Lighthouse score calculator
The performance score isn’t an average of your timings. Each metric goes through its own scoring curve, then gets weighted. Drag the metrics and watch where the points come from.
Each bar is a metric’s weight; the filled part is the points it earned.
The weights (Lighthouse 10 and later)
| Metric | Weight | Score 90 at (mobile) | Score 50 at (mobile) | Score 90 at (desktop) |
|---|---|---|---|---|
| Total Blocking Time | 30% | 200 ms | 600 ms | 150 ms |
| Largest Contentful Paint | 25% | 2.5 s | 4.0 s | 1.2 s |
| Cumulative Layout Shift | 25% | 0.10 | 0.25 | 0.10 |
| First Contentful Paint | 10% | 1.8 s | 3.0 s | 0.93 s |
| Speed Index | 10% | 3.4 s | 5.8 s | 1.3 s |
The curves come from real-world data in the HTTP Archive: a metric scores 50 at roughly the median of real sites and 90 at about the top 8%. So a “good” score isn’t absolute. It means you beat most of the web.
What the calculator teaches
- TBT and LCP decide most scores. Together with CLS they are 80% of the total. Speed Index and FCP rarely move the needle alone.
- Scores are non-linear. Improving LCP from 6 s to 4 s is worth far more points than from 2 s to 1 s.
- Desktop is graded harder. The desktop curves expect much faster timings, which balances the faster test device.
The other three scores
Accessibility, Best Practices and SEO are weighted averages of pass/fail audits: a failed audit with a weight of 10 costs more than one with a weight of 1. The Lighthouse test lists every failed audit for your page.
Questions people ask
How is the Lighthouse performance score calculated?
Each of five lab metrics (FCP, Speed Index, LCP, TBT and CLS) is turned into a 0–100 metric score using a log-normal curve based on real website data from the HTTP Archive. The performance score is a weighted average: TBT 30%, LCP 25%, CLS 25%, FCP 10% and Speed Index 10%.
Why is it so hard to get from 90 to 100?
The scoring curves flatten out near the top: going from a metric score of 0.9 to 1.0 needs a far bigger time improvement than going from 0.5 to 0.6. Past 90, the returns shrink quickly, and a 100 is rarely worth chasing over good field data.
What is a good Lighthouse score?
90 or above is good, 50–89 needs improvement and below 50 is poor. The same colour bands apply to the accessibility, best practices and SEO categories.
How is the Lighthouse SEO score calculated?
The SEO category is a weighted pass/fail list: the page must be indexable, have a title, a meta description, a valid robots.txt, descriptive link text, crawlable links, a valid hreflang and canonical if present, and a successful HTTP status. Each passed audit adds its weight.
Does INP affect the Lighthouse score?
No. INP needs real interactions, so it doesn’t exist in a lab load. Total Blocking Time is the closest lab signal and carries the biggest weight, 30%.