I don't know if I agree with you, Borivoj. I guess it depends on one's philosophy of education.
You state, analytics "could determine the methodology of teaching too much". I am not completely convinced we shouldn't provide some structures and frameworks versus allowing a more loose approach to teaching - however, even the most liberal teachers learns and uses methods on some manner- no tabula rasa there. Research has been done on how kids learn, the curriculum has been developed, and modern versions of teaching are unfolding (albeit, slowly). There is a base to work from (and improve).
If we don't use our findings from applying analytics, what will we use them for if not to inform our practice?