About Kraft Heinz
As a globally trusted producer of delicious foods, The Kraft Heinz Company provides quality, great taste, and nutrition for all eating occasions, whether at home, in restaurants, or on the go. Based in Amsterdam, the global supply chain team drives the company’s end-to-end value chain, putting consumers at the center of everything it does – from the quality of its world-class, iconic brands to its commitment to the environment and its people.
The challenge
In the highly competitive and dynamic FMCG industry, accurate demand forecasting is critical for aligning supply chains, optimizing service levels, inventory, and managing costs. Kraft Heinz faced forecasting challenges that were heavily influenced by seasonal demand, promotional pressures, and evolving consumer market preferences.
To address these challenges, Kraft Heinz wanted to implement new APS tooling that would allow them to incorporate machine learning (ML) into their demand planning process to improve overall forecast performance. Recognizing that they lacked in-house data science expertise, Kraft Heinz turned to EyeOn for expert assistance. As they prepared to transition to the new supply chain planning system, they realized that this capability was key to unlocking the full potential of the upcoming tool.
The project
EyeOn was asked to partner with Kraft Heinz in building in-house data science capabilities in forecasting, aiming for improvements in forecasting across all global sales organizations.
- Phase 1: Data discovery and validation
The project began with a data discovery process in which EyeOn worked with Kraft Heinz to identify the key demand drivers impacting their forecasts. We validated the quality of their data and looked for correlations between the drivers and forecast performance. This first step laid the foundation for understanding how seasonality and external factors, such as promotions, influenced demand shifts.
- Phase 2: Iterative machine learning modeling
Next, we used an iterative machine learning approach to assess the impact of various demand drivers on forecast accuracy and bias. During this phase, EyeOn optimized model parameters and evaluated the performance of different models, comparing the results to Kraft Heinz’s existing forecasting methods. This phase also included routine testing of new models to ensure continuous improvement.
- Phase 3: Modular forecast engine and knowledge transfer
As the transformation progressed, an internal need for a modular, scalable and maintainable machine learning forecasting engine was identified. EyeOn designed this modular forecast engine, allowing Kraft Heinz to maintain and scale the model as their business evolves. By incorporating software engineering best practices, we ensured that the engine was robust, maintainable, and adaptable to future business needs by integrating both traditional statistical and machine learning models into a unified platform. Once the model was fully developed and validated, EyeOn transferred the knowledge and code to Kraft Heinz’s internal data science team, ensuring a seamless transition and continuity.
In addition to developing the forecasting engine, EyeOn worked closely with the Kraft Heinz team to co-create several analytics reports. These reports included forecast accuracy, forecast value add, stability, and trend analysis, providing the team with valuable insights to monitor and refine their forecasting process. This focus on user adoption was critical to ensuring that Kraft Heinz could effectively use the new forecasting system in their day-to-day operations.
Results
The project delivered several key results for Kraft Heinz, including:
- Driver insights: New insights on different granularities in different markets.
- Modular forecast engine: The modular design allows Kraft Heinz to integrate traditional statistical models and machine learning models into one unified interface, increasing flexibility and scalability.
- Knowledge transfer: EyeOn successfully transferred the forecasting engine and provided training to Kraft Heinz’s internal teams, ensuring long-term sustainability and in-house capability.
- Enhanced forecast stability: By incorporating key demand drivers and machine learning techniques, Kraft Heinz saw improved forecast stability, with reduced bias and increased accuracy.
- Extendibility: The new forecast engine can easily adapt to future business needs and evolving data sources.
This project was an important step in Kraft Heinz’s journey to transform its supply chain forecasting processes. By partnering with EyeOn, they were able to leverage the power of machine learning to increase forecasting accuracy, improve decision-making, and valorize the transition to their new supply chain planning tool.
Ready to drive sustainable value from your digital planning transformation?
Our new guide provides you with an end-to-end APS digital transformation framework based on proven strategies used by industry leaders. Download your copy here or get in touch with our end-to-end transformation specialists right away.