In all the discussion surrounding the promise of AI, one thing that is not up for debate is that there is a lot of value at stake. It’s also true that while companies are investing an immense amount in AI, both in capital and leadership focus, they are seeing limited results. Indeed, McKinsey’s annual survey looking at AI ROI shows that only around 6% of respondents attribute at least 5% of EBIT to their use of AI and describe the technology’s impact as “significant” so far.
By contrast, when we see exciting examples of the successful deployment of agentic AI systems in business, these companies all share one thing in common: they have reimagined workflows end-to-end, with AI enabling the redesigned operating model, rather than simply automating the one they have today. Recent MIT MIMO and McKinsey research shows the performance gap between AI leaders and the rest is widening: Leaders achieved a 2.7x average impact multiple in 2021, 3.8x in 2023, and 4.7x in 2026.
A critical element that many AI transformation programs are missing is operational excellence (OE) as a core foundation. Operational excellence means excelling at more familiar management activities, such as having clear KPIs and data quality accountability, disciplined performance management, timely resource allocation, and robust operating systems.
Only when operational excellence sits at the heart of domain and workflow reimagination can AI-enabled ways of working translate into bottom-line impact. And in manufacturing and supply chains, where execution determines margin, the real productivity story is that AI can deliver enterprise value only if the foundations of management maturity are in place.
It’s operational excellence that enables AI to scale effectively across organizations rather than languish in experimental stages. We recently surveyed 1000 executives and managers from 696 unique companies with over $500 million in annual revenue. The results paint a clear picture of how operational excellence enables the scaling of AI through the practical production of systems, capability building, and transparent performance management.
In supply chain, you cannot simply insert a better demand-forecasting algorithm and expect transformational value. Demand planning connects to supply planning, S&OP/SIOP, roles, incentives, decision rights, handoffs and management routines.
The message for leaders is that conversations have become AI-centric, but without the requisite consideration of operational fundamentals. Too many organizations are tracking AI pilots, use cases, or adoption rates as success metrics, when those metrics say little about whether AI is improving business performance. Operational excellence is the antidote to an AI-scaling impasse.
To rally support, those in favor of a greater focus on the fundamentals may need to explain what operational excellence might do for a business. Our data shows how companies with mature management systems consistently outperform peers on productivity and profitability, while also demonstrating a greater degree of meaningful AI scaling.
Further, management systems need to connect redesigned ways of working directly to measurable business outcomes. This means an explicit connection between redesigned workflows, through operational metrics and KPIs, to financial value and P&L. AI pilots, adoption rates, or individual use cases are insufficient measures of success.
A review and definition of what comprises operational excellence may also help, stretching as it does, across shared KPIs, real-time performance management, disciplined resource allocation, and continuous improvement. Beyond this, operational excellence touches processes for business case development, workflow improvements, the most helpful reward structures, and the timely acquisition of key talent.
The Supply Chain Crucible
Supply chains are already based on the operational disciplines of planning, scheduling, inventory management, quality, maintenance and logistics. Agentic AI and generative AI can improve these functions dramatically, but only when processes are standardized, and performance management exists.
The pressure on supply chain and manufacturing teams to achieve results is of course internal as well as external. Looking deeper into the missing links for greater tech-enabled productivity will certainly help those teams ask for the right support and encourage them to input most helpfully into wider corporate strategy.
Examples can be persuasive too. We have seen AI agents unlock value for clients, reducing the cost of goods sold by 4-7%, improving productivity by 20-50%, and shrinking decision cycles from weeks to hours, and in some cases, to seconds.
Elsewhere, we have helped an electronics OEM achieve fully native supply chain orchestration through agents — a self-funding journey resulting in significantly faster planning cycles with rapid sense and response to changes across components, freight, fulfillment, replenishment, demand and more.
Other examples are of course available, but every successful example we can cite depends on clean processes, reliable data, clear ownership, and strong management systems before AI can scale effectively. Our aim is to drive operational excellence up the priority list so that AI can flourish and businesses can reap the rewards.
Leaders must be bold enough to ask, “In an unconstrained world, how should this entire process work?” They need to push their organizations beyond incremental automation, and be willing to make fundamental changes to processes, roles, workflows and management systems. At the same time, leaders themselves need greater AI proficiency so they can ask the right questions, understand what becomes possible in an agentic world, and not simply call for new ways of working, but role-model these new behaviors themselves.
At this point in time, a significant opportunity exists for businesses able to achieve and sustain management maturity. The companies pulling ahead aren't simply deploying more AI or outspending competitors, they're running better management systems. If an organization can combine operational excellence with AI deployment, it will be best placed to turn the technology of today into business success. Supply chain and manufacturing remain the ideal proving grounds for this next chapter.
Sal Arora is senior partner and global co-leader of McKinsey’s Operations Practice at McKinsey & Company.
