By Terence Tse
If you still think that AI adoption is a technology issue, think again. The companies that really succeed are all about their people and processes.
What’s driving the AI revolution is, of course, the technology. But there’s more to it than that. To really succeed, organisations must focus on two crucial factors: the people they’re expecting to make it work, and the processes they’re targeting. And, perhaps most importantly, they need to measure the right things.
Companies invested approximately $252 billion in AI in 2024 alone.1 Despite this, returns remain limited. A survey of nearly 2,000 companies across 105 countries found that 88 per cent use AI in at least one business function, but only 6 per cent report an EBIT increase of more than 5 per cent.2 Another study of 300 AI deployments found that 95 per cent of generative AI pilots show no measurable P&L impact.3 Similarly, a report on 1,250 executives found that only 5 per cent of companies are “future-built”, generating outstanding returns from AI.4 In financial services, just 14 per cent of respondents view AI as transformational to their strategy and competitive advantage.5
What has gone wrong? The issue is not with AI technology itself, but with how businesses approach its implementation.
The Wrong Clock
The most significant failure often goes unnoticed: the practice of evaluating AI programmes using the same financial criteria as conventional software projects. A survey of 1,854 senior executives across 14 countries found that typical AI use cases take two to four years to deliver positive ROI, which is three to four times longer than traditional technology investments. Only 6 per cent of respondents saw returns within a year and, even among the most successful, just 13 per cent achieved payback within 12 months.6
The most significant failure often goes unnoticed: the practice of evaluating AI programmes using the same financial criteria as conventional software projects.
This mismatch in expectations often leads organisations to cancel AI programmes just as they are about to deliver results. In a survey of 782 infrastructure and operations leaders, 57 per cent cited “expecting too much, too fast” as the reason for AI project failure. 7 However, these failures can provide valuable learning opportunities. Another report found that 61 per cent of respondents had failed with AI before succeeding in subsequent projects.8
The rise of agentic AI may further exacerbate this issue. Organisations are now adjusting ROI measurement for different AI technologies, using shorter timeframes for generative AI (GenAI), which focuses on efficiency and productivity, and longer timelines for agentic AI, which targets cost savings, process redesign, risk management, and long-term transformation.9
The Invisible Baseline
Measuring AI adoption is essential for successful transformation. However, many organisations do not track returns at all. While 89 per cent of enterprises have adopted AI tools, only 23 per cent can accurately measure ROI.10 Some deploy AI without establishing a pre-deployment baseline, making it impossible to assess true impact. In a survey of 1,006 C-suite executives, only 11 per cent assess AI value before and after implementation, while 32 per cent assess only before implementation and never validate outcomes.11 Additionally, “shadow AI” – the unauthorised use of AI tools – remains unmeasured. A previous study found that 67 per cent of enterprises lack full visibility into which AI tools employees use.12 Shadow AI can be harmful: 57 per cent of employees using unapproved platforms admit to entering sensitive company information.13
Without proper measurement, initiatives cannot demonstrate value, even if successful. The most important AI investment is in measurement, not just in models or solutions. A survey of 325 senior leaders provides some insights into the data debt problem.14 It found that 94 per cent believe that well-connected data, processes, and applications are crucial to successful AI adoption. Nevertheless, only 27 per cent say those elements are currently well connected within their organisation. Even scarier, while 65 per cent of respondents say their structured data is at least partially AI-ready, only 39 per cent say the same about unstructured data (i.e., the emails, PDFs, call transcripts, and institutional records that underpin most day-to-day business processes). Furthermore, over half of the organisations cited data silos as their most common barrier to AI adoption.
Additional surveys confirm these findings. One report states that 36 per cent of organisations see improving data quality, integration, and interoperability as the greatest opportunity to extract more value from AI in finance.15 Another study identifies data availability and quality as the leading challenges to AI adoption.16
The Wrong Order: Technology First, People Last
Our research in Becoming AI Native: Charting the Next AI Frontier shows that workforce capability is a major barrier, often greater than data debt. The most reliable predictor of AI value is the ratio of investment in people and processes versus technology.17 BCG finds that the top 5 percent of companies allocate about 70 per cent of AI resources to people and process redesign, 20 per cent to technology, and 10 per cent to algorithms. In contrast, most failing organisations spend 93 per cent of their AI budgets on technology, leaving only 7 per cent for organisational needs such as training and change management.18 By 2027, companies prioritising AI literacy among executives are expected to achieve 20 per cent higher financial performance than those that do not.19
Research shows that people-focused companies achieve better AI outcomes than those focused solely on technology. Organisations that prioritise human-machine collaboration and redesign roles and processes are more likely to realise measurable returns.20 Companies investing in AI talent plan to add three times as many full-time equivalents for upskilling and are twice as likely to involve employees in adoption.21 The key is to evolve alongside AI, not just adopt it.
The Wrong Ownership
Value creation requires clear ownership. The key question is to whom ownership should be assigned. A recent survey found that most organisations assign AI value accountability to chief data, analytics, or AI officers, or to functional executives. However, when the CFO is responsible, 76 per cent of organisations report significant value, compared to 53 per cent under CIOs or CTOs and 32 per cent under functional executives.22 This reflects the CFO’s evolving role as an orchestrator of AI-driven transformation, balancing finance, technology, and strategy while upskilling teams and managing risks.23
As noted, over half of companies failed in AI experiments before succeeding. This often results from misaligned ownership, where teams treat AI as a technology project rather than a process and change-management initiative. Early failures typically occur when technical teams lead without business ownership.24
The Pattern Is Clear. The Path Forward Is, Too.
These failure modes – mismatched timelines, absent baselines, data debt, budget misallocation, and misaligned ownership – are not random. Organisations that succeed with AI focus on organisational redesign, not just technology. By aligning incentives, redesigning processes, and reconfiguring organisational structures, they turn AI adoption into lasting business value. Treating AI as only a technical upgrade leads to underperformance.
About the Author
Terence Tse is Professor of Finance at Hult International Business School and co-founder at the AI Native Foundation. He is also co-founder and Executive Director of Nexus FrontierTech.
