What is Lighthouse?

Lighthouse, explained

Chrome's documentation describes Lighthouse as an open-source, automated tool for improving the quality of web pages. You can run it in Chrome DevTools, from the command line, as a Node module in CI, or through PageSpeed Insights. Because DevTools runs locally, it works on pages behind a login and on localhost, which PSI can't reach.

The performance score is a weighted average of lab metrics. In Lighthouse 10 the weights are Total Blocking Time 30%, Largest Contentful Paint 25%, Cumulative Layout Shift 25%, First Contentful Paint 10% and Speed Index 10%. Total Blocking Time stands in for responsiveness, because a lab run has no real user to measure Interaction to Next Paint. Scores of 90–100 are good, 50–89 need improvement and 0–49 are poor.

Scores vary between runs. Lighthouse's docs list reasons such as A/B tests or ads, network routing, testing on different hardware, browser extensions and antivirus software. Run it several times, in an incognito window without extensions, and compare ranges rather than single numbers.

The SEO category checks basics like a title, meta description, crawlable links, image alt text, valid hreflang and a successful status code. A perfect SEO score only means those basics exist; it says nothing about content or rankings. Treat Lighthouse as a checklist and debugging tool, and use field data to judge real performance.

Why it matters for founders

Lighthouse is free, built into Chrome and good at pointing to specific fixes. It's also easy to misuse as a vanity score.

Example

Before a launch, you run Lighthouse on your new landing page in an incognito window. It flags images without dimensions and a missing meta description, which you fix in ten minutes.

Common mistakes

• Running it with browser extensions enabled.
• Treating one run's score as definitive.
• Reading a perfect SEO score as a ranking guarantee.
• Optimizing lab scores while field data still fails.