Break on Through to the Other Side

Linking information silos helps Castrol better manage uncertainty in demand, which improves forecasts, reduces inventory and boosts service levels.

Companies Mentioned
Posted on Nov 03, 2006

Selling lubricants can be a slippery business, especially at a global company like Castrol Ltd. (Swindon, England). The company's European operations include multiple countries, currencies, languages, business procedures and enterprise resource planning (ERP) systems. Until a few years ago, systems for forecasting and replenishment varied from country to country, and data was trapped in information silos in production, scheduling, forecasting and inventory planning. As a result, there was virtually no collaboration between sales and marketing and supply chain in projecting future demand. Sales and marketing personnel tended to over forecast and supply chain personnel would be skeptical and filter the numbers. "There was no trust, no collaboration," recalls Alessandro Tenaglia, logistics manager at Castrol. Since safety stock levels were manually set based on intuition, and rarely adjusted to reflect changing conditions, forecasts were unreliable and pushed inventory levels off by a wide margin. Fast-moving products were constantly out of stock. Most slow-moving products had excess inventory. Planners spent a great deal of their workday expediting to compensate for poorly-derived inventory targets. To increase forecast timeliness and maximize service levels, Castrol needed to transform by linking silos of information, improving communication and automating the demand, inventory optimization and replenishment-planning process. In 1997, Castrol selected DPM software from ToolsGroup (Amsterdam, Netherlands) with the goal of improving service levels and reducing stock levels. It is used in two of Castrol's global business segments: consumer and commercial. The demand, inventory optimization and replenishment-planning system was implemented across Europe. It covers the supply chain from blended oils and packaging through to end user and it replaced 10 independent systems. ToolsGroup's DPM software bolts onto ERP and supply chain systems and relies on proprietary analytics and stochastic modeling to measure demand and uncertainties in the supply chain and set inventory targets. Rather than computing a finite forecast number, "the stochastic process creates a bell curve showing the range of all possible outcomes and their probabilities," says Jeff Bodenstab, vice president of marketing at ToolsGroup. The software, DPM release 3.14, was initially installed on personal computers, but Castrol later upgraded to DPM 5.0, a Windows XP server-based version of the product that integrates with other systems through flat files or XML. One positive of the DPM technology is interface flexibility. "This was key to enabling installation of a single, consistent technology throughout Europe despite the many ERP and satellite systems installed," says Tenaglia. "Every single country could configure input and output files to fit with local technology." In a perfect world, a consolidated database would provide a single view of Castrol's supply chain in Europe. However, creating a single database has proved to be impractical because operations in different countries use different item codes. Moreover, the effort required to keep a merged database up to date is incredibly time consuming. Not only are there a lot of stock-keeping units, the numbers change frequently as products are added and dropped. In fact, life cycles for many Castrol products don't last more than a year. Analytics within the DPM software define the relationships between inventory behavior and target service level for each SKU location, including intrinsic demand and supply uncertainty. The software statistically models up to millions of SKUs against many variables such as volumes, lead-times and lot sizes. The model takes into account a variety of demand/supply influences such as promotions, product phase in and phase out and product launches, expiration and shelf life. Modeling allows DPM software to dynamically generate optimal inventory targets and feed them back to the supply chain or ERP system. This gives supply chain, finance and marketing personnel a collaborative management process for setting global service levels as well as an optimization engine to support goals. The DPM system divides stock into three classes -- A, B and C -- from most in demand to slowest movers. It quickly showed Castrol that stock levels for A-level products needed to be increased to eliminate out of stocks and to minimize the number of emergency production runs made to keep service levels high. Inventory levels for class-B and class-C products, on the other hand, needed to be scaled back. Although specific numbers are not available, reducing inventory was critical to generating a return on investment since even small decreases in unnecessary stock translate into savings when so many products are involved. As implementation progressed, the biggest surprise came from production. DPM resulted in a reduction of raw-materials inventory and urgent production runs to respond to service problems. "Most of the warehouse is filled with components," explains Tenaglia. "In fact, the mix of labels, colors, caps and bottles is quite complex and difficult to simplify because of sales reasons," he adds. Armed with a more accurate replenishment plan, Castrol was able to reduce safety stocks for packaging components. In fact, "most of the reduction in investment capital came from components rather than finished product," Tenaglia says. Since lack of space in the warehouse often is a bottleneck, reducing component inventory also gives the company the opportunity to increase output without additional investment in production equipment. For example, Castrol's plant in Italy boosted annual output from 15,000 tons to 24,000 tons in just a couple of years by freeing up space in the warehouse formerly used to store components. Another unanticipated DPM benefit is the ability to quantify the cost of service. For example, if three warehouses must be built to provide a 99.9% service level, deciding whether the investment is worth the service improvement can be based on numbers rather than gut feel. Expected benefits also were achieved. Castrol reduced finished goods inventory an average of 35% in the first two years. Some countries had more dramatic results, like Bulgaria's 50% reduction. Service levels as defined by line fill rates increased 9%. It should be noted that line fill rates, which involve the number of orders shipped complete with all items, is a more stringent metric than case fill rates, which counts the number of cases shipped against the order. Sales and marketing is concerned about the percentage of "perfect orders," notes Tenaglia, because if orders aren't complete, correct and on time, a second shipment generally must be sent, and costs rise dramatically. Yesterday's Forecast
Improved forecasting has also made it possible to plan promotions better, which typically consist of banded multipacks, or a different size or color of bottle. This special packaging precludes the sale after the promotion ends. Without a careful match of supply with demand for promotions, the company tended to end up with a lot of unsellable product that had to be reworked -- a costly, time-consuming process that involves emptying the containers to recover the product and then repackaging it. Improved forecasting also helps Castrol optimize batch sizes to balance the trade-off between setup and stock holding costs. Castrol now has one company standard for forecasting, service planning and stock optimization. This is a particularly important unifying factor because the company still has 10 different ERP systems and more than 20 warehouse management systems in use in Europe. Now that Castrol forecast analysts use the same system, accuracy has increased dramatically. There's also a single point of responsibility. Although the benefits are significant, implementation was not without challenges. The biggest hurdles were people related. Tenaglia and his team quickly learned that support from managers is crucial to successful implementation, along with identifying forecast analysts who understood the process and could transfer that knowledge to others. Creating the correct interfaces for the systems in each operation also was critical. But, most importantly, the company's culture had to change. Accustomed to focusing on information in their own silos, people rarely looked outside their own department or even talked to each other. It took a while to build the trust necessary to do this. "For the first time, sales and marketing had an understanding of the importance of a correct forecast and were able to see and understand the cost of achieving certain service levels," explains Tenaglia. "Before, everyone was more focused on reacting to emergencies. Service levels didn't have a cost, so people didn't understand the link or trade-off between service level and level of stock." Inadequate attention to continuity resulted in some backtracking after the initial successful deployment. This was exacerbated by the merger with BP (London) in 2000 since managers from the BP side, who were unfamiliar with DPM and its advantages, often were tapped to lead the merged operations. Once merged, management refocused on DPM as the common tool the company would deploy for demand planning and inventory planning in Europe. Another hiccup occurred more recently when management decided to standardize ERP systems on SAP (Walldorf, Germany) and set the DPM redeployment aside temporarily to determine whether the ERP software had the right demand and inventory planning functionality. DPM was judged better suited to Castrol's demand and inventory planning needs. A secondary rollout is underway. The system now spans 29 installations in 25 countries and two continents. As a result of lessons learned in the initial implementation, Castrol continues to refine its processes, define key performance indicators, train personnel and create an academy of in-house experts to provide the knowledge base that will ensure success and continuity for the forecasting and replenishment process.

Top Enterprise Software Planning (ERP) Comparison