A proper AWS cost optimization assessment is not a tool you run, it is a sequence you follow. The order matters more than any single finding, because optimizing in the wrong order locks in waste: commit before you rightsize and you pay for the oversized footprint for years. This is the assessment we run across 500+ environments to produce a 31% average reduction, laid out so you can run it yourself or know what to expect when we run it for you.
This article links up to our complete guide to AWS cost optimization, the pillar for this cluster, and the broader, cloud-agnostic version of this work lives in how to audit a cloud environment for waste. The assessment maps directly to our See, Cut, Lock, Run method.
See before you Cut, Cut before you commit. Get clean cost visibility, then rightsize and remove waste, then buy commitments on the clean baseline. Reservations are always last.
Step 1 · See: get the data and make it legible
Start from the Cost and Usage Report, the most granular record of what AWS charged, broken down by service, account, usage type and resource. Pull at least the trailing three months, ideally a year for seasonality, and normalize it so every dollar has an owner. Layer in tag coverage, since untagged spend cannot be allocated or held accountable, and pull utilization data from CloudWatch and Compute Optimizer. We cover the underlying data sources in our RDS optimization walkthrough and the tooling in our visibility cluster. The output of this step is a clean picture of where the money goes and who owns it.
Step 2 · Find the waste, sized and ranked
With visibility in place, quantify each category of waste so you can rank by saving and effort. Work through idle and zombie resources first because they are pure waste with no performance trade-off, then over-provisioned compute and databases, then storage on the wrong tier, then data transfer, then commitment gaps. The point is to size each opportunity in dollars, not just flag it, so the program targets the largest levers first instead of the easiest.
| Category | What to look for |
|---|---|
| Idle and zombie | Stopped instances still billing, unattached EBS volumes, idle load balancers, orphaned IPs |
| Over-provisioned compute | EC2, RDS and EKS nodes running well below their size against p95 utilization |
| Storage tiering | S3 in the wrong storage class, old snapshots, oversized gp2 volumes |
| Data transfer | Cross-AZ, cross-Region and NAT Gateway traffic that could be reduced or rerouted |
| Commitment gaps | On-Demand spend that should be on Savings Plans or Reserved Instances once rightsized |
Want this assessment run for you, with the saving quantified first?
Our AWS cost audit runs exactly this sequence across your estate, sizes every lever in dollars, and models the saving before you touch production. On the performance model, you pay only from realized savings. No savings, no fee.
Book an AWS cost audit →Step 3 · Cut: rightsize and clear waste before committing
Execute the cuts in the order the assessment ranked them, lowest risk first. Delete the idle and zombie resources, then rightsize compute and databases against real utilization, then fix storage tiering, then address data transfer. Only when the environment is the right size do you move to commitments. This sequencing is the single most valuable output of the assessment, and the reason an assessment beats running optimization tools in isolation.
Step 4 · Commit on the clean baseline
Now the footprint is honest, model the commitment mix. Cover the steady core with Compute Savings Plans for flexibility and Reserved Instances or reserved capacity where the workload is fixed, and ladder the terms so you are not re-rating everything at once. The buying logic is in Savings Plans versus Reserved Instances. For large estates, the assessment also informs whether an Enterprise Discount Program is worth pursuing, which we cover separately.
Step 5 · Lock and Run: make the saving stick
An assessment that is not locked decays. Set budgets and anomaly detection so spend cannot drift back, put guardrails on the provisioning that created the waste in the first place, and assign owners so each line has accountability. Then review on a cadence, since utilization and pricing both move. This is the difference between a one-time cut and a unit cost that keeps falling.
AWS services, tools and reporting referenced above reflect offerings as of May 2026. Verify current features and pricing in the AWS console as part of any live assessment, since they change.
The AWS Cost Optimization Field Guide includes the full assessment worksheet, the data-pull checklist and the savings model we use on engagements. It is the downloadable companion to this guide.
The short version
To run an AWS cost optimization assessment: See first by pulling the Cost and Usage Report, utilization and tag coverage into one clean picture; size and rank every category of waste; Cut in order, lowest risk first, rightsizing before you commit; buy commitments on the clean baseline; then Lock and Run so the saving holds. When you want the assessment run for you with the number proven before any change, that is what our AWS cost optimization service delivers, with the option to pay only from realized savings.