You have probably seen the number. Around 95% of enterprise generative-AI pilots deliver little or no measurable impact on the bottom line (MIT NANDA, The GenAI Divide, August 2025). It gets quoted to sell urgency. Here it earns its place for the opposite reason: to remove the panic and replace it with a diagnosis.
Because the failures don’t cluster where you would expect. They are not mostly about weak models, missing data, or insufficient budget. McKinsey’s State of AI (November 2025) found that only about 39% of organizations could attribute any EBIT impact to their AI work — and the organizations that could were not the ones who spent the most. They were the ones who ran AI with discipline.
What “discipline” actually means here
It is unglamorous. It is tying a use case to one number with one named owner. It is writing the data rules down on a single page. It is keeping a human check on anything a customer sees, and being willing to switch off what underperforms. None of that requires an enterprise budget. Most of it requires a decision.
That is the uncomfortable, useful finding: the gap between the 5% and the 95% is a leadership-and-discipline gap, not a technology gap. And a leadership gap is exactly the kind of gap a 30-person company can close faster than a 30,000-person one.
Why this is good news if you’re small
A large enterprise closing this gap has to move budget through approval layers, retrofit governance onto legacy systems, and re-train thousands of people. You have to convince yourself and brief a team. Your lack of legacy infrastructure and your short chain of command are advantages here, not consolation prizes.
Most of the writing on this site turns that claim into something you can act on: the six questions that separate aligned AI adoption from expensive activity, and how to score yourself honestly against them.
The companies that succeed with AI don’t have bigger budgets. They have better discipline. The rest of this work is about borrowing theirs.