Dan Seville, CFP, from the Institute of Business Forecasting & Planning, in the article “Demand Planning 101: The Basics” (May 2025), provides a useful clarification on the difference between demand forecasting and demand planning. They are closely related but serve different purposes:

• Demand forecasting is the analytical process of using data, statistical models, and judgment to predict future demand. It is a probability-based estimate of what might happen and forms the foundation for planning decisions. It is essentially a predictive exercise.
• Demand planning goes beyond forecasting by integrating it into the company’s strategy and operating rules, and by aligning stakeholders — sales, production, procurement, distribution, marketing, and finance — around a shared set of expectations, while determining the necessary actions to meet that demand. It is essentially a prescriptive exercise.

Thus, they serve different objectives, even though they are closely linked. Demand forecasting feeds into demand planning, but it is not the only piece of information considered.

This text expands on the article “Choosing the Right AI Model” to think on the types of AI models that can support demand forecasting and demand planning. But mainly to highlight the differences between both set of model.

Demand Forecasting

👉 Objective: Predict future demand as accurately as possible.

👉 AI models to consider: primarily quantitative and predictive models with strong statistical, probabilistic, and time-series analysis capabilities. This comes back to the need for high-quality data in sufficient quantity. With AI tools based on these models, one can “arrive at a number” for the demand forecast.

💡 Example: predicting that the demand for road bicycles will be 10,000 units in September, with a confidence interval.

Demand Planning

👉 Objective: Translate the forecast into concrete actions within the company. 👉 AI models to consider: prescriptive, optimization, and simulation models. These must take into account the company’s operating rules, production capacities, current inventories (finished goods and raw materials), and possibly upcoming marketing initiatives. We’ll choose among the Mathematical optimization models: linear/mixed-integer programming (MILP), Multi-agent simulations to test scenarios for aligning production, inventory, and logistics, Reinforcement Learning to adjust inventory and distribution policies and finally LLMs if one wants to incorporate company policies and business rules.

Here, a mixed human-AI approach is often required to orchestrate the inputs from sales, marketing, and production.

💡 Example: deciding how much to produce, where to store it, and how to allocate resources, based on the forecast of 10,000 units — while meeting delivery commitments and minimizing transport costs.

Summary

As highlighted in “Choosing the Right AI Model”, the selection of AI tools cannot be improvised. The example of demand forecasting versus demand planning clearly illustrates this:

• Forecasting = predictive models (what is likely to happen).
• Planning = prescriptive and optimization models (how the company should respond).

This brief reflection does not intend to prescribe choices of models in any way. It merely illustrates the care that must be taken when selecting a model.

Write to you soon.

Ciao