5 AWS Cost Mistakes We See Every Week

Cloudtrim Team · 2026-02-24 · 5 min read

5 AWS Cost Mistakes We See Every Week


Every week we scan dozens of AWS accounts for agencies and consultancies. The patterns are remarkably consistent — the same handful of mistakes show up again and again, quietly draining budgets.


Here are the five most common cost mistakes we find, and what you can do about each one.



1. Idle EC2 Instances Running 24/7


This is the single biggest source of waste we see. Dev and staging environments left running over weekends, instances spun up for a demo that nobody remembered to terminate, or old projects that quietly keep billing.


  • **Average monthly waste per idle instance: $50–$200**
  • Most agencies have at least 3–5 idle instances across their accounts
  • The fix is simple: schedule stop/start or terminate what you don't need

    • We regularly find instances with CPU utilisation below 2% that have been running for months. That adds up fast when you manage 10+ accounts.



    2. Oversized RDS Instances


    Agencies love to "play it safe" with database sizing. The result? Production databases running on db.r5.xlarge when db.t3.medium would handle the load just fine.


  • **Right-sizing can cut RDS costs by 40–60%**
  • Check CloudWatch metrics for CPU and memory utilisation
  • Consider Aurora Serverless for variable workloads

    • The key metric to watch is average CPU utilisation over 14 days. If it's consistently below 20%, you're almost certainly over-provisioned.



    3. No Reserved Instance or Savings Plan Coverage


    On-Demand pricing is the most expensive way to run AWS. If you have workloads that run consistently, you're leaving money on the table without RI or Savings Plan coverage.


  • **Savings Plans can reduce compute costs by up to 72%**
  • Even 1-year no-upfront commitments save 30–40%
  • Start with Compute Savings Plans for maximum flexibility

    • We see agencies paying full On-Demand rates for workloads that have been running steadily for over a year. That's thousands of dollars in unnecessary spend.



    4. S3 Storage in the Wrong Tier


    Not all data needs to live in S3 Standard. Logs, backups, and infrequently accessed assets should be in cheaper tiers.


  • **S3 Intelligent-Tiering automates this for you**
  • Glacier is 90% cheaper than Standard for archival data
  • Lifecycle policies take 5 minutes to set up

    • One agency we scanned had 4TB of CloudTrail logs in S3 Standard dating back three years. Moving to Glacier saved them over $80/month instantly.



    5. Unattached EBS Volumes and Elastic IPs


    When you terminate an EC2 instance, its EBS volumes don't always go with it. Same with Elastic IPs — they cost money when they're not attached to a running instance.


  • **Unattached EBS volumes cost $0.10/GB/month**
  • Unused Elastic IPs cost $3.65/month each
  • These are easy wins that take minutes to clean up

    • We typically find 10–20 orphaned EBS volumes per account. It's not huge money individually, but across a portfolio of accounts it adds up to real savings.



    What To Do Next


    The good news is that all five of these mistakes are fixable without any architectural changes. Most can be resolved in an afternoon.


    Cloudtrim scans your accounts automatically, identifies exactly where the waste is, and generates the CloudFormation or Terraform code to fix it. No guesswork, no spreadsheets — just ready-to-apply infrastructure code.


    Stop guessing where your AWS budget is going. Start trimming.

    Ready to cut your AWS costs?

    Cloudtrim finds waste across your AWS accounts and generates the IaC to fix it. See it in action with our interactive demo.

    Try Cloudtrim Free