Lighthouse runs a battery of automated audits and rolls them up into category scores, but the headline number hides how it is calculated. Understanding which audits belong to which category — and how much each weighs — is the difference between guessing and fixing the right thing. This reference lists every category, the key audit IDs, and the precise weights that build the Performance score.
How it works
Lighthouse has five categories. Four are scored 0–100: Performance, Accessibility, Best Practices and SEO. PWA was demoted to informational in Lighthouse 12. The Performance score is the one with explicit metric weights: in Lighthouse 10 and 11 it is a weighted average of five lab metrics — Total Blocking Time at 30%, Largest Contentful Paint at 25%, Cumulative Layout Shift at 25%, First Contentful Paint at 10% and Speed Index at 10% — which sum to 100%.
Each metric is first converted to a 0–1 score using a log-normal curve calibrated against real-world data, then multiplied by its weight and summed. Because the curves and throttling differ, the same page scores differently on the mobile and desktop configurations. TBT carries the heaviest weight as the lab stand-in for the field INP responsiveness metric.
Tips and notes
Treat the Accessibility and Best Practices scores as a floor, not a ceiling: automated audits catch only part of what matters, so a perfect 100 still needs manual keyboard and screen-reader testing. When optimizing Performance, attack the heavily weighted metrics first — shaving Total Blocking Time and Largest Contentful Paint moves the score far more than tuning Speed Index.