- Numbers reveal a series of uncomfortable disconnects between what leaders know they should do and what their companies are actually doing.
- High performers are roughly seven times more likely to redesign workflows and reshape the business end-to-end with AI.
Nearly nine in ten CEOs report that their organisations are already capturing measurable cost or revenue benefits from artificial intelligence (AI) in targeted areas.
It is an encouraging statistic — one that underscores how far enterprise AI has come in a remarkably short time. But behind that headline lies a more sobering reality: most companies are hitting a wall when it comes to scaling those early wins into broad, enterprise-wide financial impact.
And the culprit, according to a new study from Boston Consulting Group, is not the technology itself. It is execution.
The findings, drawn from a global survey of 152 chief executives at companies with revenues of at least $500 million, map out the specific leadership behaviours that correlate most strongly with superior AI outcomes.
Collectively, they signal that the AI conversation has entered a distinctly new chapter. For a growing number of organisations, the central question has shifted. It is no longer whether AI can create value. It is whether the organisation can muster the discipline, velocity, and accountability needed to transform that potential into something durable and large-scale.
“The disciplines that drive successful transformation haven’t changed,” said Nicolas De Bellefonds, a managing director and senior partner at BCG and a coauthor of the report.
“What AI does is raise the stakes on both sides: the potential is larger, but so is the penalty for weak execution. That’s why the gap between CEOs who understand what’s required and those actually doing it is so consequential right now.”
Execution gap
For all the attention AI receives as a technological frontier, the CEOs surveyed by BCG pointed overwhelmingly toward organisational execution as the tallest barrier standing between them and scaled value. The numbers reveal a series of uncomfortable disconnects between what leaders know they should do and what their companies are actually doing.
More than half of CEOs identified the need to connect AI initiatives directly to the profit-and-loss statement as a critical barrier. Yet only 14 per cent have clearly defined the P&L impact for every AI initiative they undertake — a gap of 42 percentage points that speaks to a widespread accountability vacuum.
Similarly, 55 per cent of CEOs cited people redesign as a major obstacle, but just 30 per cent include human resources in their AI governance structures, compared with 82 per cent who include technology. The message is clear: organisations are staffing AI governance as though it were a tech deployment, when in reality it is a transformation challenge that demands deep integration with the functions that manage people, processes, and incentives.
Other barriers reinforce the pattern. Weak value tracking, insufficient funding for people and change management, and a lack of mechanisms to hold business leaders — not just technology teams — answerable for AI results all surfaced as persistent drags on progress.
Experimentation without transformation
AI pilots occupy an ambiguous space in the enterprise landscape. On one hand, they serve as essential launch pads: controlled environments where companies can test hypotheses, build confidence, and generate early proof points.
On the other hand, they can become a comfortable inertia — a way to signal innovation without forcing the hard organisational changes that true transformation demands.
Nearly two-thirds of CEOs say their companies pursue AI pilots. But only 26 per cent have embedded AI as part of a broader business transformation. That gap matters enormously because the economics of piecemeal experimentation look nothing like the economics of end-to-end reinvention.
High performers in BCG’s sample are roughly seven times more likely to redesign workflows and reshape the business end-to-end with AI, rather than layering intelligence onto existing processes and calling it progress.
The distinction is not semantic. Companies that treat AI as a bolt-on capability tend to see bolt-on returns. Those that treat it as a reason to rethink how work gets done — who does what, how decisions flow, where handoffs occur — unlock a fundamentally different order of magnitude in value.
Four transformational moves
BCG’s research, reinforced by the practices of the highest-performing companies in the survey, identifies four transformational moves that distinguish organisations scaling AI value from those still cycling through pilots.
Make the CEO the orchestrator — but make the business accountable. The chief executive has a distinct and irreplaceable role in AI transformation. It is to set an ambitious, unifying vision for what AI will make possible and then cascade delivery accountability to CXOs and P&L owners. When accountability sits with the technology function alone, AI remains a tool in search of a problem. When it sits with business leaders who own revenue, cost, and customer outcomes, AI becomes a strategic lever.
Focus AI on a few high-value domains that can genuinely change the business. The temptation to scatter AI experiments across the enterprise is strong, but the evidence argues against it. Leading companies concentrate people, capital, and organisational capability on a small number of domains where AI can deliver transformational — not incremental — impact. Critically, they fund these efforts on a multi-year horizon, recognizing that meaningful transformation does not fit neatly into annual budget cycles.
Set up AI projects so value can be tracked from day one. Before any AI initiative launches, high-performing organisations define the expected P&L impact and the value logic that underpins it. Finance is brought in at the start to validate results, not at the end to audit them. This discipline does more than ensure rigor; it forces clarity about what problem the AI investment is actually solving and how success will be recognized when it arrives.
Prioritise people and change management over technology deployment. Perhaps the most telling statistic in BCG’s research is this: high performers are 2.4 times more likely to assign their best talent to AI workstreams. They treat AI transformation as a talent magnet, not a side project. They invest in training, redesign roles, and manage the organisational disruption that accompanies any genuine shift in how work gets done. And they embed HR in governance from the outset, recognizing that technology without adoption is overhead.
A more consequential phase
“The early AI results are real, and they are encouraging,” said Matthieu Berthion, a managing director and partner at BCG and a coauthor of the report. “But companies are now entering a more consequential phase: turning targeted gains into enterprise-wide impact. That requires moving beyond the deployment of technology and building the execution discipline to reinvent how the business runs — from governance and workflows to accountability and value measurement.”
Berthion’s framing captures the inflection point clearly. The first wave of enterprise AI was about proving the technology could work. That case has largely been made. The second wave — the one most companies are now navigating — is about proving the organisation can work differently. It demands a level of integration between strategy, finance, talent, and operations that few enterprises have historically achieved.
For CEOs, the implication is bracing but also clarifying. AI success in the years ahead will not be determined by which models an organisation deploys or which vendors it partners with.
It will be determined by whether the CEO can orchestrate a transformation that links vision to execution, assigns accountability to the people who control resources, concentrates investment on a few big bets, and treats the human dimensions of change as central rather than peripheral. The technology is ready. The question now is whether the organisation is, too.