As dns0 has threat intel partners, I think even if NextDNS is tweaked, it would fall behind. Maybe you could perform the same test using NextDNS and ControlD? As far as I know both nextDNS and ControlD actually directly import from urlhause and phishtank. So the result should be very similar for both. Also, I tested from 2 sources which both of them definitely cannot import from directly like that, as that's impossible. I used Alienvault and IBM X-Force to get some malicious links manually and tested them with both of them. They both were very close in both cases. When used links from IBM, NextDNS did a little better and When used AlienVault, ControlD did a little better. It should also be taken into consideration that ControlD doesn't have AI yet(coming this month as mentioned in their blog) and NextDNS does. So, basically I am currently sticking with ControlD as they have customer support. They listen to their customers , their requests and info provided by them and take actions very quickly if needed. And their ad filter is maintained by them with aggressive blocking and false positives removed.