While annual consumer demand for a brand across the nation may be well understood, it is difficult to determine the right number of specific products to put in an individual store on a particular day of the year. Unpredictable factors such as weather, transportation delays, production problems, and administrative errors can all wreck havoc on supply and demand. But more fundamentally, promotions create massive swings in demand. A single promotion can dwarf the average weekly demand for a product. Suppliers are forced to carry unprecedented amounts of safety stock, or stay lean and risk being unable to fulfill demand. The first option raises costs for everyone; the second results in lost sales, and frustrates customers.

The CPFR Opportunity
Collaborative Planning, Forecasting and Replenishment (CPFR), a set of guidelines supported and published by the VICS Association, changes the rules, so companies throughout the supply chain can simultaneously lower costs and improve customer service. Trading part-ners share their plans for future events, and then use an exception-based process to deal with changes or deviations from plans. By working on issues before they occur, both partners have time to react. A supplier can build inventory well in advance of receiving a promotional order, and carry less safety stock at other times. A retailer can alter the product mix to reduce the impact of supply problems. In short, both sides win, and the consumer ultimately benefits from lower prices.

The CPFR process is divided into nine steps:

• Step 1 - Front-end agreement: Participating companies identify executive sponsors, agree to confidentiality and dispute resolution processes, develop a scorecard to track key supply chain metrics relative to success criteria, and establish any financial incentives or penalties.
  
• Step 2 - Joint business plan: The project teams develop plans for promotions, inventory policy changes, store openings/closings, and product changes for each product category.
  
• Steps 3-5 - Sales forecast collaboration: Retailers and suppliers share consumer demand forecasts, and identify exceptions that occur when partners' plans do not match, or change dramatically. They resolve exceptions by determining causal factors, adjusting plans where necessary. Steps 6-8 Order forecast collaboration: Retailers and suppliers share replenishment plans, identifying and resolving exceptions.
• Step 9 - Order generation/delivery execution: Results data (POS, orders, shipments, on-hand inventory) is shared, and forecast accuracy problems, overstock/understock conditions, and execution issues are identified and resolved.
  

Step 1 ensures that each company has an adequate commitment to collaboration, and that all parties are aligned around common goals. This front-end agreement might be reviewed on an annual basis. Step 2 applies good category management principles-borrowed from the Efficient Consumer Response (ECR) initiative - to create a joint plan for going to market. This would typically be revised quarterly, or semi-annually. What makes CPFR unique is that this joint business plan is used to control the day-to-day activities of manufacturing, delivering, and selling products. That's where steps 3-9 come in.

A core assumption of CPFR is that each organization will enter the details of the joint business plans into their on-line planning systems, and then share the results on a regular basis as market conditions change and logistical problems occur. Because each company may manage thousands of products distributed across thousands of locations, it is not feasible for planners to compare these plans manually and deter-mine which changes are significant. Instead, a specialized CPFR system exchanges and compares each value using thresholds that planners have set. If changes in one plan, or differences between them exceed the threshold, the CPFR system alerts the planner to the problem. Forecast revisions are exchanged on a regular-usually weekly-basis.

The CPFR philosophy is that if plans are "close enough", they probably do not require attention. Even when trading partners have identical objectives, differences in statistical forecasting or constraint-based planning algorithms will produce minor variances in plans. These are not significant relative to statistical deviations in demand, and safety stock will take care of them. CPFR technology is essential to identifying exceptions, because of the millions of product/location combinations that are planned, and because of the unique perspectives (product, location, and partner hierarchies) of each supply chain participant. Figure 1 illustrates exceptions that might be triggered when a supplier's forecast is compared with the retailer's.

Figure 1: Forecast-to-Forecast Comparison Exceptions CPFR Results

Only a few CPFR initiatives have published the results of their collaboration, but these have been eye-opening. Nabisco and Wegmans, for example, noted over a 50% increase in category sales. Wal*Mart and Sara Lee reported a 14% reduction in store-level inventory with a 32% increase in sales. Kimberly-Clark and Kmart achieved steady increases in category sales growth that exceeded market growth.

It may be surprising to see such dramatic sales increases linked to a "supply chain" initiative like CPFR. While improved in-stock levels can increase sales, they wouldn't normally be responsible for a 50% lift. Joint business planning (CPFR step 2) was behind most of these increases in the pilot projects. These increases are clearly not sustainable when CPFR is applied across the board; more telling is that in spite of the sudden sales increase, inventory did not increase in the Nabisco-Wegmans pilot, leading to higher inventory turns. Steps 3 through 9 of CPFR can take credit for keeping inventory under control in the face of sharply higher sales.

CPFR Process Synchronizes Planning
From a business process standpoint, CPFR defines how retailers and suppliers can synchronize their different planning functions. Retailers are focused on predicting consumer reaction to promotions, competitors, and product category changes, while suppliers usually concentrate on managing the level of inventory at distribution centers. The retailer's objective is to keep products in stock in stores. The supplier's objective is to create the most efficient production and replenishment process possible. These differences are reflected in each party’s sales and order forecasting processes.
• Sales (Consumer Demand) Forecast Comparisons Retailers produce very detailed sales forecasts, often including weekly (or even daily) store-level demand per SKU. Suppliers may also gather a great deal of intelligence about what sold from a syndi-cated data source (typically IRI or Nielsen), but they usually create only market- or account-level forecasts. The CPFR solution aggregates the more detailed sales forecasts from the retailer and compares the total with the supplier's number.
• Order Forecast (Replenishment Plan) Comparisons Often, retailers do not produce an order forecast at all. When retailers do produce an order forecast, it may include only base demand. Many handle promotional orders through a totally different process, tools, and personnel. Suppliers, therefore, don't often get an integrated view of the retailer's demand. A CPFR solution can improve this situation by providing a forum where replenishment order forecasts and promotional orders can be brought together and compared in full. It can also give the retailer better visibility to how the supplier makes changes to their order forecasts to meet demand.

CPFR Technology Fits Existing Applications
Putting CPFR technology into action requires sophisticated data processing and network technology. CPFR systems stand on their own but must interact with existing supply and demand chain applications in an enterprise. The points of contact depend upon if a company is a retailer or supplier.

Figure 2 illustrates how a CPFR solution can fit in with a set of existing retail applications.

• Systems responsible for merchandise planning generate promotions and sales forecasts.
  
• Distribution systems - including purchasing, warehouse management, or replenishment planning applications - can produce order forecasts, track shipments and report distribution center on-hand information.
  
• Store operations systems report store sales, store orders, and on-hand information.
  

Figure 2: CPFR in the retail IT environment

The CPFR solution collects planning and historical data from these retail systems, identifies exceptions, and forwards relevant data to suppliers for review and processing.

On the supplier side, a different set of applications contributes to the CPFR process as illustrated in Figure 3:

• Customer relationship management (CRM) applications help the sales force develop promotions and sales forecasts.
  
• Advanced planning and scheduling (APS) applications create optimal replenishment plans.
  
• Enterprise resource planning (ERP) applications produce and distribute products, based upon firm demand.

Figure 3: CPFR in the supplier IT environment

Figure 3 also illustrates how a CPFR solution could fit into the supplier's IT environment. The CPFR system compares the data it collects from the supplier's enterprise systems with the data it receives from retail partners. It generates any exceptions it finds, and communicates results with the retailer's CPFR solution.

The CPFR solution itself does not have to be deployed at each trading partner. Instead, several companies offer hosting services that run CPFR solutions for a trading community. All each company needs is access to the Internet; data is uploaded to the hosting service through a secure connection, and then planners access their company's view through a web browser. Hosting services allow companies to launch CPFR initiatives quickly and with minimum investment.

Conclusion

CPFR is rapidly becoming mainstream. One retailer already has 20 partners online, with announced plans for 200. Other industries, as exemplified by the RosettaNet high tech consortium, have adopted CPFR business practices as well. Innovative manufacturers are also beginning to do forecast collaboration with their own suppliers - resulting in end-to-end collaboration. Barriers to participation are low, and benefits are being demonstrated. Maybe all the buzz about trading exchanges is justified, at least once they take forecasting product demand and distribution into account through CPFR.