Five AWS bill surprises we see in New England SMB accounts

Most teams we talk to in Maine and New England did not set out to waste money on AWS. The bill grew because defaults, legacy choices, and “temporary” resources stacked up quietly.

Here are five patterns we see repeatedly — none of them require a massive refactor to fix.

1. Orphaned EBS volumes and old snapshots

When an EC2 instance is terminated, attached volumes are often left behind. Snapshots from a migration two years ago may still be accruing storage charges. Run a monthly inventory: anything unattached for 30+ days is a candidate for review.

2. NAT Gateway data processing on low-traffic subnets

NAT Gateways charge per hour and per GB processed. Dev and staging VPCs that route all outbound traffic through NAT are a common leak. Consider VPC endpoints for S3 and DynamoDB, or tighten which subnets need outbound internet at all.

3. Over-provisioned RDS and “just in case” instance sizes

db.r6g.xlarge made sense during a launch week. Six months later, CPU averages 8%. Right-sizing — or moving dev databases to Aurora Serverless v2 — often cuts database spend materially without a migration project.

4. CloudWatch Logs never expiring

Application teams enable verbose logging; nobody sets retention. Log groups with indefinite retention are one of the fastest ways to grow storage costs. Set 30–90 day retention unless compliance requires longer (and archive to S3 if it does).

5. Data transfer between regions and out to the internet

Cross-AZ traffic, cross-region replication, and CloudFront origin fetches all show up under “Data Transfer.” Map your architecture once a quarter: same-AZ placement, regional services, and caching at the edge reduce surprises.

---

You do not need a FinOps platform on day one. A structured cost review — workload by workload — usually finds quick wins in the first month. That is often the best conversation starter before a broader managed cloud engagement.