Back to home
All guides
☁️
Amazon Web Services · FinOps Field Guide

AWS Cost Optimization: The Complete Guide for FinOps Teams (2026)

Amazon Web Services is the largest cloud platform in the world — and the source of more than half of the cloud waste we audit. This guide gives you the 14 proven AWS cost optimization patterns we use in real customer audits, ranked by ROI.

32%
Average waste in AWS bills
60% of waste
Top 3 patterns
21 days
Median time-to-savings

The 14 patterns

01

Right-size EC2 instances

High impactMedium effort15-30%

CloudWatch CPU/memory at p95 for 14 days. If <40%, downsize. AWS Compute Optimizer surfaces the candidates.

02

Buy Savings Plans (compute)

High impactLow effort20-40%

Cover your baseline compute (the 40-60% that runs 24/7) with 1-year No-Upfront Compute Savings Plans. Coverage target: 65%.

03

Convert idle resources

High impactLow effort5-15%

EBS volumes detached >7 days, EIPs unattached, RDS instances with 0 connections — kill or snapshot+delete.

04

S3 lifecycle policies (IA + Glacier)

Medium impactLow effort40-70% on storage

S3 Standard → Standard-IA at 30 days, → Glacier IR at 90, → Glacier Deep at 180. Set once, save forever.

05

Cross-region data transfer audit

Medium impactHigh effort$0.02/GB egress

AWS Data Transfer line in Cost Explorer. Group by source/dest. Anything >$500/mo cross-region = bucket consolidation candidate.

06

Spot Instances for fault-tolerant workloads

High impactMedium effort60-90%

Batch jobs, dev/test, CI/CD runners, and stateless EKS nodes. Use mixed-instance ASGs with capacity-optimized strategy.

07

Lambda right-sizing (memory)

Medium impactLow effort30-50%

Lambda Power Tuning is free and tells you the cost-optimal memory size. Most Lambdas are over-provisioned.

08

NAT Gateway elimination

Medium impactMedium effort$32/mo per NAT GW + data

Replace with VPC endpoints for S3/DynamoDB. Massive savings if you have multi-AZ NAT GWs in dev accounts.

09

CloudWatch log retention defaults

Low impactLow effort50-90% on logs

Default = "Never expire". Set log group retention to 30/90 days max. Run `aws logs put-retention-policy` once.

10

Reserved Instances for RDS/ElastiCache

High impactLow effort30-50%

Stateful databases run 24/7 by definition. 1-year No-Upfront RIs are the safest cost lever in AWS.

11

Tag enforcement via SCP

High impactHigh effortEnables everything else

AWS Service Control Policy: deny resource creation without `owner`, `cost_center`, `env` tags. Without tags, you can't run any of the above.

12

Trusted Advisor weekly review

Medium impactLow effortvaries

Free with Business/Enterprise support. Runs the standard checks automatically. Should be a weekly ritual.

13

Graviton (ARM) migration

High impactHigh effort20% perf/$

Most workloads run on Graviton with zero code changes. Compatible with EKS, ECS, RDS, Lambda. Test in dev first.

14

GuardDuty/Inspector cost cap

Low impactLow effortvaries

These can scale silently. Cap their billing alarm at 110% of last month — surprises caught early.

Free audit

Run a free AWS cost audit

CARTIE AI runs all 14 patterns above against your AWS account in <10 minutes. Read-only IAM role, no agents.

Get the audit

THE FINOPS BRIEF

3 cost-saving tips, every Tuesday.

Built for finance & engineering teams who are tired of paying for cloud they don't use. No fluff. Just what works.

Unsubscribe anytime. We never sell your data.

We value your privacy. Cookies help us improve your experience. Learn more

Install CARTIE AI

Add to your home screen for quick access and offline support