We've run Well-Architected cost reviews across more than thirty enterprise AWS accounts, and the same handful of patterns account for the majority of avoidable spend every single time.
Oversized RDS instances top the list — provisioned for a peak load that either never materialized or happened once, eighteen months ago. Right-sizing based on actual CloudWatch metrics routinely recovers 20-30% of database spend alone.
Idle and orphaned resources are next: load balancers pointing at decommissioned services, unattached EBS volumes, and NAT gateways serving traffic from environments nobody uses anymore. None of these show up unless someone is actively looking.
The highest-leverage fix isn't a tool, it's a process: monthly cost review ownership assigned to someone on the engineering team, not finance. Cost problems compound quietly for a reason — nobody with context is watching.