I agree with your last post. When the parameters in play are too "squishy" to make a rigorous model, engineering falls back on rules of thumb. And those are based on something. Similarly, when a shop guy or other expert recommends a ski, they are making their recommendation based on something, even if they are unable to sort it out clearly enough in their own heads to write the rules down.
And, when you have a rule-of-thumb based choice, you have to build in time for experiment, rework, and retry (which is where demo-ing comes in, of course).