So, when does eventual consistency actually make sense?

Hey everyone, I've been pondering consistency models for distributed systems quite a bit lately. We usually go for strong consistency, right? But that can seriously mess with availability and how fast things respond. So, I'm really wondering about those times when eventual consistency isn't just okay, but actually the better way to go. Have you guys seen or used situations where eventual consistency worked out great? What were the main hurdles or upsides you noticed?