Lighthouse reports a score from 0 to 100 for each of four categories: Performance, Accessibility, Best Practices and SEO. Since Lighthouse 13.3 there is also a fifth, Agentic Browsing, which is still in beta and is shown as checks passed (for example 3/4) instead of a 0 to 100 score.
| Score | Rating |
|---|---|
| 90 to 100 | Good |
| 50 to 89 | Needs improvement |
| 0 to 49 | Poor |
The PWA category was removed in Lighthouse 12.
How the performance score is calculated
Since Lighthouse 10, the performance score is a weighted average of five lab metrics, each first converted to a 0 to 100 score. These are still the weights in Lighthouse 13:
| Metric | Weight |
|---|---|
| Total Blocking Time | 30% |
| Largest Contentful Paint | 25% |
| Cumulative Layout Shift | 25% |
| First Contentful Paint | 10% |
| Speed Index | 10% |
Interaction to Next Paint is not part of the score, because a page load has no user interactions to measure. Desktop runs use their own scoring curves, so the same timings can score differently on mobile and desktop.
Only these metrics affect the performance score. Lighthouse 13 also lists insights, such as LCP request discovery, Render blocking requests and Improve image delivery, which estimate how much time a fix could save but do not change the score directly.
The Accessibility, Best Practices and SEO scores are weighted averages of passing or failing individual checks.
How Lighthouse tests a page
By default Lighthouse loads the page once with simulated mobile throttling: a “slow 4G” connection (150 ms latency, 1.6 Mbps down, 750 Kbps up) and a CPU slowed down 4 times. Desktop runs use a faster connection and no CPU slowdown.
Why scores change between runs
Performance scores vary because of network conditions, CPU load, server response times, A/B tests, ads and browser extensions. Google recommends running a page several times and using the median; AiroPages uses PageSpeed Insights settings on every page, and you can re-scan any page to check how much it moves.