How to Audit Your AWS Bill: A CTO's Checklist for Finding 25–40% Recoverable Spend
Published: 2026-05-20
Author: Saascutters
Read time: 8 minutes
Keywords: AWS cost audit, cloud spend optimization, CTO checklist, FinOps, infrastructure audit
Most engineering teams discover their AWS bill is too large only after finance flags it in a quarterly review. By then, the waste has compounded for months. The good news: the patterns are predictable, and the recoverable spend is usually twenty-five to forty percent.
This checklist is what we run on day one of every infrastructure audit. You can run it yourself today.
1. Rightsize your compute
Log into Cost Explorer and filter by EC2. Look for instances with average CPU under 10% over the last thirty days. These are almost always over-provisioned. Downgrade one tier and watch for two weeks. If performance stays flat, downgrade again.
Check your RDS metrics the same way. Eight percent average CPU on a db.r6g.2xlarge is a signal, not a success story.
2. Fix your reservations
Reserved Instances and Savings Plans expire. When they do, your bill silently reverts to on-demand pricing. Go to Billing → Reservations and note every expiration date in the next twelve months. Set a calendar reminder ninety days before each one.
If you are running on-demand instances today, you are likely overpaying by thirty to fifty percent on compute alone.
3. Audit inter-region transfer
The line item no one budgets for. Go to Cost Explorer, group by service, and look for DataTransfer. Egress between regions, NAT gateway processing, and CloudFront origin fetches add up fast. One client we worked with was spending $6,800 per month moving logs between us-east-1 and us-west-2 for a staging environment that did not need to exist.
4. Kill idle dev environments
Non-production workloads do not need to run twenty-four hours a day. Tag everything with Environment: staging or Environment: dev, then filter your bill by that tag. If the number shocks you, write a simple scheduled Lambda to stop instances at 7 PM and start them at 8 AM in your team's timezone.
5. Review your observability stack line by line
Datadog, New Relic, Sentry, Honeycomb, and Grafana Cloud are all billed differently — per host, per seat, per million spans, per GB of logs. Pull every invoice for the last three months. One vendor is usually responsible for sixty percent of the total cost, and it is rarely the one delivering the most value.
6. Check your S3 storage classes
Standard storage is the default and the most expensive. Run S3 Storage Class Analysis on every bucket over 500 GB. If objects are accessed less than once per quarter, move them to Glacier Instant Retrieval or Intelligent-Tiering. This single move has cut storage bills by forty percent for some teams.
7. Verify your CDN cache hit ratio
A low cache hit ratio means you are serving content from origin that should be served from edge. In CloudFront, look for a hit ratio under 85%. Common fixes: increase TTLs on static assets, normalize query strings, and enable compression.
When to bring in outside help
If your monthly cloud and SaaS spend is over $35,000, the audit becomes a full-time job for two to three weeks. A performance-based engineering practice can run this checklist end-to-end, execute the fixes, and verify the savings against your prior invoices — with no retainer and no upfront fee.
At Saascutters, we take thirty percent of verified first-year savings. If we do not find savings, you do not pay.
About Saascutters
Saascutters is a performance-based engineering practice that audits, re-architects, and renegotiates SaaS, cloud, and infrastructure spend. We work with Series B through enterprise teams across healthcare technology, fintech, e-commerce, and media. Request an infrastructure audit →