Five ways we cut AWS bills without cutting corners

Almost every AWS account we audit is paying for capacity it doesn't use. The good news: cost optimization rarely means degrading the product. It usually means matching what you pay for to what you actually run. Here's where we start.

1. Right-size compute

Instances get provisioned for peak-day guesses and never revisited. We look at real CloudWatch utilization and move workloads to the right instance family and size — often Graviton for a straight price-performance win.

2. Buy the commitment you already have

If a baseline of compute runs 24/7, on-demand pricing is a tax. Savings Plans and Reserved Instances cut that baseline 30–70% with no code change — the trick is committing only to the steady-state, not the spikes.

3. Tier and expire storage

• Move cold S3 data to Infrequent Access or Glacier with lifecycle rules.
• Delete unattached EBS volumes and old snapshots.
• Turn on S3 storage-class analysis so tiering is data-driven.

4. Kill idle and orphaned resources

Dev environments running overnight, load balancers with no targets, NAT gateways in empty subnets. Individually small, collectively a real line item. We tag everything and schedule non-prod to shut down off-hours.

5. Make cost visible

You can't manage what you can't see. We set up cost allocation tags, budgets, and anomaly alerts so the next surprise bill becomes a Slack message before it becomes a month-end shock.

A typical Cloud Assessment surfaces these in a week and pays for itself in the first month or two of savings.