up you are stuck doing it the same way over and over."

Not all companies are able to stockpile goods, however, and have to rely on big ramp-ups in production and distribution to meet seasonal demand. One example is Old Neighborhood Foods, Lynn, Mass., which makes cold cuts, hot dogs and barbecue sausages.

"We practically make as much in the summer as we do in the other nine months combined," says Peter Katsos, MIS director. Since Old Neighborhood does not freeze product, it has to produce more of its perishable products to meet this demand. The company uses software from Ross Systems, Atlanta, for forecasting and materials planning. "We have to make sure that we have enough raw materials like pork and beef coming in," says Katsos, as well as packaging materials. "We buy a lot of films and plastics from big companies like Cryovac and they require an eight- to 12-week lead time," Katsos says.

Stock Smarts
Whatever a company stockpiles or ramps up production to meet seasonal demand, understanding how much seasonal stock to plan for and where to place that stock are paramount. "An overbuy in seasonal goods will kill your profits," says Linda Baker, director of retail merchandising strategies at Oracle/Retek, Redwood Shores, Calif. "There is absolutely nothing you can do when the customer decides that a seasonal product is dead or when you are beyond the season. Then, there is almost no bottom price to the goods."

To address this issue, a few leading companies are trying to capture early signals and then adjust forecasts for the rest of the season. Such early detection and response is the "holy grail" of forecasting, says Jerry Blanford, vice president of retail sales at software developer Manugistics, Rockville, Md. "Most companies just take a pot shot at what is going to be hot in terms of style and color and hope they have the right assortment," he says. "But what if you could put pink and blue out there and early in the season see that pink is going to take off? If you had reserved manufacturing space and time and material, then you could go back and say, 'OK, turn all that material into pink in these sizes.' That's where a lot of retailers are trying to go."

Fashion retailer The Limited has been working on this strategy for some time and recently opted to implement the full Manugistics suite of products to support this effort. "Their whole strategy is to get products in the stores quickly, understand what is going to sell and then collapse their seasonal lead time from nine months to 12 weeks," Blanford says.

The key is having total inventory visibility and collaborative tools that allow them to quickly communicate with manufacturers, says Blanford. "Having replenishment and allocation systems that tie back into transportation and to international track-and-trace capability-bringing all of that together-is a very, very powerful value proposition."

A major retailer customer of Teradata is working in a similar way with a large apparel manufacturer, says Dupee. "If they plan to sell 10,000 pair of shorts during the summer season through a particular DC, they do not pre-build all 10,000. Instead, they pre-build the first four weeks and then use actual sales history to do replenishment within the season.

"Within the first three weeks our system reacts to actual sales," says Dupee. "This allows companies to react to unpredictables like weather. If it is cooler or hotter than usual, they can react."

Another advance in this area is "key item planning," says Baker, an approach that is gaining interest in the apparel industry. The idea is to forecast around a concept instead of a particular item, she explains. "For example, a particular style of embroidered-back-pocket jeans is not as important as having 500 pieces of apparel with an embroidered back pocket," she says. "This approach allows you to manage clumps of inventory on a very tight weekly level while protecting the big ideas inside of each season." Retek's products support this type of key item planning, she says.

Regional Differences
Predicting seasonal demand also is complicated by geographic and demographic differences as well as local regulations. Not all schools open at the same time around the country, for example. Here again, increasingly sophisticated demand planning and forecasting applications are helping companies deal with such vagaries by providing greater planning flexibility and more granular information.

"The ability to manage different seasonal patterns in different regions cannot be done manually anymore. The variations are just too numerous," says CEO Sridhar Tayur of SmartOps, an inventory optimization and forecasting software based in Pittsburgh, Pa. Technology also is needed to help companies account for uncertainty in their planning and to adapt quickly to new information, he says. All of these issues are addressed by SmartOps' Multi-stage Inventory Plant Optimization (MIPO) solution. "MIPO goes after these three requirements by going for granularity, specificity and adaptability," Tayur says. "We look at previous plans and previous actuals, collect from that a typical uncertainty and embed that uncertainty in our setting of future plans. And by knowing that different regions have different uncertainties, as well as different seasonal features, we are able to produce answers that vary by region, by time and by item." Moreover, he says, because the application is completely automated and integrated, weekly updates of actual sales can be used to "realign the plan in the most adaptable way possible."

Logility also emphasizes the importance of data granularity. "In addition to looking at a seasonal pattern for a specific item or a group of items, we can break that down geographically or by channel of distribution to really analyze the demand pattern from multiple perspectives," says Karin Bursa, vice president of marketing at the Atlanta-based company, whose Voyager products enable supply chain and demand planning. Users can see, as well, if there are one or two outlying products that are artificially inflating demand for other items, she says.

The Voyager planning engine also can apply a lifecycle curve to a product of group of products, says Bursa. "That allows users to have better visibility of the sell-in, meaning how much product needs to be available to supply all retail locations, both initially and to replenish throughout that season."

As good as the technology is, however, when it comes to seasonal forecasting, some art is still involved. "There is science, but also research and intuition," says Office Depot's Reed. "We try to rely more heavily on facts, but we also have to put in the intuition."

"It is an art," agrees Manugistics' Blanford. "But you have to use science and math along with art or you are going to get killed."

 

Four Simple Steps
Paul Dittman, former vice president in charge of the supply chain for Whirlpool Corp. and now a member of the supply-chain faculty at the University of Tennessee, Knoxville, offers several steps that companies can take to help master supply chain seasonality.

1. Keep it simple. "Companies affected by seasonal demand should rationalize their SKUs and make sure their product line is no more complicated than it needs to be," Dittman says. Since most companies have to build in advance to meet seasonal spikes, having fewer SKUs will reduce forecast error, he explains, though this is an issue that most companies "fail to think through. They don't realize that they are dramatically hurting their ability to forecast accurately by allowing SKUs to multiply," he says.

2. Keep inventory clean and lean. "Once a season is nearing its end, it is important to clean things out," says Dittman. "Do whatever it takes to avoid carrying over inventory because it is a case of pay me now or pay me later. You will have to write down all that old stuff eventually and holding it simply ties up working capital and diverts dollars away from purchasing better material that ought to be stocked for the new season."

3. Build core products first. "Given that you have to build in advance and that you know the forecast is going to very inaccurate-in total and definitely at the SKU level-it makes sense to focus advance building on core, high volume SKUs that you know are going to move," says Dittman. "Save your in-season capacity to build the peripheral stuff that is much smaller volume and has much higher forecast error associated with it. Too many companies think they have to go into the season with three months' of supply and so they build three months' of supply of every SKU. A better tactic is to have more than the proportional amount of inventory in the really high moving SKUs and much less in the others."

4. Adopt modular design. "The issue of flexibility is all about how fast can you respond when you see how the forecast is wrong, either on the high side or the low side. How fast can you turn your manufacturing operations on or off, or shift production around? A lot of times this goes all the way back to product design. The simpler and more modular the design, the faster a company can react."

 

 

For many supply chains, particularly those tied to retail, the calendar rules. With a majority of annual sales occurring during short, seasonal windows, having the right products, in the right amounts, on the right shelves during peak selling weeks is a critical test of both supply chain planning and execution. That nearly every season has shortages of "hot" items and early markdowns of "cold" ones attests to the continuing difficulty of this challenge.

"Seasonality, in itself, is neither good nor bad, but it is a magnifier," says Scott Szwast, sector marketing manager for UPS Supply Chain Solutions, Atlanta. "It magnifies the differences between an exceptional supply chain and a mediocre supply chain, and it is absolutely unforgiving of a poor supply chain."

Whether seasonal peaks are around the December holidays, back-to-school, summer outdoor living or some other milestone, companies typically determine when they need product in stock to meet seasonal demand and work back from there to manage supply. This requires a solid understanding of upstream supply chain constraints.

Ed Dupee, director of demand chain management at Teradata, a forecasting software company based in Dayton, Ohio, cites three basic constraints that impact seasonal planning. These are: throughput capacity or the volume that can be moved through a company's logistics and transportation network; manufacturing capacity; and financial and procurement issues, which may require some product to be purchased in advance or some manufacturing to be moved offshore. Once this latter decision is made, Dupee says, a company often loses the ability to respond to demand as it happens.

This is particularly true of private brands, notes Ivan Reed, senior director of campaigns at retailer Office Depot, Del Ray Beach, Fla. "A big push in our industry as well as every other is not only to globally source products but move to private brands," he says, adding that lead times on such products is a lot longer because they require due diligence in terms of ensuring that quality standards are met and that factories comply with social requirements. "If it is coming under the Office Depot name, we want it to be equivalent to a national brand or better and we want to be certain that there are no children working in the factories and that the health issues are up to code," he says. "The point is that these lead times are greatly increased because we don't have a third-party factory that owns the due diligence-we own it ourselves."

In addition, he says, once the company buys from Asia, "it's ours. There is no factory or domestic partner we can return product to." Better margins on these products make the gamble worthwhile, he says, but although Office Depot has been developing and importing its own product for about five years, "we still have issues around seasonal merchandise."

This is partly due to the fact that Office Depot does not bring in safety stock and hold it, but flows inventory through a cross-dock in Los Angeles and on to stores. "It is very difficult to time those direct import products and make sure we have full shelves without loading up our stores with inventory," he says.

Nevertheless, for this year's back-to-school season-one of Office Depot's four key selling seasons-the company is front-loading its stock. "Last year, Reed explains, "we used our standard replenishment algorithms for back to school, which is more of a phased-time replenishment model-you get down to a minimum number and the software kicks out an order for an additional standard pack. We thought we had our algorithms right, but we got hit pretty hard with out-of-stocks and, of course, we live and breathe on our in-stock positions. So this year we have ordered 75 percent of our back-to-school inventory up front. This has put a little bit of a strain on our logistics and on our flow-through and also a bit of strain on our stores because they don't have a lot of back room storage. But by loading that product up front, we should be able to replenish and protect our in-stock position. Since we took a bit of a hit last year, we just don't want take any chances."

Buying Time
Rubie's Costume Co., Richmond Hill, N.Y., provides another example of a highly seasonal company with significant offshore sourcing. The peak season for Rubie's is Halloween and "our typical year starts Nov. 1," says Joel Reisman, general manager. "Right after Halloween we are off to the Orient making purchases and making customer presentations for the next year."

Customers make their commitments to Rubie's in February, March and April, says Reisman, and they want everything in their distribution centers in July and August. "That doesn't give us a great amount of time to manufacture and produce merchandise."

To try and "buy some time," Rubie's has adopted new methods of distribution. Rather than bringing everything into its East Coast distribution center, as had been its practice until a couple of years ago, Rubie's now uses Regal Logistics, a 3PL based near Seattle, Wash., to receive and ship goods imported from Asia. "This saves us about two and a half valuable weeks as well as saving us money," says Reisman. "Instead of us shipping merchandise across the country to the East Coast and then back the other way, where many of our customers are located, we now ship from the West Coast facility out to DCs around the country." An added bonus to this plan is that it frees more resources at Rubie's own warehouses to react to additional orders.

In another enhancement, Rubie's partnership with Maersk Logistics, Madison, N.J., has increased the company's visibility "to what is ready to ship and what is on the water," says Reisman. Maersk handles the movement of Rubie's goods from Asia to the West Coast and "is linked in via EDI directly to our computer system," he says. "This allows us to know the status of all of our shipments."

Ritz Camera, Beltsville, Md., relies on UPS Supply Chain Solutions to arrange ocean freight and handle customs clearance for its seasonal surge in imports. "We try to be proactive by bringing imported December holiday goods in early to our distribution centers before our domestic suppliers send their goods," says Vice President Bob Elton. The company's goal is to get imported goods in as early as July and August for distribution to retail stores in September. Domestic shipments arrive at the DCs in August and September for shipment to retail stores in October and early November. "We like to schedule domestic shipments in three waves, with a third coming in September, a third in October and a third in November," says Elton. "This keeps product flowing through to the point where our DCs can handle the volume and it doesn't load up our stores, which don't have a lot of storage room, with inventory."

UPS also is Ritz's key carrier from its distribution centers to the stores. "Our biggest peak-season nightmare is getting the maximum pick for the day in our warehouse without running over the UPS cut-off time," says Elton. UPS makes this task easier by increasing the number of pickups from one to three. "UPS will have a truck pickup, say, at 1 p.m. and again at 4 p.m., and then they give us a last-minute shot with a 7 p.m. pickup. That is really critical to moving product out of our facility during busy season."

The ability to enable fast scale-ups, with added space and transport capacity, are key reasons that companies turn to 3PLs to help manage seasonal surges. "Often the seasonal volume that must go through a company's distribution system is simply greater than what these companies can handle on their own," says Brian Piern, director of product development at Schneider Logistics, Green Bay, Wis. "We have the ability to help them surge with those situations, and we can do it without disrupting their normal distribution channels."

Where transportation is concerned, even companies that work with several core carriers can have trouble obtaining sufficient capacity to deploy product in a timely fashion during peak seasons, says Piern. That's when Schneider turns to its base of 10,000 small carriers and brokers. "A lot of companies have routing guides that demand major transportation companies so they may not have exposure to the large portion of service providers that are within our reach," he says.

Another solution in Schneider's arsenal is to use rail to forward deploy product to a market and provide temporary storage and delivery as needed.

Technology also can help with this conundrum, says Dan Sobbott, general manager at SLIM Technologies, Boston, which specializes in network optimization software. "Most of our clients design their network for the normal part of their season because this ensures that their asset base gets heavy utilization," he says. "They then use our tool to analyze capacity needs for peak periods." The SLIM tool helps companies determine what third parties to use, how much space they need at which locations, and where different products should be held, he says.

Often 3PLs can help meet seasonal surges and lower costs for customers by sharing assets between companies with opposite seasonal cycles.

DHL Solutions, Plantation, Fla., does this for numerous clients, says Chris Davy, head of sales and marketing. One example matches a large B2C e-tailer that peaks in fourth quarter with a smaller B2B retailer with seasonal demand in the summer and back-to-school. The clients share space at DHL's hub in Wilmington, Ohio. Since all workers are DHL employees they can easily flex between the two clients, says Davy. "We also can draw on thousands of employees that work in the sort operation if we need to," he adds. Space also flexes between the two. In the fourth quarter, the first client may occupy 120,000 to 140,000 square feet of space, but that shrinks by half in the summer months, when the second client begins its busy period, he says.

Yet another way that third parties can help companies with seasonal demand is to provide delayed differentiation or postponement services. "With ports on the West Coast becoming more and more blocked and lead times increasing for off-shore suppliers, there is a higher value on postponement capabilities and on being a short lead-time supplier," says Jared Schreiber, senior supply-chain consultant for Teradata. Companies should look at the trade-off between a 90-day supply chain at a certain cost point on one hand, and increased labor hours with a substantially shortened cycle time on the other, he says. "As capacity constraints grow this will become even more important," says Schreiber.

Paul Dittman, a supply-chain faculty member at the University of Tennessee and former supply-chain executive at Whirlpool, reiterates this point. "The whole issue of manufacturing flexibility is a big deal in terms of seasonality, particularly with the mass exodus to outsourcing that we are seeing," he says. "Trying to match supply and demand is pretty impossible if you are dealing with seasonal spikes and also outsourcing at the same time. Delaying differentiation until later in the supply chain, maybe even in the warehouse, would allow companies to better respond to the variability they can't predict in advance."

Stockpiling Strategy
For now, however, building stock in advance of seasonal demand is the strategy most frequently used by companies. Godiva Chocolate, for example, has 60 percent of its sales between October and Valentine's Day. To meet that demand, it begins manufacturing and freezing its fine chocolates as early as June. "After many years of not having capacity issues, we have finally gotten to the point where our seasonal requirements outstrip our ability to produce from, say, August to December," says Karen Jensen, forecasting manager. "Our sales curve has this huge spike, but you can't have a manufacturing curve quite like that. It needs to be a lot more leveled out."

Godiva uses SmartForecasts from Smart Software, Belmont, Mass., to project demand for 11 customer groups, a method that has helped improve forecast accuracy from 50 percent to 80 percent-a level that Jensen says "is probably as good as we can get." In addition to the software's statistical capability, Jensen says it has imposed a process that is perhaps even more important. "What was happening before is that our marketing group was doing the forecasting," she says. "They would put some numbers out there and never look at them again. The whole process of implementing SmartForecasts has served to establish a procedure. I'm not saying we have the best forecasts in world, but what we do have is people who are paying attention to it, spending 100 percent of their time on forecasting and really looking at the numbers. I think that is the key."

Solid Forecasting
Iowa-based Wells Dairy, maker of Blue Bunny ice cream, Bomb Pop and other ice-cream novelties, starts putting products away as early as January to meet its summer demand surge. "We have everything in place pretty much by May," says Scott Roy, director of demand planning. "After that we are replenishing, shipping out as much as we can make."

Stockpiling products that far in advance requires a solid forecasting process, he says, "because you are putting dollars into inventory before you are even in the season." Wells uses a forecasting solution from John Galt, Chicago. "We really have to have a good forecast in place in September and October for the coming year," says Roy. "We will make adjustments as we get into the year, but we have to make a lot of commitments to our suppliers and we need numbers to back that up."

Roy likes the flexibility of the John Galt solution, which allows users to define the forecast period. "We forecast in four 13-week periods," he says. Because Wells sells primarily to warehouses, he explains, a full truckload order may come in week one, followed by no orders in weeks two and three, and then another truckload in week four. "A lot of applications would see this as random spikes in demand, he says, but because Wells is able to look at 13-week buckets, it can see the patterns. "The application gives you a lot of flexibility to expand or contract how you are forecasting," he says. "With a lot of systems, once you set them up you are stuck doing it the same way over and over."

Not all companies are able to stockpile goods, however, and have to rely on big ramp-ups in production and distribution to meet seasonal demand. One example is Old Neighborhood Foods, Lynn, Mass., which makes cold cuts, hot dogs and barbecue sausages.

"We practically make as much in the summer as we do in the other nine months combined," says Peter Katsos, MIS director. Since Old Neighborhood does not freeze product, it has to produce more of its perishable products to meet this demand. The company uses software from Ross Systems, Atlanta, for forecasting and materials planning. "We have to make sure that we have enough raw materials like pork and beef coming in," says Katsos, as well as packaging materials. "We buy a lot of films and plastics from big companies like Cryovac and they require an eight- to 12-week lead time," Katsos says.

Stock Smarts
Whatever a company stockpiles or ramps up production to meet seasonal demand, understanding how much seasonal stock to plan for and where to place that stock are paramount. "An overbuy in seasonal goods will kill your profits," says Linda Baker, director of retail merchandising strategies at Oracle/Retek, Redwood Shores, Calif. "There is absolutely nothing you can do when the customer decides that a seasonal product is dead or when you are beyond the season. Then, there is almost no bottom price to the goods."

To address this issue, a few leading companies are trying to capture early signals and then adjust forecasts for the rest of the season. Such early detection and response is the "holy grail" of forecasting, says Jerry Blanford, vice president of retail sales at software developer Manugistics, Rockville, Md. "Most companies just take a pot shot at what is going to be hot in terms of style and color and hope they have the right assortment," he says. "But what if you could put pink and blue out there and early in the season see that pink is going to take off? If you had reserved manufacturing space and time and material, then you could go back and say, 'OK, turn all that material into pink in these sizes.' That's where a lot of retailers are trying to go."

Fashion retailer The Limited has been working on this strategy for some time and recently opted to implement the full Manugistics suite of products to support this effort. "Their whole strategy is to get products in the stores quickly, understand what is going to sell and then collapse their seasonal lead time from nine months to 12 weeks," Blanford says.

The key is having total inventory visibility and collaborative tools that allow them to quickly communicate with manufacturers, says Blanford. "Having replenishment and allocation systems that tie back into transportation and to international track-and-trace capability-bringing all of that together-is a very, very powerful value proposition."

A major retailer customer of Teradata is working in a similar way with a large apparel manufacturer, says Dupee. "If they plan to sell 10,000 pair of shorts during the summer season through a particular DC, they do not pre-build all 10,000. Instead, they pre-build the first four weeks and then use actual sales history to do replenishment within the season.

"Within the first three weeks our system reacts to actual sales," says Dupee. "This allows companies to react to unpredictables like weather. If it is cooler or hotter than usual, they can react."

Another advance in this area is "key item planning," says Baker, an approach that is gaining interest in the apparel industry. The idea is to forecast around a concept instead of a particular item, she explains. "For example, a particular style of embroidered-back-pocket jeans is not as important as having 500 pieces of apparel with an embroidered back pocket," she says. "This approach allows you to manage clumps of inventory on a very tight weekly level while protecting the big ideas inside of each season." Retek's products support this type of key item planning, she says.

Regional Differences
Predicting seasonal demand also is complicated by geographic and demographic differences as well as local regulations. Not all schools open at the same time around the country, for example. Here again, increasingly sophisticated demand planning and forecasting applications are helping companies deal with such vagaries by providing greater planning flexibility and more granular information.

"The ability to manage different seasonal patterns in different regions cannot be done manually anymore. The variations are just too numerous," says CEO Sridhar Tayur of SmartOps, an inventory optimization and forecasting software based in Pittsburgh, Pa. Technology also is needed to help companies account for uncertainty in their planning and to adapt quickly to new information, he says. All of these issues are addressed by SmartOps' Multi-stage Inventory Plant Optimization (MIPO) solution. "MIPO goes after these three requirements by going for granularity, specificity and adaptability," Tayur says. "We look at previous plans and previous actuals, collect from that a typical uncertainty and embed that uncertainty in our setting of future plans. And by knowing that different regions have different uncertainties, as well as different seasonal features, we are able to produce answers that vary by region, by time and by item." Moreover, he says, because the application is completely automated and integrated, weekly updates of actual sales can be used to "realign the plan in the most adaptable way possible."

Logility also emphasizes the importance of data granularity. "In addition to looking at a seasonal pattern for a specific item or a group of items, we can break that down geographically or by channel of distribution to really analyze the demand pattern from multiple perspectives," says Karin Bursa, vice president of marketing at the Atlanta-based company, whose Voyager products enable supply chain and demand planning. Users can see, as well, if there are one or two outlying products that are artificially inflating demand for other items, she says.

The Voyager planning engine also can apply a lifecycle curve to a product of group of products, says Bursa. "That allows users to have better visibility of the sell-in, meaning how much product needs to be available to supply all retail locations, both initially and to replenish throughout that season."

As good as the technology is, however, when it comes to seasonal forecasting, some art is still involved. "There is science, but also research and intuition," says Office Depot's Reed. "We try to rely more heavily on facts, but we also have to put in the intuition."

"It is an art," agrees Manugistics' Blanford. "But you have to use science and math along with art or you are going to get killed."

 

Four Simple Steps
Paul Dittman, former vice president in charge of the supply chain for Whirlpool Corp. and now a member of the supply-chain faculty at the University of Tennessee, Knoxville, offers several steps that companies can take to help master supply chain seasonality.

1. Keep it simple. "Companies affected by seasonal demand should rationalize their SKUs and make sure their product line is no more complicated than it needs to be," Dittman says. Since most companies have to build in advance to meet seasonal spikes, having fewer SKUs will reduce forecast error, he explains, though this is an issue that most companies "fail to think through. They don't realize that they are dramatically hurting their ability to forecast accurately by allowing SKUs to multiply," he says.

2. Keep inventory clean and lean. "Once a season is nearing its end, it is important to clean things out," says Dittman. "Do whatever it takes to avoid carrying over inventory because it is a case of pay me now or pay me later. You will have to write down all that old stuff eventually and holding it simply ties up working capital and diverts dollars away from purchasing better material that ought to be stocked for the new season."

3. Build core products first. "Given that you have to build in advance and that you know the forecast is going to very inaccurate-in total and definitely at the SKU level-it makes sense to focus advance building on core, high volume SKUs that you know are going to move," says Dittman. "Save your in-season capacity to build the peripheral stuff that is much smaller volume and has much higher forecast error associated with it. Too many companies think they have to go into the season with three months' of supply and so they build three months' of supply of every SKU. A better tactic is to have more than the proportional amount of inventory in the really high moving SKUs and much less in the others."

4. Adopt modular design. "The issue of flexibility is all about how fast can you respond when you see how the forecast is wrong, either on the high side or the low side. How fast can you turn your manufacturing operations on or off, or shift production around? A lot of times this goes all the way back to product design. The simpler and more modular the design, the faster a company can react."