Culture Is the Constraint

McKinsey just released data from 10,000 executives. Eighty-eight percent of organizations are deploying AI. Eighty-one percent see no bottom-line impact. The variable that predicts success more than any technical readiness metric, by nearly a factor of two: culture.

I've been in the room where the pilot works and nothing changes afterward. The tools work. The model works. The integrations are solid. And then it just sits there, collecting positive feedback on a Confluence page nobody reads after week three. What I've learned, across too many of those conversations, is that you cannot engineer your way past a culture that does not actually believe the change is real. The invisible infrastructure that determines whether AI value leaks out or compounds is not the tech stack. It's how decisions get made, who is watching, and what those people see the leaders above them actually doing every day.

The McKinsey finding that sticks with me: only 14% of respondents say their senior leaders consistently champion adoption. Not 14% of organizations that tried. Fourteen percent of the ones already deploying. That gap is not a training problem or a change management program problem. It is a question of whether the people with the most visibility in your organization have publicly committed to operating differently and then followed through. At Improving, we've found that AI adoption is directly correlated with managers' use of it. Culture follows behavior, not announcements. Your team is watching what you do, and talking about it normalized the experimentation and learning that it takes to become proficient.

The fix is not a comms plan. It is not another all-hands where the CEO says AI is a priority. The organizations landing in the 19% that see real bottom-line impact are the ones where senior leaders changed how they actually run their meetings, what they measure, and how they make decisions. Those behaviors compound. The declaration without the behavior just adds another layer of skepticism to the pile.

Who in your organization is visibly operating differently because of AI, not just talking about it?