Executive Guide to Forecasting in Manufacturing, Part 3: Choice Under Uncertainty

Forecasting is not just about predicting demand, it’s about making decisions when the future is uncertain. Every manufacturer faces this tension daily: how much to produce, how much to stock, and when to commit resources.

In Demand Forecasting for Executives and Professionals by Stephan Kolassa, Bahman Rostami-Tabar, and Enno Siemsen, explore how organizations can navigate uncertainty using probabilistic thinking, and predictive distributions. The goal is not to eliminate uncertainty, but to understand and manage it intelligently.

Key Insights from the Book

The authors’ message is simple: a forecast is not a plan. It’s an input to support decision-making under uncertainty, not a substitute for it. A structured estimate supports choices involving risk and trade-offs. This distinction is critical for manufacturing leaders who must balance efficiency with resilience.

(If you’re interested in the technical and statistical foundations behind these ideas — including how to calculate prediction intervals and service levels — the book’s Chapter 3 provides a clear, detailed walkthrough.)

1. Forecasts support decisions, not replace them. A forecast represents an informed belief about future demand. It guides (but does not dictate) decisions like production levels or inventory targets. Confusing forecasts with targets or budgets leads to distorted incentives and poor planning alignment.
   
   
2. Uncertainty is measurable, not avoidable. Instead of pretending a single number captures the future, leaders should communicate uncertainty explicitly, through prediction intervals or probability distributions. These tools quantify the range of likely outcomes, allowing better preparation for risk.
   
   
3. Judgment and algorithms complement each other. Few organizations rely entirely on models or intuition alone. In practice, effective forecasting blends both: structured statistical methods combined with human insights from sales, operations, and finance. The challenge is determining where human judgment adds value and where it introduces bias.
   
   
4. The real decision variable: service levels. The book’s “bagel baker” example illustrates the essence of operational decision-making under uncertainty. A point forecast (say, 500 bagels) gives a 50% chance of meeting demand — but that’s often unacceptable. Managers must define their target service level, weighing the cost of stockouts (underage) against excess inventory (overage). The optimal service level is where these costs balance.
   
   
5. Forecast quality is empirical. Whether a forecasting method is “good” or “bad” is not a matter of opinion but of empirical comparison. If a method consistently performs worse than alternatives, it’s not working. The same principle applies to management decisions: measure, compare, and adapt.
   
   

Ultimately, the authors encourage practitioners to treat forecasting as a structured conversation about uncertainty, not a contest for the “right number.”

Quantics Perspective

In enterprise manufacturing, uncertainty is the constant companion of every planning decision. From lead-time variability to volatile customer demand, no forecast can be perfect — but organizations can improve how they make decisions in the face of that uncertainty.

In practice, the challenge isn’t a lack of data. It’s how businesses use forecasts within their existing decision structures. Many manufacturing teams still plan using point forecasts, not because they reject uncertainty, but because their processes, systems, and KPIs have been built around single-number expectations. It’s familiar, measurable, and embedded in day-to-day planning.

At Quantics, we recognize this reality. While academic literature often advocates replacing point forecasts with full predictive distributions, the operational world moves differently. What matters most is how organizations interpret and act on uncertainty — not the format alone.

Our role is to help companies bridge these two worlds. Quantics supports both point and probabilistic forecasts, enabling gradual evolution rather than sudden disruption. For some clients, the starting point is to strengthen how they evaluate and interpret forecast accuracy. For others, it’s introducing ranges and probability thresholds into existing planning frameworks.

Change in forecasting culture takes time. For teams used to a single-number forecast, probabilistic thinking can feel abstract at first. The key is to translate ranges into practical decision signals.

For example: A sales representative updates the consensus forecast. If the new number falls outside the expected prediction range, it should raise a red flag for the planning team or forecast owner — prompting a quick review to determine whether the adjustment reflects genuine new information or simply noise. In this way, ranges become operational tools. They provide context for judgment and help teams focus on meaningful exceptions rather than every deviation.

Quantics’ platform enables this evolution by embedding uncertainty into every stage of forecasting and planning:

• ‍Point and probabilistic forecasts together.
  Probabilistic forecasts extend traditional point forecasts rather than replacing them. Manufacturers can use both approaches, depending on the maturity of their planning processes and how they incorporate uncertainty into decision-making. This flexibility allows teams to continue operating with familiar planning frameworks while gradually introducing richer information about demand variability.

• Richer context for demand signals.
  Effective forecasting requires integrating multiple sources of information. These may include customer forecasts, framework agreements, or quantity discount ranges, as well as internal operational constraints such as production capacity. External signals also matter — from public holidays and macroeconomic indicators to market growth, energy and raw material prices, and even weather patterns. Bringing these signals together provides more context for planning and inherent uncertainty.

• Scenario-based planning.
  Beyond generating multiple scenarios, planners can explicitly characterize the risk profile of forecasted demand — for example by marking situations as low, medium, or high risk. This helps decision-makers interpret forecast ranges in operational terms and prioritize where attention and mitigation actions are needed.‍

• Transparent overrides and traceability.
  When planners adjust forecasts, those overrides are logged together with their rationale. Over time, this creates a valuable feedback loop that reveals which human interventions improve forecast outcomes and which introduce noise.

• Cross-functional alignment.
  Finance, sales, and supply chain teams work from the same consistent view of demand. This shared perspective helps transform forecasting from a siloed exercise into a coordinated decision framework across the organization.

• Result tracking and empirical learning.
  Forecast value added can be tracked at each stage of the forecasting process. This allows organizations to empirically evaluate which models, adjustments, or inputs improve decision quality — and which steps add complexity without improving outcomes.

For manufacturing leaders, the real advantage lies not in eliminating uncertainty but in operationalizing it — turning statistical insight into disciplined, evidence-based business decisions.

Practical Takeaways

Executives can strengthen decision-making under uncertainty by:

• Separating forecasts from decisions. Treat the forecast as an input, not the final word.
  
• Communicating uncertainty. Use scenarios and ranges if possible to add context to point forecasts, helping teams interpret risk and make better-informed decisions.
  
• Balancing cost and service. Define service levels explicitly using overage and underage cost logic.
  
• Integrating judgment responsibly. Encourage expert overrides but document and analyze them.
  
• Evaluating forecasting methods empirically. Continuously compare model performance and adjust based on evidence.
  
  

Designing workflows around decisions. Ensure forecasts are delivered in a form that supports planning and execution, not just reporting.

In Conclusion

Every manufacturing decision involves uncertainty, from production to inventory to customer service. The goal of forecasting is not to make uncertainty disappear, but to make it visible and manageable.

While the book advocates moving beyond single-number forecasts, in practice, most manufacturers still plan around point forecasts — and that’s okay, as long as the uncertainty around them is understood.

At Quantics, we believe forecasting maturity is not measured by algorithmic complexity, but by how well organizations turn uncertainty into confident, coordinated action across the supply chain.

Disclaimer

In this post, we share highlights from the book “Demand Forecasting for Executives and Professionals” by Stephan Kolassa, Bahman Rostami-Tabar, and Enno Siemsen, together with our own reflections on how these ideas apply to today’s manufacturing supply chains. This is not a replacement for the book, but rather a guide to spark thought and discussion. For a deeper dive, we encourage you to explore the book itself - available here.