Sure, but fans also think he's like 90% kicker from that distance. He's not historically. He's a lot higher than 40% to make it, but he's a lot lower than 90% to make it also.
I'm kicking there because its 4 yards. If it were a yard or two, I guarantee John doesn't kick.
Besides, analytics is a baseline for decisions, not the be-all, end-all. If he literally took the analytics at face value, then he would be kicking a hell of a lot less than he does, because kicking league-wide sucks balls.
If analytics say that there's like a 20% deviation in expected win % for a decision, then there's an obvious course of action. If it's very close, like in this example, then it would boil down to how good your kicker is, weather conditions, game flow, etc.
I'd also point out that when the Ravens decided to NOT go for it on 4th and 1, take the delay of game, and then kick the FG, the analytics model, surprisingly, was largely indifferent on that decision.