Designing for High Availability: Active-Active vs. Active-Passive Architectures?

Hey everyone, I'm working on a new service that needs extremely high availability. I'm weighing the pros and cons of Active-Active versus Active-Passive architectures. Active-Active seems appealing for better resource utilization and simpler failover detection, but the complexity of data consistency across multiple active sites scares me a bit. On the other hand, Active-Passive is simpler to implement initially, but failover times can be longer and it leaves resources idle. What factors do you consider most critical when making this decision, especially for services with strict RTO/RPO requirements? Are there specific use cases where one clearly outweighs the other?