Look at some of the most disproportionately overrated recruits on the on3 website. Guys who have a MUCH higher rating with on3 than the consensus of the other sites (meaning the on3 is not only higher, but it also causes the consensus to increase, and the consensus would be lower if not for on3's overly high rating):
on3
11. Jven Williams, PED STATE commit (65 consensus)
20. Adepoju Adebawore, heavy OKLAHOMA lean (72 consensus)
28. Markis Deal, heavy TEXAS lean (88 consensus)
29. Darron Reed, LSU commit (170 consensus)
24. Joshua Mickens, LSU commit (118 consensus)
39. Daniel Harris, GEORGIA commit (139 consensus)
48. Eugene Wilson, FLORIDA commit (147 consensus)
49. TJ Searcy, FLORIDA commit (153 consensus)
50. Aidan Mizell, heavy FLORIDA lean (106 consensus)
54. Derion Gullette, heavy TEXAS lean (122 consensus)
55. Tyler Williams, heavy GEORGIA lean (127 consensus)
58. Dashawn Womack, LSU commit (112 consensus)
60. Yhonzae Pierre, ALABAMA commit (165 consensus)
64. Roderick Kearney, F$U commit (116 consensus)
70. S'Maje Burrell, TEXAS commit (232 consensus)
76. Kayin Lee, OHIO TAINT commit (166 consensus)
And don't even get me started on Nathan Efobi, the #84 on3 player who is #560 in the composite
So I just pulled out 16 guys from the Top 76 players who had on3 "over-ratings" of 50 spots or more when compared to the COMPOSITE of on3, Rivals, 247, and ESPN. This means the differential would be even greater if you pulled on3 OUT OF the consensus.
That means roughly 20% of the Top 75 recruits (per on3) are GREATLY overrated compared to the other three services. And these are "Top 75" type kids, usually any and all ratings services can recognize the best of the best. I'm not expecting the rankings to be identical, but you would at least not expect to see wild discrepancies for the best-of-the-best players.
And look where these kids are going. Lots of Texas and Florida and LSU in there. And, sure, one could argue "yeah, but those schools always get good players", which is not completely true (certainly not that true of Florida). But what we DO know is that the SEC fanbases are loudmouths who demand ratings bumps for all their recruits. Florida fans in particular tend to harass the website writers to give their recruits bumps, and act as if their subscriber base justifies better coverage for "their" recruits. And we also know that as on3 got off the ground, most of the staffed websites and writers were in SEC and Big 10 markets.
I hope that on3 gets better in its evaluations soon, because if on3 is going to be a part of the "consensus" rankings, then they need to try harder not to be a statistical outlier with so many recruits.