Every major research institution agrees on the failure rate. Almost none of them say what to do about it. We do.
Gartner, McKinsey, MIT Sloan, RAND, and Forrester have each, independently, documented failure rates for AI and digital transformation projects in the range of 70–80%. The explanation does not vary: the problem is human and organisational, not technical.
Projects fail because expectations are invisible. Because consultation is performed, not practised. Because leaders assume that announcing change is the same as achieving it.
More than 70% of leader expectations are never made explicit to the people acting on them. Everyone is working from assumptions. No one has agreed on anything.
The decision is made. The implementation is planned. Then a meeting is held and called consultation. People sense it immediately. Their silence is not agreement.
People adapt their behaviour in the presence of authority and revert when it is absent. Compliance that looks like commitment is the most expensive misreading in change management.
The pressure to implement quickly bypasses the listening and alignment that make implementation stick. Fast rollout followed by slow collapse is the most common pattern in AI projects.
They are the ones who did the harder work first. They listened — actually listened, not surveyed — to the people whose jobs would change. They made expectations explicit, in writing, with specificity.
When RAND recommends “deep stakeholder engagement,” when Gartner calls for “change readiness assessment,” when McKinsey points to “leadership alignment” — they are all describing Servant Leadership. They have the diagnosis. We have the treatment protocol.
“The AI is not the problem. The AI is ready. The question is whether the organisation — the people, the expectations, the accountability structures — is ready for it.”Nick Anderson, The Crispian Advantage
The four-discipline methodology at the heart of Not Without the People — Listening, Clarity, Patience, Focus — is not a theory. It is a sequence of practitioner interventions, tested in live AI implementations.
Most AI implementations load the technology first and ask people to catch up. The Crispian approach inverts the sequence — people articulate first, AI translates, people review.
The evidence for the people argument has been accumulating since the 1990s. It is the convergence of every major research institution that has examined AI and digital transformation implementation at scale.
Consistently reports that 80% of AI projects fail to deliver business value. Identifies “people and process issues” as the primary cause.
Documents that organisations with strong change management practices are 3.5× more likely to outperform peers in digital transformations.
Research on AI adoption finds that employee trust, clarity of role, and leadership credibility are the three strongest predictors of successful implementation.
Studies on workforce AI adoption in public institutions identify stakeholder alignment as the critical pre-condition for sustainable change.
Reports that two-thirds of digital transformation failures trace back to cultural resistance that was never surfaced, let alone addressed.
Huthwaite research on “Filter vs. Amplifier” meetings demonstrates that most organisational communication reduces, not transmits, the information leaders need.
Articles, perspectives and research on what AI-driven change actually requires — and why most organisations are approaching it wrong.
When 77% of employees say AI has created more work, not less — that’s not resistance. That’s feedback.
Read → AI AdoptionYou’re asking people to enthusiastically embrace something they believe might eliminate their jobs. Good luck with that.
Read → AI Adoption & PeopleMaking Work More Satisfying Through AI-Enabled Thinking
Read → Change ManagementWhy the Most Critical Success Factor Gets Only 1% of Our Attention
Read → Leadership & PeopleResearch shows the fear of job loss can be as damaging as actual job loss.
Read → AI AdoptionReimagining Consulting for the AI Age
Read →Not Without the People — forthcoming Q1 2027. Register now for publication updates.