You deal with existential risk daily. You know the cost of a failed strategy. However, the greatest barrier to true innovation is not the risk of consequence, but the fear of judgment that comes with failure. We spend too much time trying to avoid looking wrong, and not enough time trying to be right.

As a senior leader, I saw that the most effective executives didn't eliminate failure; they minimised the distance between decision and learning. This is the essence of the failure-as-data mindset.

The fear is the strategy:

If you are afraid to fail, you will unconsciously drift toward strategies that are safe, predictable, and incremental. These are often the strategies that keep the company comfortable while the market shifts rapidly around it. The cost of inaction, the lost opportunity or the missed curve, is often orders of magnitude greater than the cost of a failed pilot project. Your role is to make the right failure happen early.

How to implement failure-as-data:

1. Change the Language: Replace "Failure" with "Unexpected Data." A failed pilot is not an indictment of your team's competence; it is a critical, expensive piece of market feedback that saves you a much larger loss later. When presenting a setback to the board, frame it as a lesson learned that has now informed the next, more robust strategic pivot.

   
   

2. Model Vulnerability: Leaders who demand perfection create cultures of silence and cover-up. You must openly and quickly own your own strategic miscalculations. When you make a decision that yields unexpected data, stand up and say, "That was the best decision based on the information we had, and the data now tells us we must adjust. Here is what we learned..." This grants your team permission to experiment without fear of reprisal.

   
   

3. Define the Test, Not the Outcome: Before launching a new initiative, define the minimum necessary investment to get a reliable set of data points, and treat that investment as a sunk cost for learning. The success of the project is not the launch; it is the quality of the data and the speed of the iteration.

   
   

Your job is to accelerate the learning curve. You must be willing to make high-quality, calculated mistakes in the pursuit of high-leverage outcomes.