To compete profitably in today's global marketplace, companies need to respond to the dynamics of market demand in a timely manner. In theory, that may seem a simple proposition. But in reality, it's extremely difficult given the complexity of global supply chains coupled with increasing pressure to cut costs, decrease inventory, and raise margins. The reality is that while change is inevitable, growth is optional.

To achieve that necessary competitiveness, leading companies are seeking to eliminate manual intervention, expensive duplication, and waste from their day-to-day operations. For them, intelligent supply chain management has become the major weapon in maintaining profitability in a competitive but rapidly changing environment. The watchwords for success have evolved from "How can I cost-effectively make product and push it out to a mass market?" to "How can I most profitably move product to meet market demand?"

Those organizations that redesign their processes to be more responsive to customer demand will be able to grow their businesses faster than their less agile competitors. To that end, successful companies are reengineering their supply chains to deftly balance, align, and seamlessly integrate demand, marketing, sales, distribution, production, and supply in search of sustainable and competitively superior performance gains.

Traditionally, performance improvement has been based on internally focused programs designed to improve operational efficiencies. These programs have concentrated on measurements and controls within individual departments or functions. Although they may be able to generate incremental improvements, they fall short of achieving major change and competitive advantage. That can only be accomplished by (1) understanding the dynamics of process changes within the organization; (2) integrating the activities and optimizing the metrics supporting the processes; and (3) strategically aligning participating functions and organizations within the structure of the market in which they operate.

Understanding the dynamics of process change requires knowledge of the three main demand-management structures. Common to every business, they govern the behavior and determine the processes, activities, and linkages among the participants in a market economy:

• Demand-creation structure (the demand chain). This includes new product research and development, market research, marketing, and sales. Here is where the bulk of a company's intellectual capital and operating expenses reside. There is no flow of material within this structure, however, and the cycle moves on once the demand-creation process results in an order.
• Demand-fulfillment structure (the supply chain). Among these activities are procuring materials, creating the product, and shipping it to the customer to fulfill the order (actual demand). This structure accounts for the bulk of physical capital investment and operating costs. Efficiency in production determines a product's gross margin, while typically 10 to 15 percent of total product cost lies within distribution. Managing the demand-fulfillment processes is crucial to maintaining competitive cost and service advantage for sustaining profitable growth.
• Demand performance structure (the value chain). This structure embraces cost accounting, profitability, share price, and company economic value. These activities all rely on the effectiveness of the demand-fulfillment and -creation processes. They determine how much can be invested back into demand creation and demand-fulfillment cycle activities to support corporate growth objectives and return value to investors.

Integrating and optimizing corporate business processes across the three demand-management structures is not easy because most businesses are chaotic systems composed of many constituent parts. Each has its own momentum, metrics, and matrix of driving forces at the points of process interaction. These are the points that act separately or together to cause or influence change, creating metric tradeoffs among the various organizational functions.

To illustrate, there is a point in the marketing process (demand creation) where a promotion is developed that will create a deviation from historical shipments (demand-fulfillment results). Historical data, however, is often the key performance metric used by both production and distribution (demand fulfillment) to forecast and plan their operations and requirements for the promotion. That is why corporate dissonance occurs; the tactical metrics of one constituent part directly conflict with or change another.

Let's take a hypothetical manufacturer as an example, one with a complex portfolio of volume products in a market that can be promotional or seasonal. Driven by unit-cost metrics, the manufacturing manager wants each line run to be as long as possible and in a cost-effective sequence. He desires to make the most of available plant resources and minimize costs associated with changing from one line to another—the traditional "push" system. This means producing a lot of stock as inventory.

But that approach upsets the distribution manager who, governed by inventory "turn" metrics, wants to minimize his warehousing and inventory-carrying costs. By contrast, he would like to see product produced and shipped as it is ordered—the new "pull" system theory that requires frequently changing production lines.

The optimal solution lies somewhere between the two at the "push/pull boundary" where the tradeoffs occur and could be minimized. After all, the overall aim is to grow the enterprise most profitably.

The reality is that both managers are right. Very few companies can produce all of their products in the same day in the volume required to meet a day's consumption, let alone transport everything on that same day. In fact, it is physically impractical—if not theoretically impossible—to implement a truly "pull" system, especially given economies of scale and product movement. The trick is to be able to simultaneously compare all possible scenarios and to select the most profitable balance between time-sensitive production cycle resources and response-sensitive customer distribution cycle resources. A plan needs to be developed that takes into account all factors across the supply chain.

Demand-creation results are highly unpredictable, volatile, and influenced by a wide array of current marketing and sales actions as well as by the market environment. Demand performance, on the other hand, stems from the successful execution of business operations. This leaves demand fulfillment as the linchpin, the central component that can make the difference between market success or failure. In short, success increasingly depends on a business's ability to profitably respond to changing market opportunities within its supply chain and in conjunction with its partners' supply chain.

Today's production and supply chain managers, however, face a daunting task. They must balance the tradeoffs between planning and execution, production supply and market demand, inventory cost and service performance, and transportation cost and delivery time. What's more, they must synchronize these decisions within multiple business cycles throughout the channel. As managers endeavor to resolve this conundrum across their supply chain, the strategic deployment of information technology to support faster, informed, and more effective management decisions becomes all the more critical.

The ability to profitably respond to and meet customer demand partly depends on an accurate forecast. This holds true much to the chagrin of all operational managers, who know that the only thing guaranteed about a forecast is that it is guaranteed to be wrong. Businesses typically develop multiple forecasts by different organizational functions based upon a variety of data sources. More importantly, these are driven by different performance-measurement requirements. Want to know how many different forecasts your organization uses? Count the number of spreadsheet software licenses in your company and multiply by two—at least.

Consider the multitude of forecasts in most organizations:

• Promotional forecasts, which originate in sales and marketing, are measured on units of sales and result from multiple data sources (sales analysis, external data, promotions, market research, and so forth). These forecasts are synchronized to the time of the sale or promotional event and nearly always deviate from past performance (historical data).
• Operational forecasts usually are derived from statistically based programs using stock-keeping unit (SKU) measures analyzing historical sales or shipment data sources. These forecasts are synchronized to shipping or production dates or to schedule requirements. They generally fail to consider the timing of the event causing the order or the capacities at the locations charged with fulfilling the order.
• Financial forecasts are based on revenue measures (billing or cash receipt data), modeled against financial reporting requirements. These forecasts typically are aggregated by non-physical reporting structures, for example, product, region, country, or market.

Traditional statistical forecasts—whether promotional, operational, or financial—have attempted to predict what will be sold within a given period of time. For many reasons, time-series models have not produced the accuracy needed for supply chain optimization. The first problem is the high variability of demand closer to the point of sale due to non-recurring events, such as new product introductions, pricing actions, and promotions (not to mention weather, construction, and consumer caprice). The second difficulty is that the forecast rarely considers either the cost to fulfill or the capacity to fulfill—two details that affect profitability.

Most companies are able to develop a reasonably accurate aggregate forecast. They have a good general idea of total market volume that will be sold in a year—and, consequently, how much to manufacture. What most do not know, however, is how much of a particular line will be sold in a particular place on a particular date. Unfortunately, looking at what has happened in the past will not help much.

Take consumer goods as an example. It's possible to predict with relative certainty what a store will reorder. While store scanners track units at the UPC (Uniform Product Code) level, the shelves are generally planned to have a "pack out" quantity of a case. As UPCs are sold, the shelf is depleted to the minimum facing and the store reorders another case. With 30,000 to 50,000 items stocked and sold each day, a retail outlet could hardly be efficient using a unit pick replenishment process. That explains why most retailers have implemented a case pick environment. Furthermore, since little or no backroom storage exists at the store level, the order forecast will naturally be by the case.

The real forecasting issue, though, is determining when the store will place its replenishment order. The rate of sales (ROS) for each product, in conjunction with the balance on hand (BOH) calculated against case size (order policy), will determine the timing of the replenishment. Unless forecasting models understand and consider the rate of product sales through a planning location (rate of sale analysis, or ROSA), the policies controlling reorder unit size will prove erratic. That's because of the factors causing ROS to vary (demographics, pricing, promotional events, seasonality, merchandising, and so forth). As a result, the Mean Average Percent Error (MAPE) will remain relatively high and inconsistent, especially at the manufacturer level, for individual stock-keeping units.

What's needed is a shift from traditional DRP and statistical forecasting (predicting "what-will-be" based on what was) to supply chain optimization and dynamic demand planning (estimating "what-will-be" based on what's happening to cause history to change).

Another shift from tradition is mandatory if all supply chain participants are to reach their common goal: To reduce the cost of total product flow from the point of origin to the point of consumption, and to accelerate the response time to meet changing customer demand. In striving for that goal, they must somehow overcome the conventional approaches of reacting to marketplace uncertainty—the buildup of inventory to respond to orders on a "push" basis. Or, worse, they invest significant capital to improve manufacturing "flexibility." Based on ballpark forecasts, an organization will deploy product in anticipation of the customer order or implement costly, frequent production schedule changes.

The inherent difficulties of traditional supply chain planning are the fixed nature of supply/demand source relationships, sequential flow of information, and dependence upon the accuracy of statistical forecasts based on historical data. Information is passed from demand planning applications to DRP applications. From there, it moves on to master production scheduling and line scheduling applications, and finally to the material requirements plan. This sequential information flow results in time delays from level to level, great uncertainty, and multiple suboptimized functions—each with inefficient and uneconomic time, inventory, or capacity buffers against demand uncertainty.

This scenario is referred to as "monotonic" demand planning. Monotonic reasoning is defined as reasoning that fails to reconsider its conclusions despite the existence of new information. Essentially, customer demand is met on a purely reactive basis. It's no surprise then that the astronomical costs inherent in running an enterprise in a monotonic manner no longer can be tolerated. Competitive companies instead are looking to dynamic or integrated market response systems for a solution.

This is the central concept of synchronous logistics. It's the dynamic process of modeling the different business cycles, costs, and constraints at each level, considering the key drivers of demand and replenishment needed to meet profitability objectives. Synchronous logistics planning requires the integration of all forecast processes (promotional, operational, financial) into a single system with multiple views as well as model integration to identify discrepancies for arbitration and reconciliation. Lack of understanding of the interdependent forecast process and synchronization with manufacturing often leads vendors and clients to position demand-planning applications in competition—rather than in cooperation—with one another. This situation underscores the need for collaborative, dynamic decision-support systems.

Synchronous logistics also determines the information requirements. It establishes the points of interaction and systems required to synchronize logistics activities among the different cycles to concurrently and economically match product flow to consumption. In effect, it provides integration, optimization, and "visibility" across the supply chain so that all parts of the chain can see the impact of decisions and demand at any point. The objective of supply chain optimization is to create a whole greater than the sum of its parts.

Optimization is based upon a realistic, holistic view of the supply chain free from the information, material, or functional boundaries that inhibit planning. The goal is to dynamically determine the best response of the network as a whole. This is in stark contrast to the response-based, monotonic approach that views problems sequentially on a location-by-location, function-by-function basis.

Synchronous logistics also requires developing a cross functional, optimized global view that "enables" local management to keep in line with total business objectives. Once achieved, local responses to exceptions can be developed, quickly bringing operations back into alignment with the enterprise's strategic goals, objectives, and key performance indicators.

By adding optimization technology and decision-support capabilities to supply chain planning, companies can model the entire supply chain simultaneously. The model can consider costs, constraints, business objectives, and other variables across supply chain structures, processes, and activities. Optimization first considers all feasible options. It then determines the most profitable resource flow and measures needed to achieve desired results. If managers can view the planned production schedule and sequence of production prior to execution, the most profitable response to meet demand plans can be simulated and optimized.

By examining what has occurred and considering any current actions that may cause change (for example, new product introductions or promotions), managers can estimate demand fairly accurately. The next step involves identifying the typical mean average deviation (MAD) from forecast. With this information, production, schedule, and inventories can be simulated to meet the high-side deviation, showing resource/inventory cost and risk. Similarly, the low-side deviation can be simulated, showing customer service cost and risk. The result of this analysis is an optimal production schedule and inventory plan, with upper and lower control points to respond to deviation from the estimated demand forecast.

For those products already in planned production, for example, it makes sense to build to the lower side of the forecast as there is less service risk due to lower cost availability. For those products not planned for production, it's better to build more inventory because the inventory cost/risk usually is less than the production-schedule—change cost necessary to make those products. By setting upper and lower control limits to the plan based on potential forecast error, a business can dynamically determine its level of safety stock and service/cost risks. And in doing so, it can manage the flow of product, not just point inventory.

Furthermore, when orders are received, the planning model can identify gaps between the optimal plan and the actual order. As long as the order falls within the plan's upper and lower control limits, the business can maintain a profitable response to that order. Orders that fall above the tolerance threshold immediately will be flagged as exceptions. In such cases, inventory must be obtained from another location in the network. Conversely, if orders fall under the tolerance level, the result is excess inventory at that particular location. Quantifiably measuring the performance of the process dynamically and identifying any discrepancies can highlight the root cause of any problem. Furthermore, it can determine the most profitable solution considering all concurrent options, including rebalancing network deployment errors prior to pushing the problem back to the next level.

By dynamically rebalancing network inventory as orders are received, companies can use excesses in the network to satisfy shortages. This technique balances and manages the economic requirements of push and pull boundaries throughout the supply chain. The network becomes optimized as total network inventory is used as a buffer to minimize the impact of forecast error. Under the traditional DRP approach, when a shortage is experienced, demand is pushed back onto manufacturing.

The information architecture that supports an integrated supply chain environment has to recognize the different requirements across the channel. With visibility into supply chain requirements for the entire channel, companies can direct the flow of products to the actual point of consumption. This requires an enterprise system that integrates all of the local data collected from the various functional business units and systems.

The problem is that these systems are all different. Retail enterprise systems are based on store operations and distribution center requirements; wholesale enterprise systems are based on buying, stocking, and selling of products; manufacturing enterprise systems are based on the actual purchase, conversion, and distribution of product sourced from raw materials. Different systems have different objectives—both operationally and financially. For supply chain optimization to occur, however, all of these systems must "interoperate" and communicate with one another. In short, everyone must be able to view the results to support collaborative decision making.

From an enterprise perspective then, supply chain optimization requires integrated market response systems, not traditional ERP systems. The traditional ERP system focuses on manufacturing and financial applications. There is little or no support of sales and marketing (demand creation) activities. Although traditional ERP systems concentrate on managing the order-to-cash cycle, they provide only limited information on how product is delivered, deployed, and planned. What's really needed is an enterprise system that integrates all of the relevant data and supports concurrent supply chain planning and decision making.

The planning challenge is to develop an intelligent supply chain management-information infrastructure. Such a planning tool could capitalize on the data from the different enterprise systems, communicate information among all the functions across the channel, and develop optimal key performance indicators. It must have decision support systems that can analyze millions of bits of data, optimize those activities, and determine what logistical actions need to take place, and when. Finally, it must be able to determine the actual transaction instructions, support decisions to manage the flow, and then be able to measure the operational execution throughout the entire supply chain.

With an intelligent supply chain management infrastructure, companies can begin to match information and product flows to actual consumption. Such a system could collect, optimize, and visualize demand-change requirements across the channel. It could foster new trading relationships and implement integrated enterprise and decision-support systems, In short, this structure would produce a demand-driven supply chain system.

The new paradigm in supply chain management incorporates dynamic distribution planning. It entails looking at the entire supply chain holistically and responding to demand change optimally, while considering business objectives. This approach effectively weighs tradeoffs, showing where it is more cost effective to incur higher shipping costs, for example, than make more expensive changes to the production schedule.

By dynamically rebalancing inventory and taking advantage of deployed aggregate stock, companies can quickly respond to demand and buffer themselves against forecast error at distributed locations. Most importantly, they can use inventory and production capacity much more efficiently across the entire enterprise, thereby minimizing additional capital costs and improving profitability at a faster rate. Obviously, it's far more cost effective to buffer uncertainty with information than with inventory.

The ability to execute the dynamic distribution planning that leads to supply chain management success depends on two critical mindsets:

1. Rather than view the supply chain as a series of customer-supplier interactions, participants must see themselves as trading partners who share a common goal of satisfying end-user demand by responding quickly at the lowest cost or most profitable option. Called "partner focused logistics," this approach considers the economics of every participant in the channel.
2. Management must recognize that replenishment planning begins with a determination of the sales rate beginning at the point of consumption/sale. The basic premise of Distribution Resource Planning is that if we can forecast sales at the end-user level, all demands down the channel can be calculated and subsequent replenishments pulled through the channel in response to actual demand. Optimization logic improves the calculation to provide a more responsive system that is tolerant to variations in forecast accuracy considering time-sensitive resource constraints and costs.

In the end, trading partners can move product at significantly lower cost if they all work together. For this to happen, though, the entire structure of the customer-supplier relationship has to change. That means forging relationships with potential competitors as well as customers. Further, relationships need to be developed both inside and outside of the organization that are built more on trust and information sharing. Once these kinds of alliances have been established, the channel partners can take the initial steps to set cross-functionally optimized goals. It then becomes possible to unite functions and enterprises in a supply chain based on common objectives and "most profitable" product flow.

To date, management in most companies has relied heavily on intuition to make the critical decisions on flowing material through the supply chain in response to changing market demand. A paradigm shift has begun, however, from traditional, monotonic supply chain planning to dynamic, model-based planning and simulation systems. This combination of intuition (managerial experience) and quantitative analysis (management science) ultimately enables organizations to integrate supply chain data, visualize business flows, and optimize resources holistically. It lets them develop the key performance indicators and operations research that drive cross-functional behavior. This results in the achievement of corporate objectives and sustainable competitive advantage as the whole becoms greater than the sum of its parts.

Dynamic distribution planning optimization technologies are designed to enable synchronous logistics execution to keep production "in sync" with consumer demand. Fueled by the availability of more robust information and application integration technology—and fostered by new cooperative trading relations—these model-based technologies will drive supply chain management to the next millennium. And, in the process, they will help organizations achieve sustained profitability.