This statement is from the post by Adam DeJans Jr: “I’ve worked with many companies (from logistics to manufacturing) and one pattern stands out. There’s often too much emphasis on improving forecast accuracy, and too little thought about what really matters: the decisions being made and the cost of being wrong. Instead of asking how far off the forecast was, I ask how much it cost me. I want a metric that reflects business realities. If I miss high on a forecast, what does it do to my bottom line? If I miss low, what opportunities did I lose? The right loss function captures that asymmetry.

And perhaps most importantly, I never assume my forecasts are correct. Uncertainty is always part of the problem. Ignoring it doesn’t make it go away. Every planning system should explicitly account for uncertainty, and every good decision process should be built to handle it.” Link