To compete more effectively, Rotkäppchen-Mumm wanted to boost working capital and cut warehouse costs. How could it shrink stock levels without risking stockouts or impacting customer service levels?
Rotkäppchen-Mumm enhanced its sales forecasting with predictive analytics from IBM—enabling it to produce the optimal product quantities based on accurate insights into demand and future trends.
22%increase in accuracy of selected sales forecasts
Reducesstock levels without increasing the risk of stockouts
Boostsworking capital by releasing funds previously tied up in overstock
Business challenge story
Boosting efficiency to protect marginsFor businesses operating in saturated markets, cost-efficiency is a powerful tool to combat pressure on margins. Running lean is an important objective at Rotkäppchen-Mumm Sektkellereien GmbH, a leading producer of spirits and sparkling wines. The company wanted to maximize the profitability of its 47 percent share of the crowded German marketplace—but inefficiencies and poor accuracy in its sales forecasting process presented a challenge. In particular, a largely manual approach to running rolling forecasts was not scaling well to the challenges of a significantly expanded product catalog and business serving new markets.
André Birrenbach, Chief Information Officer at Rotkäppchen-Mumm, explains: “In the wine industry, lead times tend to be long. For example, our production cycle for sparkling wine typically runs for six months, and industry regulations require us to keep certain types of wine in our manufacturing plant for up to 12 months before we sell it.
“Like many leading enterprises in the fast-moving consumer goods (FMCG) space, we depend on robust supply chain management capabilities to ensure that every stage of the production lifecycle runs smoothly—from sourcing the wines that will be used to produce our products to bottling and shipping the finished stock keeping units [SKUs] to our warehouses ready for distribution to our retail clients.
“Every month, our key account managers and supply chain managers would use their experience to predict the volumes that our clients would purchase within a three-month period. However, because we relied on manual sales forecasting there was always a margin of error in our results. To minimize the business risk of stocking out on key product lines, we produced to a best-case plan value, which meant that we were tying up valuable working capital in excess stock and driving up our operational costs with warehouse storage fees.
“To solve the challenge, we looked for a way to augment our people’s expertise with predictive analytics. The aim was to enable them to build more accurate, granular rolling forecasts, and use those insights to produce the optimal quantity of each SKU.”
Adopting a hybrid approach to analytics
After considering potential solutions from a number of vendors, Rotkäppchen-Mumm decided to automate and enhance its forecasting processes with a new solution based on IBM® Planning Analytics (formerly known as IBM Cognos® TM1®) and IBM SPSS® Modeler software.
“We have been using IBM Analytics solutions for many years, which gave us confidence both in the technical capabilities of the solution and the quality of IBM’s product support and development,” recalls André Birrenbach. “We first performed a thorough proof-of-concept [POC] exercise, in which we extracted data from our ERP and CRM systems into Planning Analytics cubes for 46 products and four large customers as the scope of the POC. Using SPSS, we then constructed time-series models to forecast sales over a three-month horizon. To ensure a sufficient level of accuracy, we required a minimum of 12 data points for each product to be analyzed.”
He continues: “During the POC process, we assessed the significance of a number of potential sales drivers, including seasonal factors such as public holidays, special events like the FIFA World Cup and trade promotions such as advertising and discounts. Crucially, we drilled down to the level of individual SKUs, enabling us to assess the accuracy of our models at a highly granular level.
“Our initial work enabled us to refine our models by determining which factors were significant. In particular, we recognized that the variability of price promotions over the course of the year was distorting our forecasts. In an ideal world, you would be able to eliminate these data points from your forecasts, but in practical terms that’s not possible. Rather than accepting the distortion, we decided to treat the data on the price promotions as an external factor in the forecasts.
“One of our most important discoveries was that our manual approach was more accurate than our statistical model in certain scenarios. We realized that adopting a hybrid approach to forecasting that combined statistical models with our existing manual methods would deliver the most accurate results for our decision-makers – in effect combining both art and science into the process.”
Based on the positive results of the POC exercise, Rotkäppchen-Mumm partnered with an expert team from IBM and IBM Premier Business Partner Axians IT Solutions to move its new sales planning and performance management solution into production.
“Working with IBM and our partner Axians IT Solutions enabled us to start obtaining value from our new forecasting solution extremely rapidly,” comments André Birrenbach. “Today, we have a fully integrated hybrid approach to forecasting—delivering a single solution that serves both our sales and production teams with a three-month and a 24-month view of future sales. What’s more, the IBM Cognos Analytics presentation layer of the solution has a similar look and feel to our previous IBM Analytics solution, helping us to reduce our training requirements and get people up and running quickly.”
Cutting costs, boosting capitalWith its IBM Analytics solutions driving its sales forecasting and planning processes, Rotkäppchen-Mumm is already reaping the benefits of more accurate insights—enabling it to reduce warehouse stock levels without increasing the risk of stocking out on key SKUs.
“In the past, we based our decision-making processes purely on the knowledge of the Key Account Management team,” explains André Birrenbach. “Today, by offering our experts access to proven statistical models, we empower them to back up their assumptions with hard data science. And because we are adopting a hybrid approach, we combine automated forecasting based on our IBM Analytics solution with the knowledge and experience of the Key Account Management team. This approach delivers more accurate results for each customer and product.”
He continues: “During our POC, we recorded a mean absolute percentage error [MAPE] of 18 percent for the manual forecasting process for one of our key products. Using our automated forecast, we slashed the MAPE down to 13 percent.
“Better still, when we drilled down to SKU level we achieved a MAPE of 22 percent—vastly more accurate than our manual forecasts for the product. As this product alone accounts for six percent of all pallets in our warehouse, we can see that optimizing the quantity of those SKUs alone will cut our costs sufficiently to deliver a full return on the investment in the IBM solution.”
Following the early success of its new approach to sales forecasting, Rotkäppchen-Mumm is now planning ways to further enhance the accuracy of the platform as the SPSS models learn over time, with more data further increasing the accuracy levels.
“Our journey with Planning Analytics and SPSS is just beginning, but we are very excited by what the future holds,” says André Birrenbach. “We are continually reviewing the models that drive our forecasting process in Planning Analytics, which will ultimately enable us to cut down the margin for error in our reports. This will help us further optimize our stock levels and increase the return on working capital available to the business. Looking to the future, we plan to offer the IBM Analytics solution to our spirits division, which will enable us to improve operational efficiency across the enterprise.”
He concludes: “Business key performance indicators are notoriously difficult to predict. Thanks to our IBM Analytics solutions, we can combine human expertise with accurate statistical models—delivering timely, dependable information to support our decision-making processes.”
Founded in in Freyburg, Germany, Rotkäppchen-Mumm Sektkellereien GmbH produces sparkling wines, spirits and a variety of other drinks under brands such as Geldermann, Rotkäppchen, Mumm, Jules Mumm, MM Extra, Kloss and Foerster, Chantré, Mariacron, Echter Nordhäuser and Eckes Edelkirsch. Employing more than 600 people, Rotkäppchen-Mumm is market leader in sparkling wine in Germany, holding almost a 55-percent share of the market.
Take the next step
Founded in 1987 and employing more than 1,000 people, Axians IT Solutions is a leading system and consultancy firm headquartered in Ulm, Germany. An IBM Premier Business Partner, Axians IT Solutions has helped more than 3,000 public and private organizations boost the efficiency of their technology, organization and administration. To learn more about products, services and solutions from Axians IT Solutions, please visit: https://www.axians.com/en/.
IBM Analytics offers one of the world's deepest and broadest analytics platform, domain and industry solutions that deliver new value to businesses, governments and individuals. For more information about how IBM Analytics helps to transform industries and professions with data, visit ibm.com/analytics. Follow us on Twitter at @IBMAnalytics, on our blog at ibmbigdatahub.com and join the conversation #IBMAnalytics.
View more client stories or learn more about IBM Analytics