SupplyChainTalk: From Forecasting to Foresight - Rethinking demand planning in response to rapid market change 

Views on news 
The EU, Netherlands, Germany and Greece have become the latest US allies to join Pax Silica, an American-led effort to bolster AI-related tech supply chains as the west and its allies face rising competition from China. The US created Pax Silica last year to secure AI supply chains in everything from chips and critical minerals to energy.

 

To support AI innovation, the state department would also sign a memorandum of understanding with Stanford University to create a new curriculum focused on manufacturing to fill what he called a “major gap” in the US educational system. Pax Silica seems to be less about the supply chain and more about securing the supply of strategic minerals.

 

As the initiative relies on involving private companies, the key to its success is whether businesses can clearly see what benefits joining the alliance can bring them.  

 

Switching to more adaptive demand planning  

Adaptive planning means shorter feedback loops between signal and decision, which requires major changes in how a company operates. Available tools now can help businesses see the change more quickly. However, you can’t keep changing your planning continuously – sales teams and cross functional collaboration must validate whether an emerging trend is relevant for the business or not, while exception planning must decide whether the change is big enough to require any readjustment. The most rewarding strategy is to segment your market and alternate traditional and adaptive forecasting based on where they perform the best – the former can excel in a more predictable market environment, while ML-enabled forecasting can handle market volatility.  

 

To get good outcomes from investment in sophisticated technology systems, first, you need trained planners with analytical minds, as well as cross-functional support and participation on an executive level. In terms of ML, users expect more transparency on what the AI is doing.

 

Here, a transitory period, where users can see both the human and the AI forecast and compare the two may enhance trust in automated systems in the long run. Buy-in can also be increased when users see how a 10 per cent more accurate forecast can save money through, for example, inventory avoidance – dollar savings can be more convincing than percentages without a dollar value attached to them. 

 

Increasingly, companies let supply chain reality impact product design too. Changes that don’t impact the customer experience, even if only minor, can lead to significant savings in tariffs or reduced risk in the supply chain. For scenario planning to bring value, scenarios must be linked to KPIs such as financial delivery, customer health and SLAs to capture the value of foresight including the money saved by avoiding disruption.  

 

The panel’s advice 

• You don’t need to wait till your data is ready for ML – start with shipment and order history and general ERP data. 
• Find planners who understand both the maths behind MLs and the business context.