5 AWS Cost Optimization Strategies Every FinOps Team Should Know

Cloud spending has a way of quietly spiralling. One team spins up an EC2 cluster for a proof-of-concept, another forgets to delete a NAT Gateway, and before long the monthly AWS bill has doubled with nobody quite sure why.

In my work as a Cloud FinOps Analyst, I've seen this pattern repeatedly. The good news: most cloud waste is addressable with a structured approach. Here are the five strategies I return to on every engagement.

1. Right-size Your EC2 Instances

The single biggest source of waste I encounter is over-provisioned compute. Teams choose instance types based on peak load estimates (or fear), and those instances then idle at 5–15% CPU utilization most of the month.

How to find candidates:

• Open AWS Cost Explorer → Right Sizing Recommendations
• Use CloudWatch metrics to identify instances averaging < 20% CPU over 14 days
• Consider memory and network alongside CPU — don't just right-size on CPU alone

What to do:

• Move from m5.2xlarge to m5.large where utilization supports it
• Test in staging first; document the instance type change for rollback
• Set up CloudWatch alarms at 70% CPU so you'll know if utilization rises after resizing

A single rightsizing pass on a mid-sized AWS account can yield 15–30% savings on EC2 alone.

2. Commit with Reserved Instances and Savings Plans

On-Demand pricing is convenient, but you pay a significant premium for that flexibility. If you have any predictable, steady-state workloads (and most organizations do), you should be committing.

Reserved Instances (RIs):

• 1-year or 3-year commitments on specific instance families
• Up to 72% discount vs On-Demand
• Convertible RIs give flexibility to change instance type if needed

Compute Savings Plans:

• More flexible than RIs — apply to any EC2 instance, Lambda, and Fargate
• Up to 66% discount
• Recommended for most teams as the starting point

My rule of thumb: Look at your lowest-watermark EC2 spend over the past 3 months. That baseline is safe to commit with 1-year Standard RIs or a Compute Savings Plan.

Baseline monthly EC2 spend:    $10,000
Committed at 40% discount:     $6,000/month
Annual saving:                 $48,000

3. Clean Up Idle and Orphaned Resources

Every AWS account accumulates zombie resources — things that were created, forgotten, and are now silently billing you every month.

Common culprits to check:

Automation tip: Use AWS Trusted Advisor (available on Business/Enterprise support) or the open-source tool Cloud Custodian to schedule automated detection and alerting of idle resources.

4. Implement a Tagging Strategy

You cannot optimize what you cannot measure. Without tags, AWS cost reports show you what you're spending — not who is spending it or why.

A solid tagging strategy should include at minimum:

Environment:   production | staging | dev
Team:          platform | data | frontend | backend
Project:       the product or initiative driving the spend
CostCenter:    maps to your finance team's chart of accounts

Getting started:

1. Define your tag schema in a company wiki page
2. Use AWS Tag Editor to find untagged resources
3. Enforce tagging at resource creation with AWS Service Control Policies (SCPs) or AWS Config rules
4. Build a Cost Allocation report in Cost Explorer using these tags

Once tagging is in place, you can produce team-level cost reports and create accountability for spend — which is the foundation of the FinOps framework.

5. Use S3 Intelligent-Tiering and Lifecycle Policies

S3 is deceptively easy to ignore — objects are cheap individually, but at scale the storage costs and data retrieval fees add up fast.

S3 Intelligent-Tiering:

• AWS automatically moves objects between frequent and infrequent access tiers
• No retrieval fees for Intelligent-Tiering
• Best for objects with unpredictable access patterns

Lifecycle Policies (for predictable patterns):

• Objects not accessed for 30 days → move to S3 Standard-IA (saves ~45%)
• Objects not accessed for 90 days → move to S3 Glacier Instant Retrieval
• Objects not accessed for 180 days → expire or move to Glacier Deep Archive

{
  "Rules": [{
    "Status": "Enabled",
    "Transitions": [
      { "Days": 30,  "StorageClass": "STANDARD_IA" },
      { "Days": 90,  "StorageClass": "GLACIER_IR" },
      { "Days": 180, "StorageClass": "DEEP_ARCHIVE" }
    ]
  }]
}

One client I worked with reduced their S3 bill by $4,200/month simply by applying lifecycle policies to their log archive buckets.

Putting It Together

None of these strategies requires a massive initiative. You can start small:

1. Week 1: Pull rightsizing recommendations from Cost Explorer and pick 5 instances to resize
2. Week 2: Identify and delete orphaned EBS volumes and Elastic IPs
3. Month 1: Define and enforce a tag schema for new resources
4. Month 2: Purchase a Compute Savings Plan for your baseline EC2 spend
5. Ongoing: Review Cost Explorer weekly; set budgets and alerts

Cloud cost optimization is a practice, not a project. The teams that treat it as ongoing discipline — not a one-time cleanup — are the ones who keep their AWS bills predictable and under control.

Have questions about any of these strategies? Get in touch — I'm happy to discuss your specific AWS environment.