Too much talky talky sales pitch words, and not enough emphasis on how machine learning has completely changed the game with DOE. You'll sell a million Minitabs if you can make people understand how much better machine learning is. 

Traditional DOE requires the parameter space to be probed very rigidly. It's a big pain in the butt. For looking at a formulary space, you have to make the formulas that Minitab (or JMP or whatever) tells you to, and you have to make them right. You have to do all of your work up front, and then you get some answers that hopefully are useful because you performed the correct experiments so faithfully, and you chose the right region of the parameter space to explore. I'd just as soon stay in bed. It's often a lot faster to do one-dimensional ladder studies to probe the parameter space quickly, and then use intuition and experience to "Trial and Error" your way towards some optimum; navigate the interactions and complexities as they come, instead of performing a million experiments to precisely describe them. 

For formulation work, machine learning changes everything. You can take the inputs and outputs for whatever samples you happened to make, and if those samples cover a decent span of the formulary space, you can get a model that maps the correlations between inputs and outputs very well. And if you have doubts about the model, you can do more experiments to better cover the parameter space. Some softwares will even recommend which further experiments to perform to most efficiently fill in the parameter space. 

DOE is so inflexible and so involved; there are almost always faster, albeit less scientifically rigourous, paths to the same good insights and decisions. Machine learning makes the process flexible enough to actually make it worth doing at all. You should lead with that, and the people will be eating up Minitabs left and right.