Chrome DevTools documentation Official technical docs

What Google's Lighthouse performance score actually measures.

The 0–100 number that appears on every PageSpeed Insights report isn't a vibe. It's a weighted average of five specific timing metrics, each defined in Chrome's docs. Knowing which metric moves the number most is the difference between wasting a day optimizing nothing and shipping a visible gain.

Don's note

Every restaurant owner who has ever run their site through PageSpeed Insights has seen a number — 34, 61, 88 — and wondered what it means. The honest answer: it's Google's weighted roll-up of how quickly the page's main content paints, how stable the layout is while it loads, and how responsive the page feels while the browser is still working. Each of those has a definition and a weight, and once you know them, the score stops feeling like a black box.

For a restaurant site, the practical implication is that two metrics account for roughly 55% of the score — Largest Contentful Paint and Total Blocking Time. Fix those two and everything else follows. The audit tool on this site returns the per-metric breakdown so you can see exactly which bucket is dragging your score down.

Score 35 of 100 — Poor band 35

Poor0–49

Score 65 of 100 — Needs work band 65

Needs work50–89

Score 95 of 100 — Good band 95

Good90–100

The three Lighthouse bands — the colored dial in any PageSpeed Insights report uses these same thresholds.

Key findings

  • The Lighthouse performance score is a weighted average of five metrics (as of the current major version): First Contentful Paint, Speed Index, Largest Contentful Paint, Total Blocking Time, and Cumulative Layout Shift.
  • Current weights: FCP 10%, SI 10%, LCP 25%, TBT 30%, CLS 25%. TBT and LCP together account for over half the final number; optimizing them moves the score the most.
  • Each metric is scored against a log-normal curve built from the top ~10% of real-world sites on the HTTP Archive. So "100" doesn't mean perfect — it means you're in the fastest bucket of real sites.
  • Scores above 90 are "Good," 50–89 are "Needs improvement," and below 50 is "Poor." The colored dial in the report maps to these thresholds.
  • Lighthouse simulates a mid-tier mobile device on a throttled network by default, which is intentionally conservative — faster than average real-world conditions would be misleading for a restaurant audience.

How Muntin uses this

Full citation

Chrome DevTools. Lighthouse performance scoring. Chrome for Developers documentation. developer.chrome.com

Last reviewed: April 2026 — Lighthouse metric weights have shifted slightly across major versions; Muntin reviews this page whenever a new Lighthouse major is released.