Performance Optimization

Workflow

1. Define the goal and the metric (p95 latency, throughput, memory, Core Web Vitals, cost).
2. Establish a baseline (before measurements).
3. Profile to find bottlenecks (do not guess).
4. Apply targeted optimizations (small, reversible steps).
5. Re-measure and compare to baseline (quantify wins and trade-offs).
6. Roll out safely and monitor for regressions.

Quick Intake (Ask/Confirm)

• What metric is failing and where is it measured (RUM vs synthetic vs APM)?
• What is the target (SLO) and who is impacted?
• What changed recently (deploys, schema, traffic, feature flags)?
• What environment is representative (prod-like data, device/network class)?