n the manufacturing business, the company that can most closely predict not only what to make-but also how much to make, and when-wins.
With thousands of products and millions of dollars in inventory, companies such as Reynolds Metals Co. must forecast sales demand as accurately as possible to avoid overproducing. The 29,000-employee maker of Reynolds Wrap and more than 1,000 other food-packaging products ranging from foil wrap to plastic containers has found that even a 1% improvement in forecasting can translate to
millions of dollars in savings.
Accuracy in forecasting is especially important for Reynolds this year, as the $7 billion Richmond, Va., company works to reverse a 3% decline in 1996 revenue-down nearly $250 million from 1995. To get back on track, the company is concentrating on shrinking inventory levels and improving turnaround to customers, who are simultaneously trying to lighten their own inventories.
Less Is More
The goal, according to chairman and CEO Jeremiah Sheehan, is to lessen internal inventories by more than $100 million annually. To this end, business analysts in Reynolds' packaging division use Logility Planning Solutions supply-chain planning software. The package, from Logility Inc. in Atlanta, builds forecasts based on customer sales patterns. Reynolds also uses it to control inventory and distribution. In a few months, the company's manufacturing facilities also will have access to the system.
The improved ability to control the supply chain has already cut overall
forecasting error rates from 15% to as little as 5%, and has reduced inventories significantly, says Hank Chewning, a senior business consultant at Reynolds who oversaw implementation of the supply-chain planning software when it was installed last June. It also has dramatically improved customer service. "If you have a short lead time, every minute counts," Chewning says. There's also an added bonus: The Logility system, which cost Reynolds under $500,000 in hardware and software, handles year 2000 data.
Logility Planning Solutions replaces an IBM 9672 mainframe-based forecasting system, installed in 1990 and developed by Logility's parent company, American Software Inc. With the old system, inventory and sales data was entered three times a week in batch mode. Once a month, business analysts dug through a 700-page paper report for the numbers they needed. Analysts couldn't tell what customers were buying whi
ch products, or respond quickly to changes. "There was no easy way to enter data either from new business or lost business," recalls Chewning. "If we needed to make changes to what was on a delivery truck, we needed that information as soon as possible. But there was no way to do that."
With Logility Planning Solutions, Reynolds can change deployments within 24 hours-or even less, depending on customer needs, Chewning says. Customers typically receive their shipments within 10 days of placing an order, giving Reynolds a leg up on rival food-packaging companies, who deliver in two to three weeks, adds Dave Seibert, a business unit manager in Reynolds' food-service packaging division.
Forecasting In Action
The Logility server software and inventory data is located on a Hewlett-Packard NetServer LS 100 server running under Windows NT 3.51; client software is installed on IBM Pentium desktop machines and IBM Pentium ThinkPad notebooks in the packaging division's headquarters in Richmond.
Users combine customer and product purchase history data by using a component of the software called Demand Planning. Seibert, for example, uses the software to keep close tabs on projected sales of aluminum foil and other packaging products sold to hospitals, restaurants, delicatessens, and bakeries. He then uses an Inventory Planning component to determine manufacturing capacity, and to find the optimal balance between inventory levels and customer service.
When a customer wants to start purchasing a new item, customer sales representatives pass the information by phone or E-mail to the business analysts, who enter the data into the system. They can view graphical representations of trends in a spreadsheet-like interface that charts historical sales of a given item by warehouse, sales district, month, and customer, for a three-year period. This lets analysts see not only how much product a customer orders over time, but also whether the customer consistently orders more product, say, in October. The g
raphs help analysts easily determine the months of highest demand. "It's a lot more fun looking at graphs than at numbers on paper," says Seibert.
When the forecast is complete, market and distribution managers view the inventory planning and forecast in the distribution portion of the software. If a customer wants a specific level of service, managers can determine manufacturing and distribution requirements to meet the customer's needs. "If you have a requirement for 10 cases and there are only two in inventory, the Logility system takes that constraint and determines if you should expedite an existing resupply order, or whether you need to make some more," Chewning says.
The Logility system interfaces to Reynolds' American Software order-processing and Data-3 Systems Inc. manufacturing systems residing on an IBM AS/400. Within a few months, Reynolds' manufacturing plants in Beacon Falls, Conn., Grottoes and Richmond, Va., Louisville and Mount Vernon, Ky., Appleton, Wis., and Etobicoke, Ontario, wil
l be linked to the system to create a closed-loop distribution system. If more product must be made, the plant materials managers will be able to log on to the system, view the information on their PCs, then order the run. The manufactured products are sent to a central distribution facility in Richmond, and from there to one of 10 public warehouses at locations across the United States, Canada, and Mexico.
Not All Fun And Games
Implementation of the new Logility system had its difficulties. Though Logility would like to offer its product as essentially a turnkey package, Reynolds still needed the vendor to tailor the software, adding levels of detail such as slicing up records by business unit and product groups. "We purchased the software in February, and got it in June," says Chewning. "The tailoring process took a long time and cost us a 10% surcharge." Nor was the new software particularly easy to learn. "It took a few weeks to get familiar with it," says Seibert. "I knew what I wanted to
see, but I spent time just trying to find it."
Such difficulties are to be expected, not only because supply-chain planning software is still in its infancy, but also because constructing an accurate data model of business processes is no trivial task. "It takes some time to gather all the marketing, sales, financial, and manufacturing processes, then construct the dimensions of the forecast variables," says Greg Gerard, a senior analyst in supply-chain management at Advanced Manufacturing Research, a consulting firm in Boston.
Despite the potholes, Reynolds hopes to take its forecasting processes a step further later this year through a secure home page on the Web. Sales representatives in the field will be able to tap into the company's home page to view forecasting data. When a customer purchases a new item or changes the forecast on an existing item, the rep can make the change immediately. "We'll be able to look at it, review the change, add it to the forecast, and make a decision about supply
right there," Seibert says. "With more people looking at the forecast, we'll constantly improve it."
Seibert eventually hopes to get customers into the act. Using a Web browser with ActiveX technology and a password supplied by Reynolds, customers could access Web-based views of their own purchasing patterns, thereby gaining better control of their inventories. The result? A win-win situation for customers and Reynolds alike. Says Seibert: "It's another way to set us apart."
Back to News in Review
Send Us Your Feedback
Top of the Page
n the manufacturing business, the company that can most closely predict not only what to make-but also how much to make, and when-wins.
