Cloud Cost Optimisation Checklist 2026 — 25 Actions, Prioritised by Impact

This checklist covers 25 cloud cost optimisation actions, prioritised by impact. The first 10 are immediate wins you can implement this week. The remaining 15 are strategic changes that deliver larger long-term savings. Use the TCOIQ ROI Calculator to estimate your total saving potential before starting.

🚀 Immediate Wins (This Week)

1. Enable AWS Cost Anomaly Detection — Free, takes 30 minutes. Prevents surprise bills from misconfigured resources. Zero ongoing effort after setup.
2. Delete unattached EBS volumes — Filter EC2 Volumes by "available" state. Common to find hundreds of GB at $0.10/GB/month. Typical saving: $200-2,000/month.
3. Release unused Elastic IPs — AWS charges $3.65/month per unattached EIP. Check EC2 → Elastic IPs → filter "unassociated".
4. Stop idle dev/test instances nights and weekends — A script or AWS Instance Scheduler can stop dev instances from 7pm-7am and weekends. Saves 65% of dev compute costs.
5. Delete old EBS snapshots — Old snapshots accumulate at $0.05/GB/month. Filter snapshots older than 90 days and delete non-production ones.
6. Check for forgotten GPU instances — A single p3.2xlarge running idle costs $918/month. Search all regions for running GPU instances.
7. Enable S3 Intelligent-Tiering on large buckets — Automatically moves objects between access tiers. No retrieval fees, small monitoring fee. Typical saving: 20-40% on storage.
8. Review NAT Gateway costs — NAT Gateways charge $0.045/GB processed. Architecture changes (VPC endpoints for S3/DynamoDB) can eliminate significant NAT Gateway traffic.
9. Tag all untagged resources — Without tags, you can't allocate costs or identify waste. Run AWS Config rule "required-tags" to find untagged resources.
10. Set up budget alerts at 80% and 100% — AWS Budgets, Azure Budgets, GCP Budget Alerts. Free. Ensures you never exceed budget without warning.

📈 High-Impact Optimisations (This Month)

11. Purchase Reserved Instances for always-on workloads — Run Savings Plans recommendations in Cost Explorer. Purchase 1-year reservations for instances running >80% of the month. Saves 35-40%.
12. Implement Spot instances for CI/CD — Build agents are perfect for Spot — stateless, interruptible, batch workload. 65-70% saving on build compute.
13. Rightsize overprovisioned instances — Pull Compute Optimizer recommendations, implement top 20 by estimated saving. Typically saves 10-20% of compute.
14. Add S3 lifecycle rules to top 5 buckets — Transition to IA at 30 days, Glacier at 90 days. Takes 2 hours, saves 50-70% on cold data.
15. Implement AI model tiering — Route simple AI tasks to cheap models (Gemini Flash at $0.10/M vs GPT-4o at $2.50/M). Saves 65-70% of AI API costs.
16. Review and right-size RDS instances — Database instances are often over-provisioned by 2×. Use Performance Insights to identify CPU/memory headroom.
17. Enable GCP Sustained Use Discounts verification — Ensure GPU instances are covered. GCP extended SUD to A100/L4/T4 in Feb 2026.
18. Consolidate underutilised accounts — Multiple AWS accounts with low utilisation can consolidate under an Organisation for volume discounts and simplified billing.

🏗️ Strategic Changes (This Quarter)

19. Move 70%+ of compute to reserved pricing — Target: 70% reserved, 20% spot, 10% on-demand. Most organisations are at 30-40% reserved. Closing this gap saves $10K-100K+/month.
20. Evaluate OCI for high-egress workloads — If egress exceeds $2,000/month, OCI's $0.0085/GB vs $0.09/GB saves 90% on data transfer. Model with TCOIQ TCO Calculator.
21. Implement chargeback or showback to teams — Showing teams their cloud costs reduces waste by 15-20% within 90 days — no other change required.
22. Containerise and move to serverless where appropriate — GCP Cloud Run at $0.000024/vCPU-second (after March 2026 price cut) is often cheaper than always-on EC2 for variable traffic.
23. Purchase 3-year reservations for long-term stable workloads — 55-60% off on-demand. Only for workloads you're confident will run for 3+ years.
24. Negotiate Enterprise Discount Programme — At $100K+/month, negotiate directly with AWS, Azure or GCP for additional discounts. Typically 10-20% additional reduction.
25. Implement automated resource cleanup policies — Use AWS Config, Azure Policy or GCP Org Policies to automatically enforce cost governance at scale.

Measuring Success

After implementing these changes, track your progress with the TCOIQ Cloud AI Audit — retake it monthly to see your health score improve and identify remaining opportunities.