A fifth score: website performance
We have released scoring version 2.2. Every fresh audit now includes a fifth branch: a dedicated Performance audit, evaluated on the same crawl as your SEO, GEO, WCAG and Security results. The combined score is now the average of all five pillars.
Your report gains a Performance score card with a letter grade, a per-category breakdown and prioritized fixes. The lighter Performance slice inside SEO remains as a delivery snapshot; the new pillar is the detailed UX speed audit.
What we measure (and why)
Google still uses three Core Web Vitals as ranking signals, measured from real Chrome users at the 75th percentile:
- LCP (Largest Contentful Paint) — is the main content fast? Good: under 2.5s.
- INP (Interaction to Next Paint) — do clicks feel instant? Good: under 200ms. INP replaced FID in March 2024.
- CLS (Cumulative Layout Shift) — does the layout stay still? Good: under 0.1.
Those three are the ranking bar. We also report the diagnostics that usually explain a failed vital:
- TTFB — Google's diagnostic target is under 800ms. A slow server makes a good LCP almost impossible.
- FCP, Speed Index, TBT, TTI — Lighthouse lab metrics for first paint, visual completeness, main-thread blocking (the lab proxy for INP) and interactivity.
- Delivery — compression (gzip/Brotli), cache headers, HTTP/2 or HTTP/3 hints, CDN fingerprints and early hints.
- Payload — HTML weight, modern image formats, render-blocking scripts/CSS, third-party tags, resource hints and
font-display.
How the Performance branch is scored
The audit is organized into four weighted categories:
- Core Web Vitals (40%) — lab LCP, INP and CLS from PageSpeed Insights (mobile Lighthouse), plus CrUX field data when Google has enough samples. HTML fallbacks still run without an API key: images without dimensions, a lazy-loaded first image, and render-blocking scripts in
<head>. - Loading (25%) — crawler TTFB (800ms good / 1800ms needs improvement), HTML download time, and lab FCP, Speed Index and Total Blocking Time.
- Delivery (20%) — whether bytes leave the server compressed and cacheable, and whether the response looks like HTTP/2/3 behind a CDN.
- Payload (15%) — what the homepage asks the browser to download: images, scripts, stylesheets, fonts and third-party origins.
Lab vs field
Lighthouse is a single simulated phone on a datacenter network. Google Search uses field data from the Chrome User Experience Report. When PSI returns CrUX metrics we show them separately and score them: a green lab LCP with a red field LCP means real users still wait — ship the fix and wait for the 28-day CrUX window.
If the API key is missing, the quota is exhausted, or the site is too small for CrUX, you still get the HTML and header audit. Re-run after configuring PAGESPEED_INSIGHTS_API_KEY to unlock lab and field vitals.
What you will see in the report
- A Performance score card with letter grade and 0–100 score.
- Per-category scores for Core Web Vitals, Loading, Delivery and Payload.
- Performance issues mixed into the prioritized recommendation list (quick wins vs larger projects).
- Metric tables in the HTML report and the PDF export, including field vs lab when both exist.
Re-run your audit
Reports from version 2.1 or older remain valid for SEO, GEO, WCAG and Security, but they do not include the Performance branch. Generate a fresh report to see your speed score on the same crawl.
See the methodology page for thresholds and weights, browse the check library for FCP, TBT, resource hints and third-party scripts, or read the scoring changelog.