Perhaps it is my ignorance about the data-driven aspects of Deming's work, but it does not seem particularly paradoxical to me that one might achieve success in business by monitoring and measuring the heck out of things and then reacting accordingly. The trick, I suspect, is in knowing what "reacting accordingly" truly looks like for a given set of numbers. In particular, it sounds like he tried to be highly mindful of what was actually human controlled vs. what was due to a system. Deming's red bead demonstration [0] is an excellent example of how managers conflate the two; in it, he has a group of employees who are required to rigorously follow a strict procedure that is statistically going to have a particular rate of defects, and then proceeds to judge the employees by the defects that occurred during their execution of the procedure. That's an example of how you can do a great job of capturing the numbers and still fail to react accordingly. If your employees are being operationally rigorous and your defect rate is unacceptably high, no amount of reward or punishment to your employees will improve your defect rate (at least, not without inducing them to corrupt their operational rigor, which is presumed to be a defect in its own right). Rather, it is your operations themselves--the system--that must be adjusted!

It seems to me somewhat analogous to what we are seeing with Agile software development and the methodologies and metrics that have sprung up around it; the fundamental ideas seem sound, but their interpretation and implementation (the system put in place by which developers are expected to "do Agile") in many organizations has not managed to retain enough of the original spirit to convey the corresponding benefits.

[0] https://deming.org/lessons-from-the-red-bead-experiment-with...