This topic is a favorite of mine, but I'll try to keep it short ... "analytics" are used, and have been used, since the beginning of the game. All sorts of different things are "analytics." It's just that today's analytics are better than the ones they used to use.
For example, you think the 50 yard line is some magic barrier where you make one decision on one side of it, and perhaps another decision on the other side. Why? So if you're on your own 45, you'd never go for it on 4th. But if you're on their 45, you might? What if (I'm just throwing out made up numbers here) the other squad is 30% likely to score if you go for it and fail from your own 45, but 25% likely to score from their 45? And what if you're 75% likely to convert, and then, overall, 20% likely to score thereafter? We have that kind of analysis now (and much, much more). Or we can just stay in the ancient analytics world of "mid field be magic ... no go if no crossed").
And, actually, analytics can, or may in the future, "take into account ... the 300 lb All-American defensive tackles," indirectly or directly. Analytics may have looked at quality of opponent ... either overall, or in specific categories (run defense, etc.) ... and determined what the percentages are in that situation ... and maybe that percentage doesn't change enough to warrant a different decision.
But all these "traditional" decisions that have been made for years are evidence, themselves, of "analytics." You don't do X because you think the expected value of success in that circumstance is outweighed by the expected value of failure. But, modern day analytics actually looks at each time it happens or doesn't, and can give you the reality of expected outcome, rather than your guesstimate, which you've relied on for years.