Success Story: Demand Forecasting with AI in the Corporate Group

Background: The high number of materials and significant market fluctuations made program planning challenging

Our customer, an international leader in drive technology, operates in a market environment characterized by increasing dynamism and substantial fluctuations in customer order behavior. This volatility, combined with a steadily growing number of planning objects – regularly tens of thousands of items – presents planners with increasingly complex daily challenges.

Traditional planning methods and even the existing automated SAP forecasting solution reached their limits due to these highly fluctuating and extensive demands. Despite the planners’ dedicated efforts, this resulted in significant time expenditure and often suboptimal inventory situations – from overloaded warehouse capacities to supply bottlenecks. These demanding conditions highlighted the necessity to further optimize inventories and effectively address the well-known bullwhip effect along the supply chain – between customers, regional subsidiaries, and manufacturing plants. Additionally, the demand for advanced tools for precise measurement and continuous improvement of planning quality grew.

To meet these increased requirements and to utilize planners’ valuable resources more efficiently, our customer sought an intelligent solution. The aim was to significantly enhance planning quality and sustainably reduce the workload of employees through an AI-based demand forecasting solution, enabling them to focus on more strategic tasks.

“Inconsistent ordering behavior between customers and plants generates an enormous effort in manual planning for us.”

Solution: an AI planning assistant to simplify planners’ work

Approach

To flexibly respond to changing requirements and needs, we collectively opted for a step-by-step approach. The joint data science journey proceeded as follows:

1. Initial joint discussions and workshops: The idea of the final product was discussed and outlined. We also produced initial results to provide our customer with an impression of expected outcomes. This enabled our customer to internally set all necessary groundwork.
2. Prototype: We began with a selected and representative dataset, gradually expanded it, and delivered an initial version of an interactive dashboard. During this phase, we regularly provided forecasts to our customer, including evaluated influencing factors.
3. Operationalization: Interfaces were created to allow forecasts to be automated and regularly computed. Together, we selected an ONfuture solution, fundamentally leveraging the functionalities of the software future. Our data scientists tailored and supplemented this specifically for the customer. Additionally, performance tracking was implemented.
4. Continuous improvement & roll-out: The custom AI-based demand forecasting solution was expanded to other areas within the corporate group, incorporating and implementing new business areas’ requirements.

“The comparison between AI forecasts and our manual planning shows a substantial improvement in our planning.”

Result: 20% forecast accuracy improvement – and reduced planner workload

The AI-based automated and customized forecasting system achieved a 20% improvement in forecast accuracy for our customer. Regular, reliable forecasting of several thousand materials means a significant workload reduction for planners: Approximately 40% of the manual effort could be saved. They can conveniently view results in an interactive dashboard and adjust figures if necessary. The traffic-light system ensures they aren’t overwhelmed by the large number of materials but instead can specifically examine those planning objects flagged by the system.

“We will gradually connect additional business areas to the AI-based demand forecasting solution, as they too should benefit from these advantages.”

Conclusion: a customized demand forecasting solution perfectly matches our customer’s requirements

Thanks to an agile approach, changing requirements could be smoothly integrated during the project, successfully allowing it to grow.

Success factors

An AI solution is only effective if the user recognizes its value and can interact perfectly with it. This was also one of the crucial success factors in this project: planners significantly contributed to shaping the final solution and specifying requirements—as they know their daily tasks best. In close cooperation with our data scientists, the ideal solution for our customer emerged. Precisely this was our goal!

The project team

Dr. Thomas Christ

Chief Data Scientist & Lead of Consulting

André

Data Scientist

Gregor

Principal Data Scientist

Katharina

Engineering Associate

Would you like to improve your planning figures and simultaneously reduce manual workload? Then