How would you explain CAP theorem in a system design interview?

Hey everyone, I'm trying to get better at explaining CAP theorem in system design interviews, especially for data engineering roles. I feel like I understand the concepts, but articulating them clearly, with good examples, is a challenge. How do you usually break it down in a 45-minute interview? Specifically, I'm curious about real-world scenarios where you'd deliberately choose CP over AP or vice-versa. For instance, in a system like Consul or Zookeeper, CP is critical, but what about something like a distributed cache or a social media feed? What kind of depth do interviewers typically expect? Should I be ready to discuss Paxos or Raft briefly, or is it more about the high-level trade-offs and implications for system behavior? Any frameworks or examples you use that really hit home?