Guides
Audit Website Performance Without Fake Core Web Vitals
A fast HTML response is useful evidence, but it is not the same measurement as LCP, INP or CLS in a real browser.
Use server timing and markup budgets to find delivery risks. Use a browser lab test and field data before making a Core Web Vitals claim.
Start with the response path
Record DNS resolution, TLS handshake, time to first byte (TTFB), redirect hops, server compression (Brotli or Gzip), and HTML transfer size. A lightweight HTML document under 150 KB allows browsers to parse and render initial tokens quickly. Fix failed responses, unnecessary 301/302 hops, and slow origin processing before optimizing secondary assets.
Separate engineering budgets from Core Web Vitals
HTML weight, DOM complexity (keeping total elements under 1,500), and script counts are server and markup engineering budgets. Core Web Vitals describe actual user experience and require browser layout engines; field metrics (LCP, INP, CLS) must be evaluated across 75th percentile real-user visits (such as Chrome User Experience Report data) rather than synthetic server pings.
Inspect the critical rendering path
Identify classic blocking scripts in the document head that lack async, defer, or type='module' attributes. Review stylesheet requests (aiming for fewer than 6 critical CSS files) and inline script volume. Moving large inline scripts (> 100 KB) into external cached assets frees main-thread execution time during initial HTML parsing.
Optimize asset delivery and image strategies
Ensure below-the-fold images declare native loading='lazy' along with explicit width and height attributes to prevent cumulative layout shift (CLS). Preserve the above-the-fold hero image as eager loading (fetchpriority='high') so LCP is not artificially delayed. Adopt modern compressed formats like WebP or AVIF for rich media.
Prioritize by user impact and verify
Tackle slow backend database queries and render-blocking assets before pursuing micro-optimizations. Measure real field telemetry and run automated lab tests under throttling before and after each meaningful change to confirm measurable speed improvements.