A typical AWS optimization program cuts spend 20 to 40 percent through rightsizing, waste elimination, storage tiering, commitment coverage, and a handful of architecture decisions. The order matters: prove the no regret wins first, then commit, then redesign only what truly drives the bill.
Where does AWS spend leak?
The leaks cluster in five places: on demand instances that should be covered by a commitment, oversized compute that never reaches its provisioned size, storage that is never tiered or has orphaned EBS volumes and old snapshots, development environments running outside working hours, and data movement charges nobody owns. None of these require a rewrite to fix. Most are visible the moment you read the bill at line item depth.
The Cost and Usage Report is your source of truth
Cost Explorer is good for a quick trend, but it rounds off the detail you need to allocate shared costs and model commitments. The Cost and Usage Report is the complete line item record. Land it in a queryable store, tag it against teams and products, and you can finally answer the question that matters: which unit of value is each dollar buying. For more on the mechanics, read the AWS Cost and Usage Report explained and reading your AWS bill line by line.
Savings Plans versus Reserved Instances
Commitments are the biggest lever and the biggest risk. The instruments discount roughly 20 to 72 percent against on demand, but every percent of discount is a percent of utilization risk you carry. Choose the structure deliberately.
| Instrument | Flexibility | Best for |
|---|---|---|
| Compute Savings Plans | Flex across family, size, region, EC2, Fargate, Lambda | Most of the base; workloads that move |
| EC2 Instance Savings Plans | Locked to a family in a region | Stable, predictable compute |
| Reserved Instances | Specific instance type; can be sold on the marketplace | Steady state, including RDS and ElastiCache |
| On demand | Full flexibility, full price | The volatile top of the demand curve |
For an enterprise spend commitment on top, the AWS Enterprise Discount Program trades a multi year commitment for a discount tier. The leverage in that negotiation comes from a credible forecast, benchmark data, timing, and the real option of placing workloads elsewhere. See cloud commitment negotiation and on demand versus commitment on AWS.
Graviton, gp3, and other standing wins
Some moves lower the rate permanently without changing behaviour. Migrating compatible workloads to Graviton processors typically cuts the compute rate while improving price for performance. Moving EBS volumes from gp2 to gp3 decouples throughput from capacity and usually reduces storage cost outright. These are standing wins: once done, they keep paying every month. The tradeoff is engineering time to validate compatibility, which is exactly the kind of work we do alongside your team.
Data transfer and NAT, the quiet eaters
The charges that surprise people are rarely compute. Cross availability zone traffic, internet egress, and per gigabyte NAT gateway processing accumulate quietly and seldom appear in a headline review. The fix is architectural: route traffic to avoid needless zone crossings, use VPC endpoints to bypass NAT for AWS service calls, and put a real owner on data movement. Read data transfer costs on AWS explained for the detail.
A worked first 90 days
The figures below are indicative and verified against the client's own billing data, anonymized. A Fortune 500 retailer running roughly 18M USD a year on AWS saw the following.
| Lever | Mechanism | Indicative annual saving |
|---|---|---|
| Commitment coverage | Compute Savings Plans plus targeted RIs to a defensible forecast | 3.1M USD |
| Rightsizing | Oversized EC2 and RDS matched to measured load | 1.4M USD |
| Graviton and gp3 | Rate moves on compatible workloads and EBS volumes | 1.1M USD |
| Waste elimination | Idle resources, orphaned EBS, old snapshots, off hours dev | 0.9M USD |
| Data transfer | Zone aware routing and VPC endpoints to cut NAT | 0.6M USD |
That is roughly 39 percent of the run rate, with no roadmap slip. Your numbers will differ; we scope against your actual usage.
Your next step
If you want the depth in a single file, download the guide below. If you want it applied to your estate, our AWS cost optimization service turns this plan into verified savings, and the cross cloud cost optimization guide covers the same discipline across Azure, GCP, and OCI.
Get the AWS cost optimization guide as a PDF
The full guide with the commitment decision tree, the waste checklist, and the first 90 day plan. Independent and buyer side, with zero provider commissions.
AWS cost optimization, answered
Frequently asked questions
What is the single biggest lever on an AWS bill?
Commitment coverage. Savings Plans and Reserved Instances discount roughly 20 to 72 percent against on demand in exchange for utilization risk you carry. Buy too little and you overpay; buy too much against a soft forecast and you pay for capacity you never use. The win is coverage set to a defensible forecast, not the deepest discount.
Savings Plans or Reserved Instances: which should we buy?
It is a tradeoff between flexibility and specificity. Compute Savings Plans flex across instance family, size, region, and even between EC2, Fargate, and Lambda, so they survive change. Reserved Instances and EC2 Instance Savings Plans lock to a narrower scope for a slightly deeper discount. Most estates want a base of flexible coverage with targeted specific commitments on stable workloads.
Why is the Cost and Usage Report important?
The Cost and Usage Report is the only complete, line item record of what you spent and why. Cost Explorer is fine for trends but rounds off the detail you need to allocate shared costs, find waste, and model commitments. Serious AWS cost work starts from the Cost and Usage Report, not the console summary.
Does AWS Compute Optimizer just tell us what to do?
It recommends, it does not decide. Compute Optimizer is a useful signal for rightsizing, but it cannot see your roadmap, your reliability targets, or your commitment position. We treat its output as input, then make the call with your engineers. Trust, but verify.
What are the quiet AWS budget eaters?
Data transfer and NAT gateways. Cross availability zone traffic, internet egress, and per gigabyte NAT processing charges accumulate quietly and rarely show up in a headline review. Graviton and gp3 migrations, by contrast, are standing wins that lower the rate without touching behaviour.
Put a defensible number on your cloud spend.
No provider in the room, no published price list. Tell us your footprint and we will scope the savings against your billing data — we reduce your cloud spend or we reimburse our service fee.
The Cloud Spend Navigator: what changed in cloud pricing, commitments, and FinOps — no vendor spin.