Skip to content

Are you overpaying for AWS? 8 questions, an instant result.

Eight questions about the usual sources of waste on an AWS bill. You see the result right here, without giving an e-mail address. It can’t tell you how much you overpay, but it shows where to start.

The questions

Without JavaScript I can’t add it up for you. Total the points next to your answers and find your range.

  1. Do you have Savings Plans or Reserved Instances?

    A steady baseline paid at on-demand rates costs more than the same usage under a one-year commitment.

  2. Do your test environments run at night and at weekends?

    Staging and dev environments are used during working hours, but you pay for every hour they run.

  3. Does traffic from private subnets go out through a NAT Gateway?

    NAT Gateway charges for every gigabyte it processes, including traffic to S3 in the same region when there is no VPC endpoint.

  4. What type are your EBS volumes?

    gp3 costs less per gigabyte than gp2, and you can switch without stopping the instance.

  5. Does anyone clean up old snapshots, spare IP addresses and idle load balancers?

    You pay for them every month even when nothing uses them. Since 2024, AWS also charges for every public IPv4 address.

  6. How hard does your RDS database work?

    Databases are often sized with headroom at launch, and nobody checks later whether it is still needed.

  7. How long do you keep logs in CloudWatch?

    By default, log groups never expire, so you pay to store everything since day one.

  8. Do you have budgets with alerts, and tags on your resources?

    Without a budget, a rise shows up only on the invoice; without tags, nobody knows which environment or project caused it.

How to read the resultYour result and what next

Each answer is worth 0, 1 or 2 points. The result shows where to look, not how much you overpay.

  1. 0–4 points

    Your result

    The bill looks well kept

    Most of the usual leaks are closed. Waste can still hide in how services are sized, or in running on AWS at all, but general questions can’t show that.

    • Once a quarter, compare the 10 largest items in Cost Explorer with the previous quarter.
    • See in the calculator what the same setup would cost at Hetzner or OVH.

  2. 5–9 points

    Your result

    A few of the usual leaks

    Some items on the bill can usually be reduced without touching application code. What that means in money depends on your amounts, which a questionnaire can’t know.

    • Start with the signals that don’t touch the application, such as log retention or moving volumes to gp3.
    • Load a Cost Explorer export into the CSV analysis to see which items are largest. The file never leaves your browser.
    • Set a budget with an alert, so the next rise shows up the same week rather than on the invoice.

  3. 10–16 points

    Your result

    Plenty of signs of overpaying

    Nearly every question pointed to a place where the bill may be higher than your traffic needs. What that means in money only shows up in a line-by-line review of the account.

    • Before changing anything, set a budget with an alert and tag your environments, so you can see the effect of each change.
    • Start with changes that carry no risk for production: stopping test environments out of hours and setting log retention.
    • Order an AWS cost audit: you get a list of changes with a saving figure next to each.

What to do about each signalYour signals and what to do about them

  • Savings PlansTake your lowest usage of the last few months and cover that part with a Compute Savings Plan; use Reserved Instances for RDS.
  • Test environmentsStop them at night and at weekends on a schedule, with EventBridge Scheduler for example. Ten hours on weekdays is about 30% of the week.
  • NAT GatewayAdd gateway VPC endpoints for S3 and DynamoDB (they are free) and check the NatGateway-Bytes line in Cost Explorer.
  • gp2 volumesChange the volume type from gp2 to gp3. For large volumes, check first how many IOPS you actually need.
  • Snapshots and idle resourcesSet snapshot retention in Data Lifecycle Manager or AWS Backup, release IP addresses with no instance, and delete load balancers with no targets.
  • RDS sizeCompare last month’s peak CPU and memory with the instance class. If the headroom is large, consider a smaller class or Graviton.
  • CloudWatch logsSet retention on every log group, for example 30 or 90 days. Move logs you need for longer to S3.
  • Budgets and tagsCreate a budget in AWS Budgets with an e-mail alert, and tag resources by environment and project.

Want to know what it adds up to?

The questions show where to look. A figure next to each change comes only from an audit of the account, and the calculator shows what the same setup would cost outside AWS.

See the AWS cost auditOpen the calculator

The audit refund guarantee applies to AWS bills from USD 800 net a month. Terms on the pricing page.