The Challenge
RetailMax, a 12-store retail chain operating across Southeast Asia, was spending over $18,000/month on AWS infrastructure that had grown organically over five years. Resources were over-provisioned, environments weren't separated cleanly, and there was no tagging strategy — making cost attribution nearly impossible.
Our Approach
We began with a two-week discovery phase, using AWS Cost Explorer and Trusted Advisor to identify waste:
- Over 60% of EC2 instances were running at under 15% CPU utilisation
- Multiple development databases were running production-grade Multi-AZ RDS instances
- S3 lifecycle rules were absent, leaving terabytes of infrequently accessed data in Standard storage
- No use of Savings Plans or Reserved Instances despite predictable baseline workloads
The Solution
We delivered a phased migration plan across six weeks:
- Rightsizing: Downgraded 34 EC2 instances to appropriate sizes, saving $3,800/month immediately.
- Reserved Instances: Converted baseline workloads to 1-year Convertible RIs, locking in a 40% discount.
- S3 Intelligent Tiering: Enabled lifecycle rules moving objects to Glacier after 30 days of inactivity.
- Dev/Staging separation: Moved non-production RDS to Single-AZ and scaled down instance classes.
- Tagging strategy: Implemented a mandatory cost-centre tagging policy enforced via AWS Config rules.
The Result
Monthly AWS spend dropped from $18,200 to $9,900 — a 45.6% reduction — within 8 weeks, with zero production downtime. The new tagging strategy gave RetailMax's finance team per-store cost visibility for the first time.
The cloud migration project was flawless — zero downtime, 45% cost reduction, and our team was fully productive on day one. Proxy Soft exceeded every benchmark we set. — David Kim, VP Engineering
Key Takeaways
Cloud cost optimisation isn't a one-time task. We set up monthly Cost Anomaly Detection alerts and quarterly review cadences so RetailMax can stay lean as their infrastructure evolves.