Italian manufacturing runs on mature planning tools (ERP, MRP, APS, MES) but operates in a structurally unstable context, where the optimal plan loses relevance within hours. The real bottleneck is neither the machine nor the calculation engine, but the process of deciding how to adapt the plan when conditions change. The evidence shows that this constant re-deciding has become the most expensive and most poorly spent part of the job. Beyond a certain threshold, improving the forecast yields less than improving the ability to simulate scenarios and choose. The frontier is decision-driven production, with AI supporting — not replacing — human judgment

KEY INSIGHTS

• 23 percentage points of latent production capacity. Average OEE in discrete manufacturing sits between 60% and 67% against a world-class threshold of 85%, reliably reached by only about 3% of plants — a gap driven largely by operational decisions (downtime, micro-stoppages, setup changes), not by technology.
• Managers spend 40% of their time deciding, and 61% of it is judged ineffective. In a McKinsey survey of over 1,200 managers, executives dedicate roughly 40% of their time to making decisions and consider 61% of that time ineffective — an estimated cost of ~530,000 person-days per year (~$250 million) in a large enterprise; 72% admit that wrong decisions are as frequent as right ones.
• Spending on SCM software with agentic AI rises from under $2 billion to $53 billion by 2030. Over the same period, the share of companies using agentic-AI-enabled SCM software climbs from 5% to 60%, and 70% of large organizations will adopt AI-based forecasting (Gartner) — confirming the shift toward decision-centric planning