performance-budgets

A performance budget establishes measurable thresholds across time, size, and count dimensions to catch regressions early. It replaces subjective goals with enforceable contracts specifying metric, threshold, percentile, and consequence for each route.

To enforce Core Web Vitals in CI, define a p75 budget for metrics like LCP, INP, and CLS per route, then configure your deploy gate to automatically fail the build when those thresholds are breached.

Lighthouse budgets.json defines size and count thresholds for your application. You operationalize enforcement by treating budget breaches as build failures, ensuring automated deploy gates block regressions before they reach production.

Lab measurement provides controlled synthetic testing for consistent debugging, while field measurement captures real user experience at specific percentiles like p75. A complete budget defines measurement expectations for both environments.

Yes, per-route budget discipline requires specifying distinct metric thresholds for each meaningful route. This calibration ensures performance contracts accurately reflect the specific loading and interaction demands of different application views.

Use Core Web Vitals like LCP, INP, and CLS for user-facing loading and interaction metrics, and apply RAIL model limits for specific interaction classifications. Both standards define time thresholds that can trigger CI enforcement.