Arvind Sreenivasan, Senior Vice President Asia Pacific at Logility, explains how AI, machine learning and advanced planning capabilities such as demand sensing and causal forecasting can overcome weaknesses in traditional forecasting methods and unlock substantial business benefits.
When someone begins a sentence with In an ideal world…, alarm bells should sound in your head. For example: In an ideal business world, traditional methods of demand planning are flawless.
First, the business world is not ideal. Far from it. It is beset by a potent mix of post-pandemic supply chain challenges, inflation, labor shortages, geopolitical tension, demanding-yet-fickle consumers and increasingly bold regulatory bodies.
Second, traditional methods of demand planning are indeed flawed. This is much more than an academic problem. We need look no further than Gartner’s recent research among business executives that revealed their top pain point is demand volatility.
Traditional forecasting models assume that historical data offers all of the information needed to predict trends, but recent and frequent demand shocks have exploded throughout the supply chain and increased overall volatility, exposing the fragility of some models. Why is this important? More accurate demand forecasting drives more reliable downstream plans that boost profitability, satisfy customers and align supply chain partners.
Thinking differently about new demand patterns
Progress requires a comprehensive shift in demand planning assumptions, prioritising forecasts that offer flexible scenario planning to accommodate a wide range of demand. To thrive in the face of challenges presented by new demand patterns, companies are increasingly deploying artificial intelligence (AI) and machine learning (ML) capabilities into their planning processes.
Rather than relying solely on historical data and presumptions regarding how relationships in the data predict the future, AI and ML use real-time calculations to find solutions for supply chain problems. Specifically, ML uses clever algorithms to automatically recognise patterns, capture demand signals, and uncover complex correlations in large datasets. In addition, these systems improve responsiveness by constantly retraining models, adapting them to current conditions.
Put another way, newer methods are given the freedom to learn based on “life in the trenches every day” rather than being forced to accept outdated truths as timeless, universal and indisputable. Underpinning this approach is “agile” philosophy, whose practitioners learn to move quickly, iterate often, fail fast and recover immediately.
Adding short-term precision to long-term demand planning
In addition to AI and ML, demand sensing and causal forecasting have both received high marks for their ability to add short-term precision to long-term planning efforts, especially in high-velocity sectors. Demand sensing is the translation of market-based demand information to detect short-term buying patterns, allowing planners to model the impact of promotions, weather, market share and other factors on forecasts. When compared to traditional time-series forecasting, demand sensing can boost short-term forecast accuracy by up to 30 per cent or more.
Causal forecasting augments your core demand forecasting solutions. It provides a proactive approach rather than basing inventory positions and replenishment schedules solely on shipment data. Companies looking to reduce lead times, optimise inventory or understand the impact of external demand factors can use causal forecasting to improve customer satisfaction, boost cash flow, increase fill rates and reduce costly over/under-stock situations.
By investing in modern platforms powered by technologies like AI, ML, companies will overcome the weaknesses in traditional forecasting methods and improve speed and agility, greatly improving their ability to effectively respond to disruptions.
Let Logility guide you to create the most accurate demand forecasts possible with the technology you need in our modern, messy world. Learn more here.