Developing Demand Planning Capabilities with AI

More than ever, businesses face the big challenge of having to increase revenue while keeping costs under control. In the current economic situation, cost efficiency is as crucial as revenue growth.

Energy costs, labor-related costs, bureaucratic hurdles, and global competition are putting significant pressure on European industries, which have failed to adapt organizationally and process-wise in recent years. While the USA experiences rapid productivity growth through AI, advanced sensors, and bioengineering, Europe struggles with minimal productivity gains due to lack of vision and innovation (more on LinkedIn).

Like a weather forecast, slight inaccuracies in demand predictions can have major consequences. By refining forecasts, just as meteorologists adjust their models, organizations improve their responsiveness, avoid costly mistakes, and maintain better cost control. This higher forecasting accuracy (FCA) paves the way for greater efficiency overall. However, traditional approaches to improving FCA have proven costly, time-consuming, and often ineffective. To address these limitations, the integration of artificial intelligence (AI) into supply chain (SC) planning emerges as a transformative solution.

In the following chapters, we want to focus on FCA optimization through AI-assisted planning. We will show how companies can significantly improve their planning processes and minimize costs by leveraging AI, while maintaining high service levels.

Improving FCA has historically been approached through manual efforts and traditional planning methods. While these efforts may yield marginal improvements, they often involve significant additional workload for sales teams and planning analysts.

Investing more time and/or resources into manual forecasting
This conventional approach relies heavily on manual forecast creation and adjustments. Big part of commercial organization (sales managers, commercial directors, commercial heads, …) is participating in the forecasting process, where everyone provides input, but these efforts result in complex process and often fail to add value. As seen in figure 1, with the assistance of sales managers a baseline FC is created with 70% FCA, then with the inputs of the Sales Directors FCA increases by 2%, but when the Commercial Head adjusts the forecast to align with the budget, this results in a negative forecast value add (FVA) of 7%, because the forecast is influenced by bias.

These adaptations come at the cost of extensive manual planning, adding unnecessary complexity and resource consumption and often result in negative forecast value add.

Using statistical forecasting as a baseline
Statistical forecasting methods provide a structured foundation for sales predictions, reducing the impact and the need for subjective inputs. However, these approaches remain costly and share the drawbacks of manual planning: lengthy processes and significant efforts required from sales teams.

These adaptations come at the cost of extensive manual planning, adding unnecessary complexity and resource consumption and often result in negative forecast value add.

Using statistical forecasting as a baseline
Statistical forecasting methods provide a structured foundation for sales predictions, reducing the impact and the need for subjective inputs. However, these approaches remain costly and share the drawbacks of manual planning: lengthy processes and significant efforts required from sales teams

To overcome the challenges of high complexity and manual effort, companies must focus on two key elements:

First, a differentiated approach to planning based on specific product and customer characteristics. The key principle lies in situational optimization: tailoring forecasting and planning strategies according to the nature of the products, demand volatility, and business criticality. Determining the right forecasting approach is crucial and can be easily done with Supply Chain Segmentation.  Often, unbiased algorithms perform better than the planner. When considering whether a specific segment should be planned by a machine or a human or a combination of both, we recommend that you follow the simple principle:

Choose your battles wisely: where the machine performs better than the planner, let the machine do the job!

The second element is using AI as an advanced algorithm. It handles independently numerous decisions (data cleansing, the selection of suitable forecasting methods, the identification of both internal and external indicators to improve FCA on the lowest planning level, etc.) and automates the planning process.

Below is a simplified example of product groups with distinct characteristics and suitable planning approaches.

Make-to-Stock (MTS) Products
MTS products are stable and predictable and do not require detailed manual forecasts, therefore when assisted by AI-enabled automated planning, reliable statistical forecasts can be generated, ensuring efficient resource allocation without any human intervention.

Products with High Strategic Importance
Products with high volatility and strategic significance require advanced AI-supported algorithms with human oversight. AI systems incorporate external data sources, market trends, and other influencing factors to improve accuracy. Human expertise remains critical for final review and control. Sales managers and/or demand planning analysts play a specialized role, focusing on validating financial figures and providing the final sign-off.

Products with Low Strategic Importance
For products that are customized but lack strategic relevance, businesses should avoid excessive planning efforts. This category includes mostly products with low and volatile demand which are very difficult to plan, whether with AI or manually. In such cases, the cost of improving FCA often exceeds the financial benefits. The principle here is to optimize resources by minimizing unnecessary forecasting efforts.

Achieving this level of optimization requires a fundamental shift in the way businesses approach planning and forecasting. Three key enablers must be addressed:

People
Specialized roles are essential for implementing AI-supported planning. Demand planning analysts and sales experts must develop expertise in managing advanced forecasting tools and interpreting AI-driven insights.

Processes
Differentiated planning processes must be implemented, categorizing products and customers based on their characteristics. Supply chain Segmentation is crucial to ensuring that planning strategies are appropriately tailored.

Tools
form the backbone of this approach, but at the same time it is crucial to find the best-fit tool in the emerging market of software providers. Tailored AI algorithms can analyze large datasets, identify patterns, and provide actionable insights to enhance FCA while reducing manual effort.

AI-supported optimization represents a game-changing solution for improving forecast accuracy and optimizing customer deliveries. By adopting a differentiated approach to planning — tailored to product characteristics, demand volatility, and strategic importance — businesses can achieve situational optimization that reduces costs and drives revenue growth.

The successful implementation of this strategy requires a combination of specialized expertise, process differentiation, and advanced AI tools. Studies show that almost 60% of buyers regret at least one software purchase made in the past one and a half years, while one out of two of those are facing increased costs as a result of their decision.[1]

We at TenglerConsulting help you to define the relevant requirements for the software and guide you through the selection process to successful implementation. Companies that embrace this approach will gain a significant competitive edge, ensuring efficient and cost-effective supply chain operations in an increasingly challenging market environment. We are here to guide you through this new environment from idea to implementation while making sure that your strategy aligns with this transformation.

 

 

Source:
[1] Gartner, 2025: “2025 Software Buying Trends Report”  [Accessed: 07 January 2025].