Just curious, how do they detect the fact that the connection is from a VPN? Are they subscribing to IP blacklists, or is there some other method?
Basically if the IP is from a datacenter - it's likely a VPN.
And for sites with a lot of traffic, you probably don't even need an external subscription, you can deduce the ranges yourself simply by checking for IP addresses that wind up having logged-in users from all over the world in a short time frame.
Pretty easy to do accurately with a short set of heuristics.
In a previous job at an online gambling company we found that a large percentage of fraudulent user accounts came from VPNs so ended up banning them at the geo lookup stage.