Over the past few years, many organisations entered an experimentation phase with AI. That experimentation phase was not wasted time. Running pilots, exploring what the technology could do in specific contexts, building internal awareness and separating useful applications from overhyped ones all had genuine value. It was the appropriate response to a period of uncertainty about what was actually possible and what was worth pursuing. That period is over. BCG’s 2025 global survey of 1,250 senior executives found that 60% of organisations report minimal revenue or cost impact from their investments, and do not yet have the capabilities to scale what they have built. McKinsey found that only around one third of organisations report deploying at scale beyond isolated functions. The dominant characteristic of the current moment is not the absence of activity. It is the absence of progress from activity to outcome. And the organisations that have not yet confronted the reason for that gap are making a strategic error, even if they have a portfolio of impressive pilots to point to.
Experimentation is no longer sufficient. The question now is: what are we actually building, for what purpose, and what does the organisation need to look like to build it well? Execution is the new challenge.
The Gap Between Pilots and Production
The data on how many pilots fail to reach production is bleak. BCG found that only 22% of organisations have moved beyond the proof-of-concept stage with any meaningful initiative. McKinsey found that 62% of organisations remain in experiment or pilot mode. Research from KPMG found that 70 to 87% of enterprises have launched initiatives, but only a fraction have scaled to sustained production use. In 2025, approximately 95% of pilots in the enterprise context failed to produce measurable financial returns. By 2025, 42% of companies had abandoned most of their projects entirely, more than double the rate of the year before.
The consistent finding across all of this research is that the failure is not technical. The organisations that struggled did not lack capable models or sufficient budgets. The analysis from enterprise leaders reviewing 2025 found the same structural deficiencies appearing repeatedly: unclear definition of which problems justified investment, operational unreadiness to absorb the workflow changes that scaling requires, and the absence of the discipline to concentrate effort rather than scatter it. BCG found that high-performing organisations pursue roughly half as many initiatives as struggling ones, but scale far more of them. Concentration, not breadth, is what moves pilots into production. That is a leadership decision before it is an operational one, and it is the decision most organisations have been avoiding.
What Execution Exposes
The move from experimentation to execution is where organisations discover things about themselves that were previously invisible. Governance structures that were adequate for isolated experiments become consequential when systems are deployed into live operations affecting multiple functions. Ownership that was ambiguous during a pilot creates real accountability gaps at production scale, because when something goes wrong or stalls, there is no clear answer to the question of who is responsible for resolving it. Fragmented decision-making that caused only minor friction during controlled testing generates real operational cost when an initiative touches business units with different priorities, different definitions of success, and no shared framework for resolving the differences.
None of this means that leadership teams were totally negligent during the experimentation phase. Execution is a different environment from experimentation, and it tests different organisational qualities. The leaders who navigate it well are not necessarily those who were most energetic about the technology during the exploration period. They are those who can provide structural clarity under pressure, make and hold on to difficult prioritisation decisions, and sustain engagement through the inevitable periods when results are not yet visible.
BCG’s research on what distinguishes high performers from the rest identifies the core distribution of effort: 70% of the value from successful scaling comes from people and processes. That is, it comes from the workflow redesign, the change management, the governance structures, and the operating model decisions that determine whether any capability is actually absorbed into how the organisation works. The technology itself accounts for 30%. Most organisations are investing that ratio in reverse.
What The Shift to Execution Requires
Three changes in leadership behaviour mark the difference between organisations that scale and those that do not.
The first is from curiosity to commitment. Experimentation is appropriately exploratory: try something, observe, learn, adjust. That posture served the previous phase well. Execution requires something different. It requires public commitment to a small number of defined objectives, with the willingness to direct resources, remove organisational obstacles, and hold the work accountable to outcomes over time rather than to demonstration milestones. McKinsey’s research found that high-performing organisations are three times more likely to have senior leaders who demonstrate active, committed, and sustained ownership; not periodic oversight, but genuine sponsorship that continues through the periods when short-term results remain ambiguous. At struggling organisations, approximately 16% of respondents report that level of executive commitment. At high performers, nearly half do.
The second is from breadth to concentration. The portfolio of parallel experiments feels productive. It creates the impression of strategic momentum. It also dissipates resources, creates competing governance demands, and prevents the deep organisational investment that any single initiative needs to reach production. The leaders creating real returns have made the harder decision: to choose a small number of initiatives and commit to them seriously, which necessarily means declining to pursue many others. That choice of the strategic ‘no’ is as important as any ‘yes’.
The third is from technology strategy to operating model strategy. The question is no longer which tools to adopt. It is how the organisation needs to be structured, in terms of roles, workflows, accountability, and performance metrics, to derive lasting value from what it deploys. This is not a question that the technology team can answer on behalf of the organisation. It is a question that requires senior leaders to engage with the reality of how their organisation currently operates, and what needs to change in that reality for any capability to be genuinely embedded rather than layered onto old structures and old habits.
The performance gap between organisations that have made this shift and those that have not is already measurable. BCG’s research found that future-built organisations — the roughly 5% that have moved through experimentation to sustained production and are generating material returns — achieve 1.7 times higher revenue growth, 3.6 times greater three-year total shareholder return, and 1.6 times higher EBIT margins than counterparts still circling in pilot mode. And they are actively reinvesting those returns into further capability, widening the gap with deliberate intent.
The experimentation phase gave most organisations broadly comparable starting points. But the pilot was never the destination. The leaders who understand that, and who are willing to do the harder organisational work that execution demands, are the ones visibly scaling.
