Availability design begins with a failure you intend to survive and a recovery target. Separate Availability Zones protect against a zone-level failure, while another Region addresses a different set of risks and adds operational complexity. A backup supports restoration; it doesn't automatically meet the same recovery time as a running standby.
Follow the complete request path. Two web tiers still fail together if they depend on one unprotected database, a single-zone outbound path, or a manually maintained secret that expires. State the recovery time objective and recovery point objective before choosing a replication or backup design. Then test failover, recovery, and the return to normal operation. The difficult case is a diagram that looks redundant but shares a critical dependency.