The 5% difference: what winners do differently
Neil D. Morris
February 23, 2026
After two years of research for the book (50+ peer-reviewed studies, published implementation case studies, and my own work as a three-time CIO and advisor), I've reached a conclusion that should be heretical but isn't:
The technology matters least. The same models are available to everyone.
The same platforms. The same capabilities. Access to AI is table stakes.
Yet 95% of the organizations MIT NANDA studied were getting no measurable P&L return from generative AI (MIT NANDA, 2025).
The difference is organizational execution.
The framework that emerged
Over 25 years in technology leadership, I organized what I learned into the Seven Pillars of AI Leadership. Each pillar addresses a specific failure mode that derails implementation.
Strategic Clarity means knowing exactly what problem you're solving before you choose a technology. Define your North Star before touching technology.
Leadership Alignment means executives who show up when things get hard, remove blockers personally, and make difficult decisions about resources and organizational change. Build the coalition that makes transformation possible.
Capability Building means building skills alongside technology. Create permanent assets, not vendor dependencies. BCG's 10-20-70 rule puts 10% of the effort on algorithms, 20% on infrastructure and 70% on people, process and culture (BCG, 10-20-70 rule).
Pilot Discipline means measuring real business outcomes against honest baselines with complete cost accounting. Kill zombie projects before they multiply.
Scale Strategy means changing workflows as you grow. AI layered on top of broken processes produces broken processes with AI. Bridge the pilot-to-production gap.
Risk Management means protection that enables progress. The 5% build governance frameworks that manage real risks while creating space for experimentation. Enable progress while building guardrails.
Continuous Evolution means the work does not end at launch. Compound advantages through learning discipline.
Why the pillars work together
Individual pillars aren't revolutionary insights. You've heard versions of all of them.
What matters is integration. Organizations fail when they excel at one or two pillars while ignoring others.
Strong technology capabilities without leadership alignment produces pilots that never reach production. Excellent data without process redesign produces insights no one acts on. Good governance without capability development produces policies no one can implement.
The 5% who succeed treat the pillars as a system. They assess themselves honestly against each one. They invest in their weakest areas rather than over-investing in strengths.
The pattern in the research
The published case studies I've analyzed, including JPMorgan Chase and Walmart, share common elements.
They started with infrastructure, not applications. JPMorgan built a data platform managing 500 petabytes of proprietary information and made it "AI-ready" years before generative AI emerged. When large language models arrived, they had a foundation to build on. Organizations trying to build foundations while simultaneously deploying applications struggle with both.
They chose partnerships strategically. MIT NANDA found that buying from specialized vendors and building partnerships succeeds about 67% of the time, while internal builds succeed only one-third as often (MIT NANDA, 2025, as reported by Fortune). The successful organizations didn't treat "build versus buy" as a religious question. They built where they had genuine competitive advantage and partnered everywhere else.
They invested in learning by doing. The successful organizations built adoption through senior leader modeling, onboarding integration, and continuous feature development driven by employee feedback. Training was embedded in daily work.
They accepted realistic timelines. Even JPMorgan, with $17 billion in annual technology spend (JPMorgan Chase, 2024), acknowledges most of their 450 use cases remain developmental. The 12+ months to measurable value is reality for everyone.
They measured multiple dimensions. Beyond cost savings, they tracked employee satisfaction, customer outcomes, strategic positioning, governance compliance, and capability development. ROI was important but not the only measure that mattered.
The counterintuitive truth
Slower starts lead to faster results.
This sounds backwards. We're told to move fast, fail fast, iterate quickly. And that's true for individual experiments.
But organizations that rush into AI deployment without foundational work spend their speed building on sand. The rework, course corrections, and failure recovery consume more time than doing it right from the beginning.
The 5% who succeed don't have a higher velocity of initial deployment. They have a higher percentage of initial deployments that work. They take longer to start and finish faster.
In martial arts, there's an old saying: "slow is smooth, smooth is fast." You build speed by perfecting technique slowly first. Rushing to speed before technique is solid produces sloppy movement that's actually slower.
The same principle applies. Master the fundamentals (data readiness, governance, process design, capability development) before trying to move fast. The organizations that skip fundamentals in favor of speed end up slower than those that invested the time upfront.
What AI-ready actually means
Every organization thinks they're ready for AI. Few actually are.
Being AI-ready means:
- Your data is accessible, clean, and documented, not trapped in silos with quality issues no one wants to address.
- Your leadership is aligned on priorities and willing to make hard decisions about resources and organizational change, beyond enthusiasm for the technology.
- Your governance frameworks are established and functioning, not theoretical policies that haven't been tested.
- Your processes are understood well enough to redesign, not legacy procedures that exist because "we've always done it that way."
- Your people have the skills and psychological safety to experiment, beyond familiarity with current tools.
Honest self-assessment against these criteria is uncomfortable. Most organizations overestimate their readiness because acknowledging gaps means acknowledging work to be done.
But the organizations that succeed start with honest assessment. They know where they're weak and invest accordingly. The organizations that fail often don't know what they don't know, and discover it too late.
The permanent economic expansion
Penn Wharton's economic modeling projects AI will permanently expand GDP by 1.5% by 2035, 3% by 2055, and 3.7% by 2075 (Penn Wharton Budget Model, 2025).
That's not a temporary productivity boost. It's a permanent increase in economic capacity.
IDC projects a cumulative global impact of $22.3 trillion by 2030, with each dollar spent on AI solutions and services generating an additional $4.90 in the global economy (IDC, April 2025).
This is the prize. It's real. It's enormous. And it will be captured by organizations that execute well.
The question isn't whether AI matters. The question is whether your organization will be in the 5% who capture the value or among the many who invest without returns.
The research is clear about what separates the two groups. The framework exists. The path is knowable.
What remains is execution.
Neil D. Morris. I wrote Why AI Fails, and as a fractional technology executive I show boards and executive teams one hard truth: when AI fails to deliver value, the cause is leadership. Read more at neilmorris.ai →
Ready to assess your AI readiness?
Take the free AI Leadership Readiness Assessment and get personalized insights powered by the Seven Pillars of AI Leadership.
Take the free assessmentWant to discuss this topic?
Schedule a consultation with Neil to explore how these insights apply to your organization.
Schedule a ConsultationGet more insights like this
Weekly AI leadership insights delivered to your inbox.