How thinking scientifically drives better business decisions

An experimentation culture can help executives make better decisions and harness AI more effectively, Harvard Business School professor Stefan Thomke says

November 20, 2026
A man sits at a desk in an office, looking thoughtful while holding a pen in his mouth. A laptop is open in front of him. A fan is also visible nearby.

Business leaders face growing pressure to make decisions quickly, particularly as artificial intelligence accelerates the pace of change. Harvard Business School professor Stefan Thomke, author of "The Scientific Mindset," recently spoke at a Leadership Reimagined discussion on November 11 about how executives can use scientific thinking to drive better business decisions.

“Test more, assume less”

The human brain has a bias for believing false positives, and the increasing pressure to act quickly in the AI age may exacerbate this tendency, Thomke said. Hence his maxim, and the five planks of his scientific mindset. CEOs and senior executives should be knowledgeable skeptics and question the decisions they make and don’t make. They should articulate testable hypotheses and then run experiments to produce solid evidence. Probe cause and effect to avoid confusing correlation with causation. And, finally, build an experimentation culture that empowers people to question and experiment. 

Humility trumps arrogance every time

Beware the temptation to start with a conclusion and fit the test to “prove” your assumption or, even worse, not bother to test it at all. “Information is most valuable when you least expect an event,” Thomke said. He cited Ron Johnson, who built the Apple Store into the leading US retailer in terms of sales per square foot, but whose bold plan at mid-tier department store JCPenney led to plunging sales and his departure within 17 months. Johnson has since acknowledged what Thomke calls “situational arrogance” – assuming that success in one setting guarantees you know what’s right in another. “Humility is really at the heart of the scientific method,” Thomke said. That means being willing to go into an executive meeting and say, “I don’t know.”

Learn to listen and ask good questions to get to the essence of the business

Thomke said the art of listening is a big differentiator among leaders, and it’s consistent with humility. A leader who doesn’t think he or she has all the right answers should be eager to listen carefully to employees, suppliers, and customers, and ask questions that get to the heart of a problem or opportunity as well as a company’s essence: what it does that’s different from everybody else. He cited the example of Jørgen Vig Knudstorp, who led a turnaround of toymaker LEGO in the 2000s. The company had been increasing the number of products and components in its portfolio, believing that would drive further growth.

He questioned that premise with managers, dug into the data, and found that complexity was a problem, not a growth driver. By slashing complexity in half, he sparked a turnaround that has become a classic case study.

AI can supercharge the scientific mindset, not undermine it

The key lies in how you use it. Leaders can instruct their AI to be a knowledgeable skeptic and prompt it to examine operations critically, come up with testable hypotheses, and create tests that can show the right way forward. Thomke cited how Amazon uses generative AI to update its catalog, automatically check the reliability of AI output, and test product page changes on customers with actual sales data. It’s a scalable solution for a company that sells some 600 million products, but boards remain accountable for results and leaders need to bring the humility, not AI.