How one user's AWS bill went from ~$8K to ~$6K with our free CLI

I've been building GreenOps Scan, a free AWS waste scanner, on the side. The idea is simple: you run npx greenops-scan with a read-only AWS profile, and it gives you a list of idle or oversized resources with dollar estimates. No write access, no cloud uploads, no signup.

A few weeks ago I got to run it twice against the same account, eight weeks apart. I think the numbers are interesting for anyone here who has ever watched a cloud bill drift upward and not known why.

The account belongs to a mid-sized software company. No FinOps person, just a platform team keeping the lights on. Standard stack: ECS/Fargate, DocumentDB, ElastiCache, RDS, an ALB, Transfer Family, the usual. Their bill had gone from about $2K/month in late 2023 to nearly $8K by mid-2026.

First scan: 53 findings worth an estimated $1,177/month.

• 15 over-provisioned Fargate tasks — $756.83/month. One of them was 16 vCPU / 32 GB.
• 37 S3 buckets with no lifecycle policy — $415/month.
• 1 stopped EC2 instance still billing EBS — $5/month.

The team went through the list and made the changes by hand. The scanner doesn't touch anything; it just gives them a list with dollar amounts.

Second scan, eight weeks later: 1 finding worth $105.12.

That's a 91% drop in flagged waste. The one remaining item was a DocumentDB cluster on an Intel db.r5.xlarge. They later moved it to a Graviton db.r6g.large and downsized, so the actual savings were even bigger.

AWS's own Cost Explorer forecast has the next month landing around $5,000–$6,000. That's not a closed month, it's a forecast, but it's still AWS's model reacting to the cleanup. The $1,177 → $105 drop is the confirmed part.

What I take from this: the waste wasn't from anyone being lazy. It was from shipping. Prototypes, launches, "just in case" sizing. The kind of stuff every bootstrapped/SaaS company does. The difference was just having a list.

If you're paying AWS bills, the CLI is free and read-only: npx greenops-scan.

Full case study with the Cost Explorer graph: https://greenops.cloud/blog/blog-cost-drop-8k-to-6k?utm_source=indiehackers&utm_medium=post&utm_campaign=8k_to_6k_case_study

What's the biggest waste pattern you've seen in your own stack? For me it's Fargate sizing and S3 buckets with no lifecycle rules.