The $100 Billion Problem
Cloud waste is staggering. Gartner estimates that organizations waste over 30% of their cloud spend — and that number climbs to 40-60% for companies without a dedicated FinOps practice.
We've seen it firsthand across 50+ engagements: idle resources running 24/7, oversized instances handling 5% load, untagged resources that nobody owns, and commitment plans that expired months ago.
The good news? Most of this waste can be eliminated in 30 days with the right approach.
Week 1: Gain Visibility
You can't optimize what you can't see. The first week is about building a complete picture of your cloud estate.
Step 1: Implement a Tagging Strategy
Tags are the foundation of cloud cost management. Without them, you're flying blind.
# Minimum viable tagging policy
required_tags:
- team # Who owns this resource?
- environment # prod / staging / dev / sandbox
- service # What application does this support?
- cost-center # Who pays for it?
- managed-by # terraform / manual / helmEnforce tags at deployment time using policy-as-code tools like OPA or Kyverno. Any untagged resource should be flagged for review within 48 hours.
Step 2: Deploy Cost Visibility Tooling
Set up Kubecost (for Kubernetes) and cloud-native cost explorers. Build dashboards that show:
- ●Daily spend by team, service, and environment
- ●Week-over-week trends
- ●Top 10 most expensive resources
- ●Untagged resource inventory
Step 3: Identify Quick Wins
Run a rightsizing analysis. Look for:
- ●Idle resources: EC2 instances, RDS databases, and EKS nodes with < 5% CPU utilization
- ●Oversized instances: Resources consistently using < 30% of allocated capacity
- ●Orphaned resources: Unattached EBS volumes, unused Elastic IPs, stale snapshots
- ●Dev/staging environments running 24/7 when they're only used during business hours
Quick win: In one engagement, we found $180K/year in orphaned EBS volumes alone. They'd been accumulating for 18 months.
Week 2: Execute Quick Wins
Rightsizing
Downsize instances that are consistently underutilized. Start with non-production environments — they're lower risk and often the most oversized.
# Example: Check average CPU utilization over 14 days
aws cloudwatch get-metric-statistics \
--namespace AWS/EC2 \
--metric-name CPUUtilization \
--dimensions Name=InstanceId,Value=i-1234567890abcdef0 \
--start-time 2026-03-01T00:00:00Z \
--end-time 2026-03-14T23:59:59Z \
--period 3600 \
--statistics AverageScheduling Non-Production Environments
Dev and staging environments don't need to run at 3 AM on a Sunday. Implement automated scheduling:
- ●Business hours only: 8 AM - 8 PM, Monday - Friday
- ●Savings: Typically 65-70% reduction in non-prod compute costs
- ●Tools: AWS Instance Scheduler, custom Lambda functions, or Kubernetes CronJobs with scale-to-zero
Delete Orphaned Resources
Create a cleanup script that identifies and removes:
- ●Unattached EBS volumes (> 7 days old)
- ●Unused Elastic IPs
- ●Snapshots older than your retention policy
- ●Empty S3 buckets with no access in 90+ days
Week 3: Optimize Commitments
Reserved Instances & Savings Plans
Analyze your steady-state workloads and purchase commitments:
- ●1-year no-upfront: Good balance of savings (20-30%) and flexibility
- ●3-year partial upfront: Maximum savings (40-60%) for stable workloads
- ●Convertible RIs: Slightly lower savings but you can change instance families
Spot Instances
For fault-tolerant workloads (batch processing, CI/CD, dev environments), spot instances offer 60-90% savings:
- ●Use spot fleets with multiple instance types
- ●Implement graceful interruption handling
- ●Mix on-demand and spot in auto-scaling groups
Week 4: Governance & Sustainability
Budget Alerts
Set up tiered alerts at 50%, 80%, and 100% of budget thresholds. Route them to team Slack channels, not just email.
Anomaly Detection
Configure automated anomaly detection to catch cost spikes within hours, not at the end of the month.
FinOps Review Cadence
Establish a weekly FinOps review:
- ●Review top cost changes (up and down)
- ●Track optimization progress
- ●Celebrate wins and assign new action items
Results We've Seen
| Client | Starting Spend | Savings | Timeline |
|---|---|---|---|
| Series B SaaS | $120K/month | 38% reduction | 28 days |
| Enterprise FinTech | $450K/month | 42% reduction | 30 days |
| Healthcare Platform | $85K/month | 35% reduction | 21 days |
The Bottom Line
Cloud cost optimization isn't a one-time project — it's a practice. But the first 30 days are where you capture the biggest wins with the least effort.
Start with visibility, execute the quick wins, optimize commitments, and build governance that prevents waste from creeping back.
If you want a personalized cost optimization roadmap for your environment, book a free assessment.