Cost Optimising Your AWS Bill: 10 Practical Tips

Why Cloud Bills Spiral

AWS makes it trivially easy to provision resources and deceptively hard to understand what you're actually paying for. Teams spin up dev instances that never get torn down, leave S3 lifecycle rules unconfigured, and forget that NAT Gateway charges per GB. A typical client we onboard has 30–40% of their bill coming from resources that are either unused or dramatically over-provisioned.

Quick Wins (Do These First)

Before touching architecture, go after the low-hanging fruit. These usually take an afternoon and save thousands per month.

• Enable Cost Explorer and set billing alerts for every account
• Delete unattached EBS volumes (they charge even when stopped)
• Remove orphaned snapshots older than 90 days
• Downgrade RDS instances to Graviton3 equivalents (same price, 40% faster)
• Move infrequently accessed S3 data to Intelligent-Tiering
• Reserve capacity for any EC2 or RDS instance running >6 months

Architecture-Level Savings

Deeper savings come from rethinking the architecture. Moving from always-on EC2 to Lambda for bursty workloads can cut compute costs by 80% for the right use case. Replacing NAT Gateways with VPC Endpoints for S3 and DynamoDB eliminates per-GB charges entirely. Using CloudFront in front of S3 and API Gateway caches responses and reduces origin calls significantly.

Governance That Sticks

One-time audits don't stick. Put in place AWS Config rules that flag untagged resources, Service Control Policies that prevent production-scale instances in dev accounts, and a monthly cost review ritual where engineers own their team's bill. Cost accountability at the team level changes behaviour faster than any centralised optimisation effort.