performance-optimization

Replace per-record queries with a single query using joins or eager loading, such as Prisma's include option. Detect N+1 patterns by enabling database query logging and looking for repeated similar queries within one request.

For LCP, optimize hero images with responsive srcset, modern formats like AVIF, and fetchpriority="high". For INP, break up long tasks over 50ms using scheduler.yield() and defer non-critical work out of event handlers.

Recommended budgets are under 200KB gzipped JavaScript, under 50KB CSS, API responses under 200ms at p95, and a Lighthouse score of at least 90. Enforce them with bundlesize and Lighthouse CI in your pipeline.

No. React.memo helps only for expensive components that re-render with identical props, and overusing it adds comparison overhead. Profile first with the Performance tab and apply memoization only where measurements show benefit.

Avoid optimizing without profiling evidence of an actual problem. Premature optimization adds complexity and maintenance cost without measurable gains, and changes that show no improvement over baseline should be reverted.

The change may have targeted the wrong bottleneck, or the gain is within run-to-run variance. Re-measure under identical conditions as the baseline, compare against variance rather than the mean, and revert if the delta is noise.