Companies live or die by the Almighty Plan. From ordering raw materials to delivering finished product, everything is tied to the best possible forecast of demand. Yet consumers are a notoriously unpredictable lot. Who can say which new product or promotion will take off, and which will die on the shelf? In the end, planners must rely on a combination of experience, technology, knowledge of markets and just plain luck.

That said, leading companies minimize the luck factor through a series of best practices that draw on the latest thinking in management techniques and a wide array of technology tools. Here are 10 of them, culled from conversations with leading supply-chain executives, analysts and vendors.

1. Collaborate internally. Think this part is easy? Think again. Bob Moffat, head of IBM's Integrated Supply Chain organization, has been quoted as saying that internal integration is the single biggest challenge that companies face, when overhauling their supply chains. And the business world still has more silos than the Kansas wheatfields.

Noman Waheed, senior manager of the supply-chain service line with Accenture in Chicago, calls for an integrated sales and operations planning (S&OP) process that can address mismatches between supply and demand. That means regular meetings between managers hailing from all the major functions - manufacturing, marketing, procurement, sales, fulfillment, logistics, finance. IBM has had particular success in this area, says Steve Buckley, manager of supply-chain analysis with IBM Research in Yorktown Heights, N.Y. Despite its enormous size, the company has fostered a culture of collaboration through the use of Lotus Notes, e-mail and instant messaging.

To Larry Lapide of AMR Research in Boston, Mass., an effective S&OP process cuts across discrete functions and embraces plan creation, collaboration around a forecast, and management of the follow-up. If it's done right, the process should yield lower inventories, reduced operating costs, better customer service, and increased profitability. Samsung Electronics has dismantled its organizational silos through the creation of a "command center" which combines sales and manufacturing business processes, says Adeel Najmi, supply-chain solutions executive with Dallas-based i2 Technologies. The process took nearly four years, but a multi-pronged strategy, including planning and profit optimization software from i2, has helped Samsung to become the world's most profitable manufacturer of consumer electronics, according to Najmi.

2. Collaborate with external supply-chain partners. Again, it seems like a no-brainer. Aren't supply chains collaborative by their very nature? True enough, but real collaboration - the kind that drives competitive supply chains - means more than an arm's-length business relationship.

Start with the most obvious supply-chain partner: the customer. Lapide says a workable S&OP process will incorporate accurate "downstream data" about the customer, its own customer and the ultimate consumer. Point-of-sale (POS) data, the kind generated by barcode scanners at the checkout counter, are one important source. But Lapide says companies must also know the location and movement of inventory at points throughout the chain, including customers' distribution facilities, in order to determine how much product to make. Harley-Davidson receives regular reports, sometimes daily, from its 1,000 aftermarket dealers, according to Lori Mitchell-Keller, senior vice president of market and technology strategy with Manugistics in Rockville, MD. Dealers systematically inform Harley of what's been selling and what they expect to sell. The result has been a 25 percent increase in product availability, even with a 10 percent rise in the number of individual items stocked.

Companies must also be in constant touch with their upstream suppliers, in order to understand manufacturing constraints and the amount of product they can realistically expect to receive. Such links have been missing from traditional S&OP systems, Lapide says. Andy Carlson, vice president of supply-chain management with PeopleSoft, Inc. in Denver, urges companies to "take a holistic view of the supply chain." Failure to account for the global network will lead to sub-optimal decisions, he says. Graftech, a manufacturer of carbon and graphite products for electric furnaces, was letting its individual plants make decisions about production. By centralizing the decision-making process and adopting lean-manufacturing practices, it cut inventories by 40 percent and lead times by 20 percent.

3. Measure the right things. You probably know how well you did in forecasting demand. (Assuming you're willing to admit it.) But do you measure the S&OP process itself? Many companies have no idea whether they're executing against plan, Lapide says. Did a promotion actually run? Was it on time? Did the supplier produce the promised amount of product? Lapide recommends five key metrics: forecast accuracy, variance to the baseline forecast (in other words, how it was adjusted to meet reality), actual demand versus budget, adherence to the demand plan by marketing and sales, and adherence to manufacturing, operational and logistics plans.

The trick lies in limiting the metrics. "We often walk into companies with 200 'key performance indicators,'" says Sri Aparjithan, principal with Celerant Consulting in Chicago. "You need to understand what's going to drive operational performance and have a direct impact on your financial statement." While there's no right number of KPIs, Aparjithan says, they should cover everything from top management concerns, such as overall sales and cash flow, to operations on the shop floor. Yet they shouldn't be so numerous as to inundate the company with numbers.

Simon Bragg, European research director of ARC Advisory Group in Cambridge, England, recommends that companies use the metrics laid out in the Supply-Chain Operations Reference (SCOR) model, developed by the Supply-Chain Council. "You can show management that these metrics work," Bragg says.

And don't forget the importance of data accuracy as a key metric. As Waheed points out, poor data will undermine the most elaborate planning process. Inaccurate bills of material are likely to result in endless headaches with suppliers and the manufacturing process, leading to crippling product shortages.

4. Integrate supply- and demand-planning tools. At the distant dawn of the modern IT age - that is, the 1990s-these were often separate pieces of software, as were the enterprise resource planning systems that served as the user's accounting and financial backbone. Supply-chain planning software was bolted on to ERP and usually required some form of code to make the systems work together. Newer versions are more integrated. According to Carlson, PeopleSoft combines transactional, analytical and planning databases into a single model. That allows for faster reaction and real-time decision-making, he says. Fastenal, a maker of industrial fasteners, can manage a complex supply chain involving more than 6,000 different products and 12 distribution centers serving 1,300 stores in North America. It replaced a batch-oriented framework with its current PeopleSoft system.

By synchronizing supply and demand, companies can become more competitive, says Bob Ferrari, director of supply-chain product marketing with SAP. Brown-Forman Corp., a large producer of wines and spirits whose products include Jack Daniel's and Southern Comfort, drew on an integrated series of tools from SAP to access data from various parts of the supply chain, as well as from its distributors. It now receives constant feedback from the execution side, helping it to plan production runs and generate bills of material to its suppliers.

Demand and supply-side planning might still be separate applications, but they blend better than before. To Lapide, they are part of an "integrated application architecture," aligned by an S&OP "workbench" which utilizes dashboards and scoreboards to track the overall planning function. Through such an approach, companies can better collaborate with outside partners, creating what AMR calls "the demand-driven supply network."

5. Use one-number, consensus-based forecasting. Or, as managers are apt to call it, "one version of the truth." No longer can companies support the luxury of dueling forecasts, generated by various departments with personal agendas. While planners need the perspective of marketing, sales, customer service, finance and operations, all of those pieces must come together in a single number. Otherwise, adds Bragg, "your collaboration is entirely meaningless." BP Chemicals used to take two months to produce a monthly plan, he notes, with various departments ignoring one another's input and devising forecasts based on their own needs and expectations. Such dissonance can be alleviated simply by altering management's expectations of those departments. Salespeople, whose incentives were based entirely upon the amount of product they pushed out the door, might now be judged as well on the accuracy of their forecast. The goal is to have a number that is trusted by all departments, says Bragg.

PeopleSoft's Demand Consensus application creates a framework for true collaboration throughout the distribution channel, says Carlson. By adding input from downstream partners, whose information is stored as historical records in order to assess its accuracy, companies can devise a more believable number for planning. F.W. Murphy, a maker of industrial controls, saw improved forecasting accuracy in the range of 20 percent, compared with the results from its previous statistical forecasting tool.

6. Use science and math in your planning process. Human instinct and experience are indispensable, but they can be helped along by a set of rigorous optimization formulas. That's especially true for companies with a large number of products, plants, set-up times and batch sizes. Bragg sees three levels of optimization among providers of planning software: manufacturing resource planning, known as MRP II; heuristics, or rules of thumb for managing production bottlenecks and capacity constraints; and optimized production planning, employing algorithms to determine the most cost-efficient sequencing of items on the production line.

The statistical approach can yield an economic order quantity (EOA), which takes into account such factors as the cost of a manufacturing change order and the product's lifecycle. (For high-tech goods, the latter can amount to mere months.) In the process, says Waheed, companies can determine how much safety stock they need on hand, a number that can be shrunk but will likely never go to zero for most industries. And they'll eliminate a substantial number of manual processes, creating a more nimble supplier in tune with market trends.

For those industries where unpredictability seems to come with the territory - in microchips, for example - companies can apply a new theory that Lapide calls risk-adjusted stochastic planning. Farmers and commodities traders are well acquainted with the practice of hedging their bets through the futures market. In high-tech, as practiced by Hewlett-Packard Co., supply-chain managers pre-commit a set number of critical components to production, without taking possession of inventory. They can then make quick changes to actual production when market conditions shift unexpectedly.

7. Don't forget people. Waheed stresses the need for "a well-defined organization." Employees should know whom to approach when a problem arises. Roles should be clearly defined up and down the organizational ladder. Accountability, too, must be clearly delineated. Managers are less likely to generate flawed plans if they know they'll be held responsible for the results.

SAP offers a tool which allows sales to be measured, not just on the amount of product sold, but on that department's contribution to the S&OP process, says Chris Foti, supply-chain solutions principal. All key individuals in the chain, whether inside or outside the company, can alter the forecast according to their unique perspectives. But each, too, is responsible for the final result - and the company's ability to plan with confidence.

Once planners' roles are established throughout the organization, top managers must resist the urge to meddle in the details. "Let the schedulers do their job," says Waheed. Executives should resist the temptation to schedule by the hour. They should instead provide general direction, while allowing those below them to translate the commands into daily plans. Although planning must be tightly integrated with execution, the two are distinctly different processes. Warns Waheed: "Do not use planning tools as execution tools."

8. Make use of event-management tools. It's the link between planning and execution in a truly adaptive supply chain, experts say. The latest iteration of event management is the "sense and respond" model pioneered by IBM researcher Stephan H. Haeckel in the late 1990s. He outlined a means by which companies can quickly adapt to the business environment, and fill in the gaps between supply and demand. His work was followed by the development of software tools at IBM and elsewhere, Buckley says. First to deploy the model was IBM's Microelectronics division, which has acquired more accurate information about demand through better access to customer data. Then, based on historical data, the company proactively analyzes what the customer really wants, instead of relying on a forecast developed six months in advance. Finally, using simulation techniques, the company can perform a risk analysis of its various options for meeting demand. Under the Microelectronics pilot, the business unit achieved a 4 percent reduction in inventory over a year's time, Buckley says. IBM's Personal Computer division plans to deploy the sense and respond model in the first quarter of 2004, allowing it to identify problems in the supply chain much earlier than before, he says.

Event management assumes that no forecast can be 100 percent accurate; it seeks to combat the inevitable error through quick response. According to Mitchell-Keller, the system can be deployed when damaged goods have been shipped to the warehouse, when the company has sold more than anticipated, or when production machines are down, to name but a few possibilities. Some incidents can be cured through automated resolution, with the system triggering a reorder, for example, in the case of low stocks. Such responses are based on business rules established by programmers. Most of the time, however, companies prefer to utilize event management as a warning system, with humans taking action. "The over-arching concept is being adaptive," says Foti. "It's no longer looking backward. It's looking two steps forward."

9. Get down to the details. The big picture is all well and good, but companies won't wring maximum profitability out of operations until they attend to the "small" things - like individual customers and SKUs. Supply-chain planning bears little worth if planners aren't shaping operations to favor those areas with the greatest return, and minimizing or jettisoning those with an unacceptably high cost to serve.

Aparjithan says forecasting ultimately must be done at the stock-keeping unit (SKU) level. Planning can either begin or end there, but it must occur at each level of the organization, with particular emphasis on individual products. Many companies, he says, focus solely on revenue forecasting without regard for the unit level, either in terms of price or volume. Proper SKU segmentation can highlight those products with the greatest volatility, where demand is seasonal or otherwise erratic. Gross margin and product lifecycle are also important factors to toss into the mix, Aparjithan says.

Profit optimization is another fast-emerging concept that supports better planning. According to Najmi, it helps companies to understand which customers and orders are most profitable, as well as which supply lanes are the least expensive for getting product to market. Users of such systems might alter their raw materials sources in order to minimize cost and make best use of limited capacity. Using i2's profit-optimizing software, steel producer Posco saved an estimated $249m, mostly in inventory reductions and reduced production cycle times, all within a period of six months, Najmi says.

10. Get a handle on logistics. The best forecasting and manufacturing systems can be undermined by an inadequate system of warehousing and delivery to the ultimate customer. You might have a superior product, but customers judge their suppliers when the truck pulls into the dock, or when they reach for goods on the shelf.

Good planning must be backed up by well-considered contracts with a stable of reliable carriers, says Bragg. Companies must work with their carriers to plan each day's shipments in the most economical way possible. Proper load building, with an emphasis on full truckloads, can slash transportation costs without hurting customer service. But it's difficult to achieve that goal, says Bragg, without centralized control of the transportation procurement process. A "global" view of logistics can help companies to decide exactly how many carriers are needed throughout their supply chains, and which ones are doing the best job.

New optimization techniques for transportation come at the perfect time for logistics managers. Increasingly, customers are asking for deliveries within narrower time windows, along with smaller, more frequent shipments. All of these demands threaten to boost the cost of shipping. Optimization software can help suppliers meet those requirements while keeping costs down, says Bill Pritz, vice president and general manager of Logility's transportation group in Atlanta. Opportunities for cost cutting include pooling, multi-stop truckload carriers and continuous moves. But only sophisticated software can spot those options in a high-volume operation, Pritz says.

Pritz divides optimization packages into three categories - strategic, for network design and sourcing decisions; tactical, for localized solutions such as pool points and carriers serving particular regions; and operational, for shipments that must go out in a matter of days. Together the modules allow companies to deploy transportation as a competitive tool. One maker of personal care products realized significant savings both in transportation and warehousing in its first year of using the Logility software, in part by consolidating orders from different departments for the same customers, says Pritz.