Our client — a B2B SaaS platform with about 500 active companies — was spending 45L per year on AWS. After a two-week audit, we reduced that to 27L annually without changing a single user-facing feature.
Phase 1: The Audit
We started with AWS Cost Explorer and set up tag-based cost allocation. Most teams skip this step, which means they can't attribute costs to specific services or teams. Once costs were tagged, the breakdown was illuminating: 60% EC2, 25% RDS, 10% data transfer, 5% everything else.
Phase 2: Right-Sizing EC2
The client was running m5.2xlarge instances at 15-20% average CPU utilisation. We moved to a mix of t3.large and m5.large instances, reducing EC2 costs by 45%.
Phase 3: Reserved Instances and Savings Plans
For workloads that were clearly stable and long-running, we purchased 1-year Reserved Instances. This delivered a 37% discount compared to On-Demand pricing.
Phase 4: RDS Optimisation
We validated that the client's RTO requirements didn't require instant failover and migrated to a smaller standby configuration. We also enabled RDS storage autoscaling and removed 40% of provisioned IOPS that were never being consumed.
Phase 5: Data Transfer
The client was serving static assets from EC2 instead of CloudFront. Moving assets behind CloudFront reduced data transfer costs by 70%.
Final Numbers
Total annual saving: 18L (40% reduction). Implementation time: 3 weeks. Zero features changed. Zero downtime.