Cut server costs by 88% while making the product 2–3× faster.
- Problem
- API responses were crawling past 600ms, concurrency was capped, and a $500/month infrastructure bill was growing faster than revenue.
- Approach
- Ran a performance analysis to find where time and money were actually going
- Prioritised a rebuild around the slowest, most-used paths
- Validated outcomes with SQL-based reporting and stakeholder feedback
- Outcome
- Response times dropped to 200–300ms, the monthly bill fell to $60, and the platform now handles more concurrent users on less hardware.
+ higher concurrent-user capacity on the same traffic



