One way to think about it is the following: imagine you stand on a field and you have a soccer ball and you kick it. You close your eyes and you kick it and then you open your eyes and you try to predict, where did the ball fall? Imagine you do this a thousand times; after a while you know exactly the relationship between your kick and where the ball is. Those are the conditions in which intuitions are correct—when we have plenty of experience and we have unambiguous feedback.
That’s learning, right? And we’re very good at it. But imagine something else happened. Imagine you close your eyes, you kick the ball, and then somebody picked it up and moved it 50 feet to the right or to the left or any kind of other random component. Then ask yourself, how good will you be in predicting where it would land? And the answer is: terrible.
The moment I add a random component, performance goes away very quickly. And the world in which executives live in is a world with lots of random elements. Now I don’t mean random that somebody really moves the ball, but you have a random component here, which you don’t control—it’s controlled by your competitors, the weather; there’s lots of things that are outside of your consideration. And it turns out, in those worlds, people are really bad.So what is the solution? We should experiment more and test our gut feelings before we go all out and implement a pervasive solution.
This actually, I think, brings us to the most important underutilized tools for management, which [are] experiments. You say, I can use my intuition, I can use data that tells me something about what might happen, but not for sure, or I can implement something and do an experiment. I am baffled by why companies don’t do more experiments.I think the reason why many R&D executives I know do not experiment more is the lack of information - both about factors driving the decision and potential impacts of the decision. For example, executives are normally forced to rely on gut feelings to decide future R&D investments. It is difficult to experiment because R&D projects are interlinked. It is difficult to see the impact of changing one program on all the other linked programs. Funding decisions also need to satisfy a multitude of often conflicting requirements. There are no tools to quickly understand the impact of investments of staffing or on competitive position. Even when information is available, it is normally at the wrong level of detail to actually make a difference. We need tools to help executives experiment effectively in R&D management.