How to Cut Your AWS Bill: 9 High-Impact Savings

Most AWS bills carry 20–40% in avoidable waste. Not because teams are careless, but because cloud spend grows in a thousand small decisions that no one circles back to review. Here are the highest-impact places to look, roughly in order of effort-to-savings — the same checklist I run in a cost audit.

The 80/20: idle resources, the right purchase commitments (Savings Plans), storage tiering, and rightsizing typically account for the bulk of recoverable spend. Start there before micro-optimizing.

1. Delete idle and orphaned resources

The fastest money you'll ever save is on things you're paying for and not using:

Unattached EBS volumes — they bill whether or not a server uses them.
Unassociated Elastic IPs — AWS now charges for all public IPv4 addresses, idle or not.
Old snapshots — backups accumulate silently; add a lifecycle policy.
Stopped instances — no compute charge, but their EBS and any EIPs keep billing.
Idle load balancers and NAT gateways — both bill per hour even at zero traffic.

2. Right-size over-provisioned compute

Teams routinely run instances 2–4x larger than the workload needs "to be safe." Check CloudWatch for instances sitting under ~20% CPU and memory, and step them down a size (or two). AWS Compute Optimizer will surface candidates automatically. Don't forget RDS — databases are some of the most over-provisioned resources in any account.

3. Buy the right commitment (Savings Plans)

If you have steady-state compute running 24/7 on on-demand pricing, you're leaving 20–30%+ on the table. A one-year Compute Savings Plan discounts that baseline while staying flexible across instance families and regions. Commit to your floor of usage, not your peak. (See our guide on Savings Plans vs Reserved Instances.)

4. Migrate gp2 volumes to gp3

gp3 storage is about 20% cheaper than gp2 and comes with better baseline performance — and the migration is an in-place modification with no downtime. If you still have gp2 volumes, this is nearly free money. (Full walkthrough: gp2 vs gp3.)

5. Tier your S3 storage

Most S3 buckets hold data that's rarely touched after the first month. Turn on S3 Intelligent-Tiering for unpredictable access patterns, and add lifecycle rules to move cold objects to Infrequent Access or Glacier. On large buckets this routinely cuts storage cost 30–60%.

6. Tame data-transfer and NAT costs

Data transfer is the silent line item. Cross-AZ traffic, NAT gateway processing charges, and egress to the internet add up fast. Add VPC gateway endpoints for S3 and DynamoDB (they're free and bypass NAT), keep chatty services in the same AZ, and put a CDN in front of internet-facing traffic.

7. Modernize to Graviton

AWS's ARM-based Graviton instances typically offer ~20% better price-performance than equivalent x86. Many workloads — and managed services like RDS, ElastiCache, and Lambda — support Graviton with little or no code change.

8. Schedule non-production environments

Dev, test, and staging rarely need to run nights and weekends. Auto-stopping them outside business hours can cut their cost by ~65–70% — they're idle two-thirds of the week.

9. Catch it early with budgets and anomaly detection

Turn on AWS Budgets alerts and Cost Anomaly Detection so a runaway resource or a misconfigured job pings you in days, not at the end of the month. Tag resources by team/project so you can actually attribute spend.

Do this quarterly. Cloud cost isn't a one-time cleanup — it's a habit. New waste creeps in with every deploy. A recurring review keeps the savings from eroding.

How much could you recover?

The honest answer depends on your account, but if you've never run a structured review, expect meaningful double-digit percentages. Our free AWS Cost Checkup reads an exported cost file and gives you an itemized estimate in minutes — no account access required.

Pricing and features referenced here change; always confirm current details in the AWS console for your account.