AWSCostAWSCost · example report
The whole report, before you connect anything.
14 findings worth $1498.26/month on an AWS bill of $8420.55/month, across us-east-1 and eu-west-1. Below is exactly what a customer sees: the free scan first, then everything the $99 report adds.
The account is invented - I don’t publish a customer’s. Every dollar on it is not: each figure was priced against the AWS Price List on 13 August 2026, by the same code that prices a real scan, and the two panels are rendered by the same components your own report would use.
Free: what the scan says
The scan costs nothing and you see the total before deciding. Findings stay closed until the report is paid for.
Findings
We found $1498.26/month of potential savings across 14 findings in account 111122223333.
The full report - every resource, what it costs, and what to change - is $99, one time. The price is set by your measured AWS bill of $8420.55/month.
Not useful? Email within 14 days and we refund it. No questions.
Reading your account cost you about $0.06 - 6 Cost Explorer requests. Every other call we made was free.
$99 one time: what it opens
Every finding, the evidence behind it, what changing it would cost you - and the two lists most reports leave out: what the scan couldn’t read, and the services it doesn’t check yet.
Findings
We found $1498.26/month of potential savings across 14 findings in account 111122223333.
- OpenSearch$486.18/month
logs-archive · us-east-1 · med confidence
no searches and no writes over 14 days - peak SearchRate and IndexingRate were both 0 across all 6 nodes (3 x r5.large.search data, 3 x t3.small.search dedicated master), which bill by the hour whether the domain answers a query or not; the estimate covers instance hours only, so the domain's EBS storage is on top of it
Recommended: delete the domain if whatever indexed into it is gone - this is not reversible and the indices go with it, so take a manual snapshot to S3 first and check nothing dashboards or alerts off it; keep it if it is a deliberate warm standby, which looks identical from out here
- RDS$287.62/month
reporting-prod · eu-west-1 · med confidence
avg CPU 4.2% with 1.30 connections over 14 days (class db.m5.xlarge)
Recommended: consider the next smaller class after checking memory and connection headroom
- EFS$240.00/month
fs-0a1b2c3d4e5f60718 · us-east-1 · high confidence
no mount target in any subnet, so nothing can reach the data - 800.0 GiB on Standard
Recommended: delete it if the data is finished with - it cannot be undone, so copy what matters to S3 first; if it is being kept deliberately, a lifecycle policy moves untouched files to Infrequent Access at about a tenth of the price
- EC2$156.22/month
i-0f2e4c6a8b1d3e5f7 · eu-west-1 · med confidence
avg CPU 1.6% over 14 days (type m5.xlarge)
Recommended: right-size to a smaller type, or stop/terminate if unused
- EKS$73.00/month
staging-blue · us-east-1 · high confidence
no managed node groups, no Fargate profiles, and no EC2 instance tagged for this cluster, created 287 days ago (the control plane bills its hourly charge regardless)
Recommended: delete the cluster if its workload has moved - the control plane charge stops immediately, but deleting it destroys the cluster's configuration (add-ons, access entries, logging, auth), so bringing the workload back means rebuilding that
- ECS$72.08/month
prod/legacy-importer · us-east-1 · med confidence
CPU never rose above 0.4% over 14 days - 2 tasks at 1 vCPU / 2 GB bill by the second whether anything calls the service or not; the estimate covers vCPU and memory hours only, so ephemeral storage, data transfer and any load balancer in front of it are on top
Recommended: set the desired count to 0 if whatever called this is gone - the charge stops and the service and its definition survive, so it can be scaled back up; delete the service once you're sure. Check first that it isn't a consumer waiting on an empty queue or a standby kept for failover, which look identical from out here
- EBS$40.00/month
vol-0d3c2b1a09f8e7d6c · us-east-1 · high confidence
state=available, size=500 GiB, created 2025-11-04T09:12:44.000Z
Recommended: delete or snapshot-then-delete if not needed
- NATGateway$35.04/month
nat-0c9d1e2f3a4b5c6d7 · eu-west-1 · high confidence
0 B outbound over 14 days (fixed hourly charge bills regardless of traffic)
Recommended: remove if the private subnets no longer need outbound, or replace with a VPC endpoint / shared NAT
- EBS$32.00/month
34 stale snapshots · eu-west-1 · med confidence
34 snapshots older than 180 days, none backing an AMI or managed by AWS Backup; snapshots are incremental, so 640 GiB of source volume is the upper bound on what deleting them frees
Recommended: confirm your retention window, then prune with a Data Lifecycle Manager policy so they age out automatically
- VPCEndpoint$24.09/month
vpce-08b7c6d5e4f3a2b1c · eu-west-1 · high confidence
0 B processed over 14 days for com.amazonaws.eu-west-1.secretsmanager, billed in 3 subnets (the hourly charge bills per subnet regardless of traffic)
Recommended: remove if nothing in the VPC still calls this service - callers then fall back to the public endpoint over your NAT/internet path, which bills per GB, and private DNS for the service stops resolving inside the VPC
- PublicIPv4$21.90/month
6 public IPv4 address(es) · us-east-1 · high confidence
4 attached to running instances, 2 on network interfaces with no instance
Recommended: drop the public IP / move behind a NAT gateway or load balancer / use IPv6 where inbound public reach isn't needed
- ALB$16.43/month
arn:aws:elasticloadbalancing:us-east-1:111122223333:loadbalancer/app/checkout-old/1a2b3c4d5e6f7a8b · us-east-1 · high confidence
no healthy targets in any of its 2 target groups (the fixed hourly charge bills regardless)
Recommended: remove if the backend service is gone, or consolidate onto a shared load balancer
- CloudWatch Logs$9.30/month
12 log groups with no retention · us-east-1 · high confidence
12 groups never expire, 310.0 GiB stored in total. Storage is what a retention policy reclaims; ingestion is already spent
Recommended: set a retention period that matches what you would actually go back and read - 30 days covers most debugging, and anything you must keep longer belongs in S3 at a fraction of the price
- EBS$4.40/month
vol-07a6b5c4d3e2f1a09 · eu-west-1 · high confidence
volumeType=gp2, size=200 GiB
Recommended: modify to gp3 (same size); ~20% cheaper, no downtime
We don’t check these services yet, so their spend isn’t in the total above.
- Amazon Redshift$612.40/month
- AWS Glue$188.22/month
- Amazon Kinesis$96.31/month
- AWS WAF$41.05/month
This check ran, but the data it needs isn't available in this account. We left it out rather than guess.
- Lambda memory sizing - 23 of the 31 functions here have no Lambda Insights, so nothing publishes what they actually use (us-east-1)
This check measured part of what qualified, not all of it - the reads it makes are billed to you, so we bound how many it makes. There is nothing to fix; just read its total as a floor rather than the whole picture.
- CloudWatch Logs ingestion class - we measured the 20 largest of 46 log groups, so that total is a floor (eu-west-1)
Every price above is AWS's list price for that resource's region, read during the scan. We assume on-demand list prices and that S3 sits in its first storage tier.
Reading your account cost you about $0.06 - 6 Cost Explorer requests. Every other call we made was free.
Yours takes about five minutes: deploy one read-only role, and the scan reads every region you have switched on. You see your total before you pay for anything.
The role can only read - Get, Describe, List, nothing else. Read the exact permissions before you deploy it, and delete the stack whenever you want.