An AWS cost optimization playbook for FinOps Leads who need durable savings, not one-time wins — rightsizing every fleet, sizing commitments to the steady-state workload, and the storage discipline that stops the bill from coming back.
The math is not working.
AWS spend grows quietly. Instances get oversized in incident response and never get resized. RDS classes step up at upgrade time and never step down. ElastiCache nodes get provisioned for a launch and outlive the workload they were sized for.
Most AWS optimization programs are too narrow. Compute alone leaves the storage tail. Storage alone leaves the commitment gap. Commitments alone leave the rightsizing on the floor. The 38% savings landed because the program ran all three together.
Every EC2 instance, RDS instance, ElastiCache node, and OpenSearch domain gets a rightsizing recommendation. Most production fleets we audit have 20 to 35 percent of instances oversized, and the savings concentrate in the fleets nobody has looked at in two years.
Reserved instances and Savings Plans are leaving money on the table at most real estate SaaS companies. We size commitments to the steady-state workload, leave the variable portion on on-demand, and treat unused commitments as a metric the team is accountable for.
S3 lifecycle policies. EBS volume right-typing. Snapshot retention rules. The storage tail is rarely the headline, but it compounds — and it is the part of the bill that grows even when usage does not.
Every EC2 instance, RDS instance, ElastiCache node, and OpenSearch domain gets a rightsizing recommendation. Most production fleets we audit have 20 to 35 percent of instances oversized.
Reserved instances and Savings Plans are leaving money on the table at most real estate SaaS companies. We size commitments to the steady-state workload, leave the variable portion on on-demand, and treat unused commitments as a metric.
S3 lifecycle policies. EBS volume right-typing.
Tagging contract, monthly review cadence, owner-assigned for new spend over a threshold. The discipline that stops the bill from coming back.
If your AWS bill is growing faster than your customer base, the gap is FinOps discipline.
Not when shipped behind a feature flag with rollback. We have not had a customer-visible breach in our last 9 FinOps programs.
We use them. New commitments augment, not replace, existing commitments unless they are mis-sized.
Yes. Rightsizing rolls behind feature flags. Commitments are a finance operation. Storage policies are background changes. The team shipping product does not stop.
We work jointly with the team that owns the metric. Anything actively used stays. Anything unused for 30 days gets archived rather than deleted, with a path to restore.
Yes, when the operating model lands with the program. Tagging contract, monthly review, ownership on new spend — without those, savings erode in two quarters.
