It's still an inexact science and it is skewed towards population bases, but it is based on actual on field performance as well as some projection of what may happen in the future.
It's just a balance of school offers, production, and eye-ball tests. If Rivals/Scout evaluators were that amazing, they'd be working for schools and earning much more money. Their evaluations are weighted heavily on the schools offering. Then they tweak them based on camp production, in-season production, and word of mouth.
As a whole, they are not terribly accurate predictions of future talent. Way too much development happens between 18 and 22 for a football player, not to mention off the field distractions.
Another thing with offers, schools will also pounce all over someone with an impressive offer. In baseball, I've seen a top 10 out of state school offer a kid based on camp performance. It was his first major offer, and then all the offers came flooding in.
Football example of that: Frankie Hammond Jr. Was rated low, no one had heard of him. Miami offered. Then Florida threw an offer. BOOM, 4 star. He was a huge bust, btw.
It's a very inexact science, and certainly not worth the $100 / year or whatever they charge, at least to me.