The gap nobody's measuring
22% vs 67%. The difference between AI that works and AI that doesn't comes down to expertise, not ambition.
Most organisations can tell you how many AI tools they've trialled. Far fewer can tell you whether any of them actually worked.
A 45-point gap
MIT's 2025 research put numbers on it: 22% of in-house-only AI implementations reach measurable success, against 67% when expertise leads. (Source: MIT NANDA, "The GenAI Divide: State of AI in Business 2025." Global data, implementation success rates.)
That gap isn't about who has the smartest people. It's about whether someone is measuring adoption properly and adjusting course, instead of declaring victory at the pilot and moving on.
What to measure
Start with the boring question: is the tool actually used, by whom, for what, and is it better than what it replaced? If you can't answer that, you don't have an AI strategy, you have AI spending.
Want to know where your team actually stands?