Journal of Business Forecasting Articles Issues

Despite AI being able to synthesize large datasets, it falls short when it comes to effectively forecasting new products where little relevant information is available. AI’s inability to manage demand uncertainty and adapt to changing scenarios is what inspired my Hybrid Human-AI Forecasting Framework. This approach combines Demand Planners’ real-world expertise and strategic insight with AI’s gathering and analysis of emerging responses to the product to better manage uncertainty, compressed product life cycles, and limited historical data.

Artificial Intelligence is rapidly becoming the centerpiece of demand planning discussions, yet measurable business results have struggled to keep pace with enthusiasm. Are we witnessing the next evolution of planning, or are we inflating expectations faster than organizations can create value? Drawing on industry trends, research, and practical experience, this article examines whether demand planning is developing its own AI expectations bubble. It challenges readers to look beyond bold claims, focusing instead on the fundamentals that ultimately determine success: solving the right business problems, improving decisions, and delivering measurable business outcomes.

In today’s environment where ethics are often used to make business decisions, S&OP meetings might become less functional than they were previously. Managers need to be cognizant of this, because not everyone might be wearing their ‘corporate hat’—that is, doing what is best for the company. In this article I advocate for working towards executive-led, financial and operational targets, not moral ideals. To help facilitate constructive S&OP meetings, I highlight research into mindsets taught in MIT’s Masters of Engineering in Logistics program that reveals peoples’ mindsets and strengths in business using colors, including a type that is focused on doing what is considered to be right, perhaps at the risk of business performance. I suggest using this framework to identify such activists, and help ensure S&OP teams act responsibly, not ethically.

Replacement demand for low-voltage automotive batteries is governed by the physical world—temperature, vehicle age, and point-of-sale behavior. Demand Planners for this product typically rely on conventional time-series forecasting which is structurally blind to these drivers. To help solve this problem, this article introduces Multi-Signal Causal Demand Sensing (MSCDS), a methodology that makes causal drivers explicit and the output decomposable. It integrates vehicle-in-operation, weather, point-of-sale, inventory, and pricing signals into a single interpretable forecasting layer. I discuss how deployment of this model has delivered a high-single-digit gain in forecast accuracy at Clarios, the world’s largest manufacturer of low-voltage automotive batteries, with estimated savings in the tens of millions of dollars across the deployed network.

For decades, S&OP/IBP has aligned supply, demand, and financial plans through periodic forecasting and review cycles. While effective in stable environments, this model is increasingly inadequate in industries defined by uncertainty, complexity, and high-stakes outcomes. In this article, I discuss how Contract Development and Manufacturing Organizations (CDMOs) must evolve their planning into a continuous decision-making system focused on execution under uncertainty. I propose a new framework that integrates SISPQ and OTIF objectives, hybrid planning capabilities across S&OP, S&OE, and vendor-centric collaboration. This ‘always-on’ control tower provides real-time visibility into risks across supply, capacity, quality, and compliance, enabling faster scenario-based decisions and proactive trade-off management.

Machine Learning, using ‘Big Data’, has been around for a decade and yet real improvements in demand forecasting remain elusive. Over-reliant on large data sets, they are rarely capable of forecasting new products for which there is no historical data. I propose a solution to this challenge, utilizing AI agents that incorporate factors derived from domain expertise that could impact demand like marketing investment, products positioned as the centerpiece of negotiations led by the sales division, favorable reactions from business partners, and social media sentiment prior to launch. When analyzing how similar products performed, considering the differences in conditions between the similar cases and the target product, AI agents can help forecast demand in a way that machine learning cannot.

Several growth industries have witnessed a transformation in the central planning problem. Demand is no longer, by default, the hardest variable to understand. Often, demand is visible, urgent, and well-funded, but the ability to serve that demand is limited by scarce components, immature manufacturing capacity, supplier readiness and more. This article argues that in a supply-constrained environment, planning teams must evolve from demand forecasting to orchestrating growth by integrating demand priorities, design scalability, and execution risks. S&OP and IBP must connect the dots to form a unified operating model. The companies that scale best will not simply predict demand more accurately—they will decide which demand matters most, which designs can actually scale, and which execution risks must be converted into reliable capacity.

In the second quarter of 2026, real GDP grew at annualized rate of 1.80%, reflecting, in part, the impact of the conflict with Iran. The labor market, however, has remained relatively resilient with the unemployment rate holding steady at 4.10% in August. At the same time, inflation has eased to and held at 3.40 % for the 12-month period ending August 2026.

In many organizations, the monthly forecast accuracy report shows improvement while the planning system continues to struggle. Inventory still swings unexpectedly and supply teams still expedite orders. How can this happen if we’re hitting our forecast accuracy targets? In this article, I reveal why forecast accuracy is a less-than-perfect KPI for planning teams because, by the time supply teams execute the number, it reflects more than demand alone. I discuss how consumption signals are adjusted for distribution changes, promotions, and commercial expectations before being converted into shipment requirements. I advocate for planning leaders to view accuracy as one metric, not as judgement of planning quality.

Over the past decade, the way supply chain Management (SCM) professionals and Demand Planners have engaged with AI has continuously evolved. As classification/prediction-oriented AI has developed into generative AI and now agentic AI, with these technologies increasingly being applied in business, SCM professionals must reconsider their approach to these tools. In this article, I propose a framework to effectively utilize generative and agentic AI, along with the key skills we need to develop as demand planning and supply chain professionals.

In recent years tech company executives have pushed back against employee activism and pressure to act ethically, with Google firing more than 50 employees who protested the company’s contracts with Israel in the wake of the Gaza war. It raises the question of whether companies should act ‘ethically’ or ‘responsibly’. I discuss the recent dispute between AI company Anthropic and The Pentagon, the former taking a moral stance on the use of its technology, which ended in The Pentagon designating the firm as a supply chain risk. I argue that corporations should favor acting responsibly, taking accountability for an outcome and managing its consequences rather than reducing everything to a black or white moral issue.

After decades of working in consulting for companies across a range of industries, I know that proper analysis of the right data can address many complex business challenges. Through four real-world case studies from my own experience, I discuss how the real challenge lies not in the lack of data but in identifying what is relevant, and then extracting, analyzing, interpreting, and applying it effectively. I also highlight the importance of being willing to challenge core assumptions and generate practical, actionable insights rather than purely theoretical conclusions.

Forecast accuracy is an important metric in demand planning and while error metrics like Mean Absolute Percent Error (MAPE) have value, they are not enough. In this article I discuss how the same MAPE can tell very different stories about demand and have very different implications for the performance of a business depending on the industry and product type. I make the case that forecasts should be measured on bias and stability, and aim to meet the needs of the business rather than treating accuracy as the goal.

How did you end up working in demand planning? I moved into a demand planning role when I moved from working in retail to working for a manufacturing company. I worked for many years in retail replenishment and order forecasting. So when I made the change, I had to move from managing and planning orders in the near-term of a few days to looking “over the horizon” and anticipating future demand 18-24 months in the future. It took me a couple of years to fully make the transition and develop the basic tools required. Fortunately, I had some very talented teammates who helped me develop a new way of looking at the business.

In many companies, demand forecasting and inventory management are often separate processes. With teams on either side of the demand/supply equation working towards to their own score-cards, supply chain performance can suffer. To solve this problem I propose a proactive monitoring framework designed to bridge this “perspective gap” between demand planning (focused on forecast accuracy) and inventory management (focused on turns and coverage). By evaluating demand and inventory signals simultaneously at the SKU level, the framework transitions supply chain operations from retrospective reporting to automated, actionable recommendations.

Real GDP expanded at a 2.00% annualized rate in the first quarter of 2026, reflecting a generally solid pace of economic activity prior to the onset of the Iran conflict. The labor market has remained relatively resilient, with the unemployment rate holding steady at 4.30% in May. At the same time inflation has edged higher, rising to 4.20% in May, the highest in three years, highlighting persistent underlying price pressures despite moderating economic growth.