2374. Optimizing Forecasting Efforts: Focusing on What Really Pays Off
Invited abstract in session TB-38: Forecasting, prediction and optimization 2, stream Data Science meets Optimization.
Tuesday, 10:30-12:00Room: Michael Sadler LG19
Authors (first author is the speaker)
| 1. | Johann ROBETTE
|
| VEKIA |
Abstract
Forecasts are central to business decision-making, and improving their accuracy is often seen as a direct path to better outcomes. Many companies pursue ever-higher accuracy, believing that "the sky is the limit." However, research shows that the link between accuracy and business impact is far from obvious. Due to decision-making mechanisms, many forecast improvements generate no tangible value — despite the significant investment they require.
This raises a key question: When is it worth improving a forecast, and when is “good enough” sufficient? Our empirical studies (Foresight Issue 74, JBF Spring 2024) and past conference presentations (ISF 2023, MOFC 2023) demonstrated that in supply chain replenishment, only a small fraction of decisions truly benefited from better forecasts.
Yet, such ex-post analysis is not directly actionable. What if we could predict, in advance, where improving forecasts matters? This would allow companies to focus their efforts, reducing costs and complexity while maximizing business impact.
Existing segmentation methods like ABC/XYZ classifications attempt this, but are they truly effective? Can we do better? And if so, is the juice worth the squeeze?
In this talk, we present findings from our 2024–2025 studies at Vekia and explore how businesses can refine their forecasting strategies and make sure to focus on what really pays off.
Keywords
- Forecasting
- Decision Analysis
- Supply Chain Management
Status: accepted
Back to the list of papers