Operating Principles

What the research shows about organisational AI value

The findings below are drawn from surveys of more than 30,000 executives conducted by McKinsey, BCG, Deloitte, Gartner, Bain, the World Economic Forum, Stanford HAI, MIT, IBM, Kyndryl, Accenture and others between 2023 and 2026. Eight structural findings emerge when the evidence is read together.

Pierre Habib · setmode.io

Tool deployment does not create value

88%

of organisations use AI regularly
in at least one function

17%

attribute 5% or more of EBIT
to generative AI

72% of CIOs report breaking even or losing money on AI. Corporate AI investment reached $252.3 billion in 2024. The technology works. The returns are governed by something else.

McKinsey State of AI (Nov 2025, n=1,993) · McKinsey State of AI (Mar 2025, n=1,491) · Gartner CIO Survey (May 2025, n=506)

The operating model determines the return

Algorithms Technology and data People and process

High-performer investment ratio. 70% of AI implementation challenges are people and process problems. Most organisations invert this ratio.

2.5x

higher revenue growth for organisations with AI-led processes

BCG Where's the Value in AI (Oct 2024, n=1,000) · Accenture Reinventing Enterprise Operations (Oct 2024, n=2,000) · McKinsey State of AI (Mar 2025, n=1,491)

Governance precedes deployment

18%

have an enterprise governance council with authority over AI

28%

have a CEO directly responsible for AI oversight

12%

of Fortune 100 boards disclosed AI education or training

CEO oversight is the element most correlated with higher EBIT impact. Among high performers, nearly half of senior leaders show clear ownership. Among the rest, 16%.

McKinsey State of AI (Mar 2025, n=1,491) · McKinsey State of AI (May 2024, n=1,363) · EY / Harvard Law School Forum (2025)

Workflow redesign produces the most value

Redesigned workflows

25–30%

productivity gains

55%

of high performers have reworked processes

Tool deployment alone

10–15%

productivity gains

tools per worker: gains reverse, errors rise 39%

McKinsey State of AI (Mar 2025, n=1,491) · Bain Technology Report (2025) · BCG / HBR "Brain Fry" study (Mar 2026, n=1,488)

People adopt what they understand

15%

→

55%

Positive employee sentiment when leadership support is present

25%

of frontline employees receive that support.
The gap is about what is communicated, not how often.

Accenture Pulse of Change (2026, n=7,000) · BCG AI at Work (Jun 2025, n=10,635) · WEF Future of Jobs (Jan 2025, n=1,000+ employers)

Alignment converts investment into capability

95% of senior executives report investing in AI

14% have aligned workforce, technology and business goals

The 14% who achieve alignment are three times more likely to have a fully implemented change management strategy.

42%

of companies abandoned the majority of AI initiatives in 2025

Kyndryl People Readiness (May 2025, n=1,000+) · S&P Global 451 Research (2025, n=1,006)

The advantage compounds

Revenue growth
over laggards

1.5x

→

1.7x

EBIT margin
over laggards

1.4x

→

1.6x

Three-year total
shareholder return

1.6x

→

3.6x

20242025

The gap widened on every measured dimension. Only 7% of organisations have fully scaled AI.

BCG Where's the Value in AI (Oct 2024, n=1,000) · BCG The Widening AI Value Gap (Sep 2025, n=1,250) · McKinsey State of AI (Nov 2025, n=1,993)

Structure sustains capability

51%

frontline regular AI usage
stalled

78% for managers. The gap is structural.

36%

of employees believe their AI training is sufficient.
18% of regular AI users received no training at all.

What disappears is the operating rhythm. Designed during the programme, it inherits authority. Designed after, it rarely establishes.

BCG AI at Work (Jun 2025, n=10,635) · WEF Future of Jobs (Jan 2025)

The structural finding --mulberry: #70384E; --mulberry-tint:#EEE7EA; --mulberry-hover:#5C2E40; --sage-deep: #4C644A; --clay-deep: #735245; --slate-deep: #48535B; --sage-tint: #EDF0EC; --clay-tint: #F3EDEB; --slate-tint: #ECEEEF; --dark-ground: #242628; --dark-border: #3A3C42;

The evidence from more than 30,000 surveyed executives across 16 institutions converges on a single structural finding. The organisations generating value from AI are different from the majority in how they are designed, rather than in what they have deployed.

The design differences are specific: governance established before deployment, workflows redesigned rather than augmented, measurement baselines captured before the first deployment, human roles explicitly specified, and an operating rhythm that sustains capability after the programme concludes.

These are operating model decisions. They take days to weeks when made at the outset. They take months, or never happen, when deferred. The gap is structural. The remedy is structural.

Sources

McKinsey — The State of AI: Agents, Innovation, and Transformation (Nov 2025, n=1,993)

McKinsey — The State of AI: How Organizations Are Rewiring to Capture Value (Mar 2025, n=1,491)

McKinsey — The State of AI in Early 2024 (May 2024, n=1,363)

BCG — Where's the Value in AI? (Oct 2024, n=1,000)

BCG — The Widening AI Value Gap (Sep 2025, n=1,250)

BCG — AI at Work 2025 (Jun 2025, n=10,635)

BCG / HBR — When Using AI Leads to "Brain Fry" (Mar 2026, n=1,488)

Gartner — CIO Survey on AI ROI (May 2025, n=506)

Deloitte — State of AI in the Enterprise, 7th Ed. (Mar 2026, n=3,235)

Bain — Technology Report 2025; Parsing How Winners Use AI (2025)

WEF — Future of Jobs Report 2025 (Jan 2025, n=1,000+ employers)

Stanford HAI — AI Index Report 2025 (Apr 2025)

Kyndryl — Why Most Businesses Are Not Yet Winning with AI (May 2025, n=1,000+)

S&P Global / 451 Research — Voice of the Enterprise: AI & ML (2025, n=1,006)

IBM IBV — CEO Study 2025 (May 2025, n=2,000)

Accenture — Reinventing Enterprise Operations (Oct 2024, n=2,000); Pulse of Change (2026, n=7,000)

IDC / Lenovo — CIO Playbook 2025 (Feb 2025, n=1,520+)

EY / Harvard Law — How Boards Can Lead in a World Remade by AI (Feb 2026)