Managing Automation - December 2008 - (Page 20)

[ COVER STORY ] At the same time, widespread globalization of manufacturers’ suppliers and customers is throwing a wrench into demand planning efforts. As manufacturers look offshore for lower-cost production and supplies, supply chains and cycle times are lengthening. That means manufacturers must be able to accurately predict demand trends further and further into the future. “We’re seeing a need for manufacturers to send forecasts to suppliers almost 12 months in advance,” Deloitte’s Simrose says. “As you try to forecast further out, you’re likely to get more error.” Globalization also means manufacturers are working with new distribution partners, not all of which are ready to provide high-quality demand data. “As emerging markets represent more growth, manufacturers are moving to a build/procure anywhere model,” says Kelly Thomas, senior vice president for the manufacturing sector at supply chain management software provide i2 Technologies. “But aggre- gating all that data is becoming a nightmare. Even just getting it into some kind of credible form on a monthly basis is a big issue.” When home and office networking equipment supplier Linksys, for example, launched a globalization push in 2003, it saw forecast accuracy drop to 20% and inventories soar in part because it couldn’t get accurate demand signals from new global territories, says Worldwide Operations Vice President Mark Payne. Under Payne, the Cisco Sys- “Instead of demand planners focusing on resolving specific issues, they’re dealing with the data.” — P&G’s Mills tems subsidiary got the chaos under control by introducing a consistent, collaborative planning process and replacing spreadsheets with the Financial Sales and Operations Planning tool from Symphony Metreo Inc. Since then, the company has been able to cut inventory costs by 35%. Globalization also contributes to a proliferation of product variety, which, in turn, complicates forecasting. Manufacturers across vertical markets are finding that customers in different geographies demand different product features and packaging options. Tracking and managing demand for more and more SKUs down to the distribution center or even the store level presents an increasingly complex mathematical problem. In fact, in some cases, experts say, as manufacturers have sliced and diced their product mixes more and more finely — a specific product configuration for a specific retailer, for example — they’re left with not enough reliable data to make an accurate forecast for a given SKU. “Statistical forecasting requires large volumes of data, but as manufacturers offer hundreds of times more products, they are finding they don’t have enough data to forecast individual SKUs,” Simrose says. “The result is higher forecasting errors.” Particularly among consumer goods manufacturers, the rise in demand forecasting complexity is something of a self-inflicted problem. In order to keep retail customers happy and product moving through channels, CPG manufacturers have taken to mounting more and more sales promotions. While such promotions help drive sales, they also make predicting demand tricky since many promotions are one-of-a-kind or apply only to particular regions or stores. Recognizing Demand Forecasting Success ntil recently, one of Procter & Gamble’s biggest planning challenges was to come up with accurate yet detailed short-term demand forecasts. Like most manufacturers, the $83 billion consumer products giant wants to understand projected demand for each of its SKUs at each of its distribution centers, particularly within 40 days or so of shipment of goods. The problem is that as the number of P&G’s SKUs and distribution centers has exploded, the time it takes for planners to pull data together and calculate demand forecasts using traditional micro-forecasting and rule-based systems has begun to skyrocket as well, says David Mills, associate director of global business planning. Two years ago, as part of its Consumer-Driven Supply Network initiative, P&G decided to test a new approach to short-term demand forecasting. The company began testing the Demand Sensing application from Terra Technology, a 7-year-old supply chain software vendor that was a 2007 Managing Automation Company to Watch. Terra’s product uses fundamentally different math than most demand planning applications, which tend to extrapolate from historical order and other information to predict demand. Instead, Demand Sensing takes in current demand information and then applies pattern recognition algorithms to predict demand in near-real time. P&G has been using the Demand Sensing tool to generate daily demand forecasts for its Western Europe and North American fabric and home care, baby care, family care, feminine care, hair care, snacks, and cosmetics businesses. P&G, which also uses SAP’s Advanced Planner and Optimizer (APO) for demand forecasting, feeds daily order, shipment, and shipment forecast information into the Demand Sensing tool. The tool then searches for patterns in the data that help to predict demand by SKU down to the distribution center level. So far, Mills says, the new approach seems to be working. It has helped P&G improve forecast accuracy by 30% and decrease safety stock inventory by more than 10% in pilot tests. Next, P&G plans to feed overnight POS data into Demand Sensing along with current data feeds. That should help P&G make use of what is often difficult-to-interpret POS data and generate even more accurate and detailed forecasts. “With the inclusion of POS data, we will connect even closer to the consumption pattern and be able to improve even more in our accuracy and ensuing results,” Mills says. Recently, Mills says, P&G approved funds to implement Demand Sensing globally beginning next August. U ma December 20 2008 Photo courtesy: Procter & Gamble

Table of Contents Feed for the Digital Edition of Managing Automation - December 2008

Managing Automation - December 2008 Contents Take 1 Business Objects Chief Says Union with SAP Meets Objectives After One Year Yes, Emerson, Too, Is in the MES Market Infor Chief Puts Off IPO, Restarts Buying Plans Kronos Now Tracks Shop Floor Machines IQMS Rolls Out User Interace, Other Upgrades Notes Five Ideas for Demand Planning Building on the SOA Blueprint Innovation Now A Team Effort Lean %2B Technology = LEAN^2 Finding Flaws Before They Spread Product Scan Advertiser Index Next

Managing Automation - December 2008

For optimal viewing of this digital publication, please enable JavaScript and then refresh the page. If you would like to try to load the digital publication without using Flash Player detection, please click here.