Agreed that writing test coverage and running twin systems against would get you this. But the benefit is that when you have no or low test coverage, if you hav

Agreed that writing test coverage and running twin systems against would get you this. But the benefit is that when you have no or low test coverage, if you have a drop in way of just testing behavior matching then it's probably a more economical approach in many projects. Also, I doubt any test suite exists with the variance of inputs that live user testing generates, including error cases.