Last updated: Forecasting and replenishment: How hypermarket Kaufland nailed it

Forecasting and replenishment: How hypermarket Kaufland nailed it


Listen to article

Download audio as MP3

Ever wondered how your favorite cereal is always available when you need it, or how the store seems to never run out of fresh produce? Behind the scenes, demand forecasting and replenishment technology is working to keep shelves stocked.

Advances in this core technology are making it possible for retailers like Kaufland, a German hypermarket chain, to optimize their operations for increased efficiency, accuracy and insight into shopper behavior.

Banner illustration of a woman wearing sunglasses that says: JOIN THE RETAIL REVOLUTION: Predictive technology and AI are yielding REAL-LIFE retail outcomes to seamlessly engage customers online, in-store, and virtually. Get all the details in our powerhouse session! Register now.

A balancing act: Forecasting & replenishment

Demand forecasting is the process of analyzing historical sales data, along with a multitude of factors such as seasonality and market trends. Replenishment is about making sure the right amount of product is ordered and delivered to the store at the right time to restock the shelves as products are sold.

Together, demand forecasting and replenishment create a cycle that keeps a store running smoothly. Accurate forecasting helps order the right amount of stock, reducing the chances of overstocking, which can lead to waste, understocking, and unhappy customers.

It’s a delicate balance, but when done right, it means you can get your weekly grocery supplies without a hitch.

For departments like fresh produce, meat, and baked goods that have significant fluctuations in sales each day, accurate prediction down to the individual SKUs are vital. Kaufland relied on software from an SAP rival to calculate forecasts for fresh products, but the algorithms weren’t providing sufficient granularity for all fresh products.

Unified Demand Forecasting

For Kaufland, SAP’s Unified Demand Forecasting (UDF) has proven to be a game-changer. UDF leverages powerful algorithms and machine learning to predict future sales – i.e., how much of each product will sell in a given timeframe.

SAP’s UDF goes beyond historical sales data to produce predictions. It takes into account a wide range of influencing factors, like weekdays, price effects, trends, seasons, payday effects, and weather. It can even make educated guesses about the sales of new products based on similar items.

All this means that UDF can produce incredibly accurate + granular forecasts, which are then used for everything from stock replenishment to ad planning. The solution can run what-if scenarios, testing how different factors might impact sales.

As a unified solution, these forecasts can be used across multiple use cases, from auto-replenishment to promotion quantity planning, assortment planning, price calculation, and warehouse requirements.

Better algorithms = better forecasts 

Kaufland was the first retailer to use SAP’s new UDF forecasting system, which replaces forecasts for all assortments from the old SAP F&R system and also the rival’s fresh products forecasts. The system forecasts the expected sales quantities for all assortments at all 1,450 Kaufland stores in Germany and Eastern Europe.

The improved algorithms provide “better forecast quality,” says Michael Hahn from Schwarz Group. Hahn is responsible for supply chain management systems within Schwarz IT, operating both the Lidl and Kaufland brands.

Based on SAP HANA database technology, the UDF system calculates up to 35 million store-product combinations a day. In the first step in this process — modeling –UDF evaluates the sales history from the past 800 days.

In the second step — forecasting — the system calculates expected sales for the next 101 days at a rolling daily level. The long forecast horizon is intended to support advertising planning. Purchase orders to suppliers and supply to stores are then controlled by forecasts for the near future or even the next day.

According to Hahn, the results were so good that they became the basis for production control in Kaufland-owned meat plants and for the scheduling of baking machines in stores.

Kaufland’s replenishment software, which provides optimized order quantities based on factors like rebates and truck utilization as an auto replenishment, now also leverages forecasts from the UDF system.

Optimizing forecasting and replenishment

For Kaufland, the huge number of influencing factors that SAP’s UDF system can include in its calculations are a huge plus. But to streamline the initial deployment, Kaufland limited the factors to those that are relevant for the given assortment.

Also, Kaufland aims to use intelligent modelling for products recalculated daily in order to reduce the current computing time of eight to nine hours by focusing only on products with significant changes.

Hahn says that in the case of promotional products, Kaufland managed to push the error rate to less than 38%. Here, the error rate refers to the WMAPE (weighted mean absolute percentage error). The most important KPIs for Kaufland in assessing the forecast software are:

  1. Out-of-stocks
  2. Inventories
  3. Inventory costs
  4. Revenue

Kaufland uses SAP’s Unified Demand Forecast solution for promotion quantity planning, assortment planning, price calculation, and warehouse requirements.

The retailer also uses the software to run what-if scenarios for a number of use cases, including advertising planning and pricing.

Happy customers.
Connected experiences.
See real-life retail outcomes HERE.

Frequently asked questions (FAQs):

Demand forecasting is defined as a strategic process that leverages predictive analysis techniques to predict future demand for products and services. The process involves analyzing historical sales data and a multitude of other relevant factors that help businesses project the expected sales and revenue for a specified future period. This process not only aims to predict the quantitative demand but also to understand customer preferences, needs, and purchasing intent. The essence of demand forecasting lies in its ability to guide businesses in making informed decisions across various business processes, from inventory management and warehousing requirements to product direction, pricing strategies, and potential market expansions.

The importance of Demand forecasting cannot be overstated, it is an essential business process that helps businesses anticipate customer needs by predicting customer interest in products or services. It serves as a compass for businesses to navigate the complex landscape of customer needs, market dynamics, and operational challenges. This foresight not only helps in predicting the numbers but also helps in making informed, strategic decisions that drive growth and customer satisfaction.

Demand forecasting at its core provides businesses with the foresight to make informed decisions on several fronts:

  • Inventory Management: efficient demand forecasting ensures that businesses strike the right balance between having enough stock to meet customer demands and avoiding overstocking, which can lead to increased storage costs and potential wastage.
  • Financial Planning: by predicting demand, businesses can make sound budgetary decisions, optimizing cash flow and resource allocation. This is especially crucial for growing businesses where inaccurate forecasts can strain financial resources.
  • Pricing and Marketing: understanding demand patterns allows businesses to adjust pricing strategies based on competition, seasonality and a host of several factors thereby maximizing profit during high demand periods and promoting sales during anticipated lulls. It also guides marketing efforts, ensuring they align with product availability.
  • Operational Efficiency: in the prevalent dynamic and fast-paced business environment, demand forecasting aids in streamlining operations, from production schedules to staffing needs. It’s a tool that helps businesses stay agile, competitive, and responsive to evolving customer behaviors and market trends.

Demand forecasting is an essential tool that helps businesses anticipate future demand for their products and services. Accuracy in predicting demand can help businesses make informed decisions about inventory levels, product pricing, marketing strategies, or even potential market expansion. Following are simplified steps about how to forecast demand:

  • Set clear objectives: define what you’re forecasting (specific product, service, or category), the time frame, and the target audience; know your end goal.
  • Collect data: gather both quantitative and qualitative data to inform the model. Qualitative data can range from market trends, competitor research, and customer feedback, whereas for quantitative data use in-house data like sales figures, peak shopping periods, and web analytics. Analyze the data to identify patterns; leverage tools such as demand forecasting software, to help automate and refine this process;
  • Make adjustments: use your findings to make informed decisions; for instance, increase inventory for predicted demand surges or adjust marketing strategies for anticipated declines.
  • Utilize a modern intelligent technology stack: implement advanced analytics, AI, and machine learning to enhance accuracy and process complex and massive data sets.
  • Budget and plan: align your business’ resources with your forecast; this ensures you’re prepared for predicted demand fluctuations.
  • Review and iterate: regularly compare forecasts against actual sales and refine your methods based on variances.

It is important to note that while demand forecasting isn’t an exact science, consistent effort, trial and error, along with the right tools can improve its accuracy over time.

Unified demand forecasting integrates various data sources and forecasting techniques to holistically predict demand for a wide range of products and services, such as those offered by hypermarket retailers like Kaufland. This approach emphasizes data consolidation, method standardization, and cross-departmental collaboration to enhance prediction accuracy. Particularly, the SAP Customer Activity Repository solution includes a specialized module called the Unified Demand Forecast (UDF). Powered by SAP HANA, UDF provides advanced demand modeling and forecasting capabilities, giving retailers precise insights into shopper behavior. These insights are crucial for tasks ranging from stock replenishment to advertising planning.

Forecasting and replenishment is an integrated business process for predicting demand and ensuring adequate product availability in retail stores and distribution centers; it is a critical process in supply chain management (SCM). Forecasting predicts future product or service demand, guiding decisions about inventory quantities and timings. Replenishment ensures that inventory is restocked after sales, maintaining sufficient levels to meet customer demand. These processes are highly interconnected; accurate forecasting leads to effective replenishment, preventing stock-outs or excess inventory. From analyzing customer behavior, historical sales data and a multitude of other influencing factors, and by leveraging powerful algorithms and machine learning techniques, we have tools to predict demand accurately in combination with replenishment strategies like just-in-time help optimize inventory. SAP’s Unified Demand Forecast (UDF) module, for instance, offers advanced demand modeling and forecasting, enhancing these processes.

Stock replenishment is the systematic process of ordering and restocking inventory to maintain optimal stock levels. An efficient stock replenishment process seeks to prevent shortages, ensure products are available for customers, and optimize inventory-related costs. This involves forecasting future product demand, maintaining desired inventory levels, and using technology to automate and refine the replenishment process. An effective proper stock replenishment is crucial for efficient supply chain operations and customer satisfaction.

Share this article


Search by Topic beginning with