The 6% Playbook
This is Part 5, the final post in this series. Part 1 exposed the $37 billion failure. Part 2 diagnosed five root causes. Part 3 revealed the winning architecture. Part 4 showed why the agentic wave makes platform architecture urgent. This week: what the winners actually do and how you can join them.
What the Top 6% Do Differently. Turn AI into measurable business value.
Over the past four weeks, I've walked through the evidence on enterprise AI: $37 billion spent [1], 88% adoption [2], 6% winning [2], and five structural failure modes that have nothing to do with the technology itself.
Now the question that matters: what do the 6% do differently?
The answer, drawn from McKinsey, BCG, MIT, and every other serious research programme, is consistent, replicable, and — crucially — not about having better models.
1. They start with the process, not the technology.
High performers identify a high-volume, measurable business process and redesign it around AI capabilities before selecting the technology. They never start with "we have an AI model, let's find a use case." McKinsey [2] found that workflow redesign is the single biggest predictor of EBIT impact across 25 organisational attributes tested. Not talent. Not budget. Not the model. The process.
This is counterintuitive in a world obsessed with model benchmarks. But it's the finding that separates the 6% from the 94%.
2. They buy platforms, not projects.
The shift from 47% build to 76% buy in a single year [1] reflects hard-won wisdom. MIT NANDA [3] found that purchased solutions succeed 67% of the time versus 22% for internal builds. Mid-market firms on platforms achieve production in 90 days. Large enterprises building internally take nine months.
That six-month gap isn't a scheduling difference. It's six months of learning, six months of compounding value, and six months of competitive separation that the slower organisation may never recover.
3. They govern from day one.
This is the counterintuitive one. Governance sounds like a brake. The data says it's an accelerator. Deloitte [24] finds that enterprises where senior leadership actively shapes AI governance achieve significantly greater business value. McKinsey's high performers try to protect against more risks — because they're deploying at scale and learning from real failures, not sitting safely in pilot mode.
Start with minimum viable governance: basic accountability structures, interim role assignments, clear risk categories. Then build out as maturity grows. With the EU AI Act going live in August 2026 and 31 U.S. states having enacted AI legislation, this isn't optional. Gartner predicts enterprises will invest $5 billion in compliance by 2027 [23]. You can spend that money retrofitting or scaling. Your choice.
4. They empower the line, not just the lab.
AI can't be a central team's hobby. McKinsey's high performers [2] distribute AI ownership to business-unit leaders who understand the workflows, the customers, and the metrics. Central teams provide infrastructure, governance, and standards. Business teams drive adoption and iteration.
MIT NANDA [3] confirmed this: empowering line managers — not just central AI labs — to drive adoption is a key differentiator of the 5% that succeed. The people closest to the process are best positioned to redesign it.
5. They invest disproportionately in data readiness.
Informatica [6] reports that 86% of data leaders plan to increase data management investments, with 44% citing AI readiness as the primary driver. The winners allocate 50–70% of their AI programme timeline and budget to data extraction, normalisation, quality, and governance. It's not glamorous. It's the foundation everything else depends on.
What does this playbook deliver?
The benchmarks are clear: IBM [26] reports an average $3.50 return per $1 invested. BCG [4] finds 26–36% of professional time reclaimed. McKinsey [2] documents 10–20% cost reductions at the function level and revenue uplifts exceeding 10% in marketing, sales, and product development. Bloomberg's survey of 604 senior executives [18] projects a realistic ~7% profit improvement over 2–3 years and ~10% productivity improvement over three years.
These aren't transformative numbers individually. But they compound. An organisation capturing 10% productivity improvement while competitors run failed pilots doesn't just save money — it reinvests, iterates, and pulls away. BCG [4] calls it a "winners-take-most" dynamic. The 5% generating real value are pulling away from the 60% generating none. Every quarter the gap widens.
The platform imperative
Every trend line points in one direction: from projects to platforms. The organisations generating compound AI value operate on a platform that provides governed workflow orchestration, pre-built enterprise integrations, embedded data governance, security and compliance by design, and the ability for business users to build and iterate without engineering dependency.
This isn't a feature list. It's the architectural answer to every failure mode we've discussed: pilot paralysis (the platform provides a path to production), the bolt-on trap (the platform enables workflow redesign), data unreadiness (the platform embeds data governance), disconnected teams (the platform unifies them), and integration debt (the platform handles it).
The strategic question has shifted. It's no longer "should we use AI?" Everyone is. It's no longer "which model?" They're increasingly interchangeable.
The question is: do we have the platform, governance, and organisational willingness to extract compound value from AI at scale?
The companies that answer yes will define the next decade. The companies that don't will find themselves on the wrong side of a widening gap that becomes impossible to close.
Stop buying AI tools. Start building AI operations.
The model isn't the product. The workflow is the product. The platform is the product. The integration is the product. Everything else is a very expensive demo.
This is what we built Pullstream to do. We're an AI-enabled low-code enterprise automation platform. We provide the Layer 3 that most enterprises are missing — workflow orchestration, governed integration, and the ability to embed any AI capability into real business processes. If anything in this series resonated, I'd welcome a conversation about what this looks like for your organisation.
Sources:
[1] Menlo Ventures, Dec 2025 → https://menlovc.com/perspective/2025-the-state-of-generative-ai-in-the-enterprise/
[2] McKinsey, Nov 2025 → https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai
[3] MIT NANDA, Jul 2025 → https://fortune.com/2025/08/18/mit-report-95-percent-generative-ai-pilots-at-companies-failing-cfo/
[4] BCG, Sep 2025 → https://www.bcg.com/publications/2025/are-you-generating-value-from-ai-the-widening-gap
[6] Informatica, "CDO Insights 2025" → https://www.informatica.com/resources/articles/cdo-insights-2025.html
[18] Bloomberg/MILL5, Dec 2025 → https://www.mill5.com/2026/01/12/what-enterprise-leaders-said-about-ai-in-2025/
[23] CIO Dive, Dec 2025 → https://www.ciodive.com/news/5-cio-predictions-for-ai-in-2026/807951/
[26] IBM, widely cited across industry analyses
Ready to Join the 6%? PullStream helps organisations turn AI into profit through workflow automation, integration, and governance that actually scales.