Figure 1. McKinsey reports a large gap between AI experimentation and enterprise-wide scale.
Almost 90% are experimenting with AI. Only 7% report enterprise-wide scale.
The adoption gap is now an operating problem
Artificial intelligence has moved quickly from experimentation into the mainstream of management agendas, but enterprise-wide use remains uncommon. McKinsey’s June 2026 Operational Excellence Survey found that almost 90 percent of organisations are at least experimenting with AI, while only 7 percent report scaling it across the enterprise. That gap matters because experimentation proves that a tool can work in a controlled setting; scaling requires the organisation itself to work differently. Processes, decision rights, data, measures, skills and management routines have to absorb the technology without losing control of day-to-day performance. The research therefore shifts the AI discussion away from access to technology and towards the operating conditions that determine whether technology becomes routine work.
Operational excellence is the enabling system
McKinsey surveyed 1,000 managers and executives at companies with at least $500 million in revenue and 100 employees. It also analysed performance data for 677 companies from 2014 to 2024. The study found that stronger operational-excellence practices were associated with higher profitability, while organisations further along in AI deployment also showed stronger productivity and financial performance. The important point is not that operational excellence and AI are interchangeable. They solve different problems. Operational excellence creates a coherent system for managing work, and AI can increase the speed, reach and quality of some activities inside that system. When the operating system is weak, adding technology can simply automate inconsistency or move bottlenecks elsewhere.
The strongest results appear when both mature together
The survey reports a distinct performance difference between organisations that keep AI confined to a few functions and those that embed it more broadly. Companies with enterprise-wide AI deployment generated nearly double the profit margins of peers using AI in only a few departments, while their three-year return on invested capital was more than five times higher. High operational-excellence maturity showed a parallel pattern, including more than three times the three-year return on invested capital of less mature peers. McKinsey is careful to describe these relationships as correlations rather than proof of causation. Even with that qualification, the pattern is useful because it shows that scaled technology and disciplined operations tend to appear together in higher-performing organisations.
Figure 2. Performance patterns reported in McKinsey’s survey; the relationships are correlations, not proof of causation.
Productivity is where the connection becomes practical
The research identifies productivity as the main mechanism linking operational maturity and AI adoption to financial outcomes. Higher-maturity organisations were concentrated in the stronger productivity-improvement bands, and technology-enabled operational excellence amplified the effect. This makes operational sense. A process with clear measures, stable inputs, visible exceptions and defined ownership creates better conditions for automation and machine-assisted decisions. AI then has a clearer objective, cleaner feedback and a place in the workflow. The resulting productivity gains can create additional capacity and capital for further improvement, which strengthens the system again. McKinsey describes this relationship as a reinforcing performance loop rather than a one-off technology implementation.
Process redesign comes before meaningful scale
One example in the research is a global automotive manufacturer that began by simplifying processes and redesigning workflows after identifying technical debt and highly manual internal work. AI was then introduced into the redesigned environment. McKinsey reports software-development productivity gains of up to 44 percent, while AI-supported sales tools contributed to roughly 40 percent growth in incoming orders. The organisation identified long-term efficiency potential of about 20 to 25 percent as adoption expanded. The sequence is important. The company did not treat AI as a layer to place over unchanged processes. It first examined what customers needed, mapped work end to end, removed unnecessary complexity and then used AI where it could materially improve execution.
Operational discipline also controls the downside
Scaling AI changes more than task speed. It changes who makes decisions, what information they use and how quickly an error can travel through a process. Operational-excellence practices provide control points for that change. Clear key performance indicators make outcomes visible. Defined decision rights establish accountability. Real-time data supports faster intervention. Continuous customer feedback tests whether efficiency is improving the experience or merely reducing internal effort. These practices are familiar, but AI increases their importance because poorly governed automation can amplify variation at a scale that manual work rarely reaches.
Figure 3. Operational excellence and scaled AI can reinforce each other through productivity, feedback and reinvestment.
The practical lesson is sequencing, not delay
The research does not support waiting until an organisation reaches some ideal level of maturity before using AI. It supports choosing operational priorities, building the minimum management discipline required around them and scaling technology in step with the organisation’s ability to absorb change. McKinsey highlights focused outcomes such as throughput, yield, service levels and asset utilisation as useful anchors. That approach keeps the business case tied to performance rather than novelty. It also creates a testable question for every AI initiative: which part of the operating system becomes measurably better because this technology is present?
AI does not remove the need for operational excellence
The more organisations use AI, the more important their operating systems become. Enterprise scale requires consistent processes, reliable data, clear accountability and routines that detect whether performance is moving in the intended direction. AI can strengthen those capabilities, but it cannot substitute for the organisational choices that create them. McKinsey’s 2026 research therefore points to a practical conclusion: the organisations most likely to convert AI investment into sustained results are not simply those adopting the newest tools. They are the organisations that can integrate those tools into a disciplined system of work, learn from the results and improve the system again.
Sources
Primary research: McKinsey & Company, “Putting AI to work: The operational excellence imperative”. 19 June 2026. https://www.mckinsey.com/capabilities/operations/our-insights/putting-ai-to-work-the-operational-excellence-imperative
Research context: McKinsey & Company, “Next-generation operational excellence”. 2026. https://www.mckinsey.com/capabilities/operations/our-insights/next-generation-operational-excellence
Leave a Reply