I looked at which players were eventually drafted.... and it tracks really well with a power series where the likelihood of being an NFL drafted player goes exponentially higher as the player rating increases.
Granted, I went with 247 composite and therefore binned them into bins which spanned 1.99 ratings point, but the relationship was REALLY strong and provides a pretty good estimate for each recruit rating based on the trendline. R^2 of 0.9699
So, yeah the difference does get larger as ratings get higher.... but it's the same level of exponential growth at every level.
https://www.cougarboard.com/board/message.html?id=33199569
As you mentioned, there's variance among the individual players and a real chance that a specific 85 could end up better than a specific 87.... but in large numbers the 87s are statistically and notably going to be better