Performance Optimization

Overview

Measure before optimizing. Performance work without measurement is guessing — and guessing leads to premature optimization that adds complexity without improving what matters. Profile first, identify the actual bottleneck, fix it, measure again. Optimize only what measurements prove matters.

When to Use

• Performance requirements exist in the spec (load time budgets, response time SLAs)
• Users or monitoring report slow behavior
• Core Web Vitals scores are below thresholds
• You suspect a change introduced a regression
• Building features that handle large datasets or high traffic

When NOT to use: Don't optimize before you have evidence of a problem. Premature optimization adds complexity that costs more than the performance it gains.

Core Web Vitals Targets

| Metric | Good | Needs Improvement | Poor |