In consumer-packaged goods markets, demand-management savvy means the difference between bloated stockpiles and the appropriate amount of product moving succinctly to retailer shelves and into consumer hands; or, worse, missing inventory and missed sales. For short-life product manufacturers, demand planning and management has become a way of life. But as all companies increasingly scratch for cost savings, market share and leaner supply chains, even "longer" manufacturers are exploring demand-management processes and the software tools that help facilitate them. "Traditional demand-management leading-edge practices come out of consumer industries . . . and now we're seeing a lot of interest within the industrial manufacturing segments, within high-tech and electronic segments," says Andy Carlson, director of product marketing, J.D. Edwards & Co. (JDEC) supply-chain solutions, Denver. "Those guys have been left out in the cold with statistical forecasting, and history has been a bad predictor of the future in a lot of those industries. Now that they have the ability to structure the way they bring their stakeholders -- sales force, channels, distributors or even direct customers -- into the process, we're seeing a lot of interest [beyond simple] forecasting." Demand management is the intersection of supply and demand, and demand-management software the traffic cop pointing operational resources squarely at markets. The tools do indeed help assemble forecasts and demand plans, but they also drive plans into the organization, trigger execution against the plans, and, ultimately, ensure the right goods arrive at the right time and place. Organizational benefits can include more accurate forecasts, improved customer service, reduced inventories, more predictable and smoother production and more cost-efficient use of manufacturing capacity. Realizing any benefit in any industry, though, requires process improvements as manufacturers ascend from forecasting to demand planning to demand management and then collaborative planning with customers. Toward Demand Management Applications start at the critical first step of generating a forecast based on historical data, which feeds the sales and operation planning process and serves as the organizational guide. "Everybody is doing some kind of forecasting," says Larry Lapide, vice president and general manager, supply-chain strategies, AMR Research, Boston. He estimates 50% to 60% of the Fortune 1000 are running a forecasting application. The most prevalent forecasting tools across industry, though, remain spreadsheets. Steen Jorgensen, planning solutions manager for software provider Intentia U.S., Schaumburg, Ill., points out that automated forecasting does not equate to demand planning. "The calculation of the forecast is just a minor part of that." Demand-planning applications enable a consensus forecast, one in which multiple constituents share and reconcile forecasts to create a single plan. Demand-planning algorithms incorporate future-oriented data -- promotions, distributor incentives or even climate projections. Constituents can slice the plan based on their functional needs, such as by brand or sales territory. Demand-management tools further add capabilities such as the ability to balance supply with demand, influence demand through pricing and promotions, view supply in real time and respond to unexpected demand signals, and seamlessly integrate operating systems -- such as financial planning, material replenishment, production scheduling -- and revise as necessary. "Demand management is not just about the planning piece," says Henry Bruce, vice president, strategy and marketing, Industri-Matematik International, supply-chain solutions provider, Mt. Laurel, N.J. "It's about my ability to execute and how I tie the two pieces together so that I have a feedback mechanism and I can log week by week, month by month, my assumptions, how I am executing, and what changes I need to make." Manufacturers are at varied stages of demand-management evolution with little clear-cut borders, and software capabilities, too, fluidly range from narrow forecasting tools to product suites that address planning to collaboration. Vendors include firms with best-of-breed forecast and planning platforms, supply-chain management companies whose broader solutions extend out to the customer relationship, and ERP vendors claiming integrated solutions and bolt-on benefits for their user base. Costs range from hundreds of dollars to hundreds of thousands. Start With A Plan Telltale signs of poor planning are disproportionately high inventories and/or low service levels. For industrial markets, that could be 30% on-time delivery, says Carlson, while a poor consumer-packaged goods (CPG) performance might be 88% to 94%, "which means they're just walking away from 6% to 12% of their potential customers." Improved demand planning helps maintain lean inventories while supporting customer demand, and it opens cost-saving opportunities that cut across a supply chain. Lynne Taylor, manager of product marketing, demand management, JDEC, says, "With every little bit of improvement you can get with your forecast, you're going to drive your costs down all the way through your supply chain -- through production, distribution, scheduling and overtime." Expansion of benefits is evident at Hyundai Motor America, Fountain Valley, Calif. In 1998 Hyundai's service parts unit implemented Demand Solutions from Demand Management Inc. George Kurth, parts director of supply chain and logistics, says prior to that he had 30,000 different part numbers in three warehouses and demand history back to 1986; planners could not cope. "We knew we needed to get something to automate the forecasting so they could just work exceptions." Now, each month Demand Solutions digests 41 months of demand history, executes several simulations for the previous six months, and then finalizes a future forecast by part number for four warehouses that is rolled into a requirements planning module. That module creates an order plan for about 50,000 parts, and a legacy system sends the parts order to Korea. Kurth says the system produces a smoother and more reliable forecast, saves labor, and "We've taken a couple months of supply out of our inventory since we bought the product. We're running right now 2.8 months of supply and a 98.3% fill rate" based on complete line order fill. Hyundai is pulling inventory information and point-of-sale (POS) data from dealer systems and plans to refer orders between dealers, creating a virtual warehouse. Kurth says Hyundai also will debut a program that combines Hyundai data with local dealer signals to indicate when a dealer should stock parts as well as capabilities to more closely track inbound shipments from Korea and returns from dealers, giving greater visibility of supply-chain inventory. Similarly, GM Brazil, three months into an implementation of Finmatica Mercia's MerciaLincs application, was beginning to more accurately calculate a forecast, predict safety stock and establish signals for GM to build service parts. Denio Nogueira Jr., planning and materials management manager for parts and accessories operations, says the primary data driving the system is final customer demand through the dealer network, a collaborative process made easier because GM Brazil has implemented a vendor-managed inventory system with dealers. About 75,000 part numbers go into MerciaLincs, and it currently forecasts 18,000 parts with sales history. The system will forecast more parts as more histories are built. "Since [implementation], the accuracy ratio is something around 94%, which we consider very good," Nogueira says. He says the division is working to improve the mapping of old items to new items, generate forecasts on a weekly basis rather than monthly and improve forecasting of part life based on application and usage. Planning Practices As with any IT implementation, there are challenges. Upfront hurdles involve extracting data nestled in disparate data warehouses, legacy systems and fiefdoms and then mapping and refining data to feed demand plans. Organizations then need discipline to follow the plan. "Just getting all the data in one place is a huge task," says Mike Campbell, CEO of Demand Management Inc., St. Louis. "Typically when we walk into a company that's not doing a good job of planning, we find it very fractionalized. We find that each department has its owns scheme, and each interest group has its own agenda." Additionally, many believe their data is too complex to move from spreadsheets into more rigid software formats. "Certainly there are pieces of knowledge that a demand analyst or forecast analyst might have that the software has a more difficult time replicating, but that in large part is an exception," says Lori Mitchell-Keller, senior vice president of market strategy for Manugistics Group Inc., Rockville, Md. The majority of products have steady, available demand information, she says, and software, like Manugistics Demand, automates the process for those products but also flags the exceptions "so they can move toward an exceptions-management environment as opposed to having tools for every single item." Todd Inlander, CIO of Fleetwood Enterprises, a maker of recreational vehicles and manufactured housing, says demand-planning tools are more inflexible than a spreadsheet, which can be a hurdle for individual users, but downstream benefits require upfront work. "That's a common theme of all my problems -- standardization -- and standardization comes at a cost as does consistency of information." The motor homes division of the Riverside, Calif., manufacturer implemented Demand Planner from Demantra Inc. and purchased through JDEC, and Inlander says the business unit has first focused on improving the timeliness and reliability of retail information it pulls. Due to difficulties in getting POS information from hundreds of dealer systems, forecast accuracy had not initially improved, Inlander says, but forecasting time has been cut from 75-90 days to 30-45 days. Rolling out a Web portal to capture dealer information and, eventually, electronically integrating with dealer systems should further accelerate the cycle and hone the forecast. St. John Knits International Inc., a high-fashion manufacturer for retailers such as Nordstrom, twice a year sets new fashion lines. Buyers review products and project orders for each line, and then St. John prepares a forecast. St. John chose not to map legacy data to its new system, Movex Demand Planner from Intentia International AB, opting instead to more swiftly implement for a new selling season with fresh data. "In a couple of years time we'll have some pretty relevant data," says Scott Huckleberry, senior vice president and CIO, which will enable St. John to compare customer responses to past lines by style, cut and body type and eventually trigger production and support manufacturing-capacity decisions. The St. John forecast currently is used to order long lead-time materials, such as yarns. Four make-to-order plants don't "cut, sew or do anything" until they get an actual order, says Huy Vu, senior vice president manufacturing. Orders "consume" the earlier estimate in Demand Planner, which then redistributes estimates to prevent duplication. Although deploying on time, Vu says the system still required upfront work and training because of the SKU-level detail being extracted. For example, the knit division produces 800 styles, 8,000 SKUs and about 400,000 total pieces per line. "The more information you want to pull out and the more flexibility and functionality you want, the more effort it is to put in data at the right level and in the right places," adds Huckleberry. Karen Laucka, solutions marketing for demand-management products, i2 Technologies Inc., Parsippany, N.J., says another major challenge is "getting a company organizationally ready to do best-practices demand planning." Organizations need to define their processes, motivate individuals to comply and contribute the truest numbers for the good of the organization, and regularly monitor individuals' inputs, she says. Otherwise an inaccurate or biased demand plan emerges that leads to increasing variability, uncertainly and inaccuracy as it goes downstream. And that, in turn, reduces buy-in to the plan. Customer Collaboration Some manufacturers are tackling the challenge of integrating real-time customer data, such as POS information and in-stock inventories, which, at the very least, reduces the frequency and severity of customer surprises. But even counting CPG markets where mass retailers have a vested interest in sharing data, true collaborative planning appears to be more buzz than business practice. Internet-based tools are helping collaboration, but, estimates AMR's Lapide, only 5% of the Fortune 1000 are using collaborative packages and getting real-time information from customers. "In an academic theoretical environment, [collaboration] works, and it is what we should all be striving for," says Laucka. "Most companies still aren't there yet. While the manufacturer and retailer have to work together, there is a hint of adversarial nature there because they're both trying to maximize their profits. That age-old trust issue of getting two large companies to work together is a real challenge." Barry Boehm, partner, integrated supply-chain planning practice, IBM Business Consulting Services, St. Louis, says to first approach large customers, and "if you the manufacturer can pass through a lower cost position to your customer, then the willingness to work with you in a collaborative way goes up significantly."
|Demand planning and forecast management|
|All supply-chain management|