One of the chief problems in cybersecurity is the inherent reactivity of most forms of defense. An attack has to be observed, analyzed and reverse-engineered. THEN, protection can be developed. This means attackers are successful, and inside environments, for a period of time before the attack is noticed, before the indicators for that attack can be extracted, and before a policy can be disseminated to stop it.
There has been a wide variety of research in recent years around this problem. How to speed up the cycle to recognize attacks and to potentially get out in front of attackers to block them before the attacks start. Both my own PhD research and other researchers have noticed that one attribute that is overwhelmingly an indicator of maliciousness in DNS is “newness,” that is to say, the newer a domain is, the more likely that it is bad. More importantly, when a domain is new and otherwise benign, it is rarely in meaningful use except by the organization that’s setting up whatever will go there.