Google Lighthouse
Google Lighthouse is an open source, automated tool from Google for assessing the quality of web pages. It checks performance, accessibility and SEO, among other things, and presents the results in a report with scores from 0 to 100. Lighthouse runs in Chrome DevTools, from the command line, as a Node module and in PageSpeed Insights. The performance score comes from a simulation of page loading, not from real-user data. The tool is updated regularly: version 13.6.0 from 27 October 2026 changed, among other things, how loading is simulated.
What is Google Lighthouse?
Google Lighthouse is an open source tool that runs a series of tests on a web page and generates a report with the results. According to the Chrome for Developers documentation, you can run it on any page, public or requiring authentication.
Each failed test, called an audit, points to something you can improve. For every audit, the documentation explains why it matters and how to fix the issue.
| Feature | Description |
|---|---|
| Creator | Google, Chrome team |
| Type | automated web page auditing tool |
| License | open source, code in the GoogleChrome/lighthouse repository on GitHub |
| Areas checked | performance, accessibility, best practices, SEO |
| Score range | 0–100 in each category |
| Performance data source | simulated page load (lab data) |
| Extensions | Stack packs, Lighthouse plugins, Lighthouse CI |
Lighthouse is extensible. Stack packs detect the site's technology, such as a CMS, and give recommendations for that platform. Plugins let you build your own audits as new report categories. Lighthouse CI helps catch score regressions across deployments.
What does Lighthouse measure?
Lighthouse rates a page in several categories, each scored from 0 to 100. Performance is the category discussed most often.
The performance score is a weighted average of several metrics. The latest weighting table in the Chrome for Developers documentation applies to Lighthouse 10.
| Metric | What it measures | Weight in performance score |
|---|---|---|
| First Contentful Paint (FCP) | when the first content appears | 10% |
| Speed Index | how quickly the page visibly fills in | 10% |
| Largest Contentful Paint (LCP) | when the largest content element appears | 25% |
| Total Blocking Time (TBT) | how long the main thread is blocked | 30% |
| Cumulative Layout Shift (CLS) | how much the page layout shifts | 25% |
According to the documentation, the scoring curve is based on real-website data from HTTP Archive. The 25th percentile of that data maps to a score of 50, and the 8th percentile to a score of 90.
| Score range | Color | Rating |
|---|---|---|
| 0–49 | red | poor |
| 50–89 | orange | needs improvement |
| 90–100 | green | good |
The documentation notes that around a score of 96 the simple relationship between a metric and the score ends. Raising the score further takes increasingly large improvements in the metric itself.
How does the simulation work and where is Lighthouse used?
By default, Lighthouse uses simulated throttling. It first loads the page without restrictions, then simulates how it would load on a slower connection and a less powerful device.
According to the Lighthouse documentation on GitHub, this method is fast and repeatable. It does, however, have inherent inaccuracy, because it predicts a page load that was not measured directly. The simulation is handled by the Lantern engine.
| Default setting | Value according to the Lighthouse documentation |
|---|---|
| Method | simulated throttling |
| Network profile | "Slow 4G", formerly called "Fast 3G" |
| CPU slowdown | fixed 4× multiplier |
| Consistency | the same settings as PageSpeed Insights and the default CLI |
According to the documentation, the 4× multiplier is meant to move a typical result from a powerful desktop to the class of a mid-tier phone. The "Slow 4G" profile roughly matches the slowest quarter of 4G connections and the fastest quarter of 3G connections.
| How to run it | Best for |
|---|---|
| Chrome DevTools, Lighthouse panel | quick tests in the browser, including pages that require login |
| Command line (npm install -g lighthouse) | automating tests with scripts |
| Node module | integration with continuous integration systems |
| PageSpeed Insights | testing in the browser without installing anything |
| Chrome extension | testing the current tab |
Next to Lighthouse results, PageSpeed Insights also shows field data from the CrUX report. According to Google's documentation, the two can differ, because lab data comes from a single device and fixed network conditions.
How is Lighthouse updated?
Lighthouse is developed as an open source project and released in successive versions on GitHub. New versions reach Chrome DevTools with new browser releases, and PageSpeed Insights.
Updates can change a page's scores without any change to the page itself. This happens when the simulation, the metric weights or the set of audits change. The performance scoring documentation, for example, shows different metric weights in Lighthouse 8 and Lighthouse 10.
| Version | Weight or metric change according to the documentation |
|---|---|
| Lighthouse 8 | Time to Interactive 10%, CLS 15% |
| Lighthouse 10 | Time to Interactive removed from the score, CLS 25% |
One example of a simulation update is version 13.6.0 from 27 October 2026. According to the release notes, it improved the accuracy of the Lantern engine. Lantern now uses, among other things, the render-blocking status reported by Chrome and calculates text LCP differently. The notes said the version would reach PageSpeed Insights within a week and DevTools in Chrome 159.
The same version updated the Agentic Resource Discovery audit to the ARD v0.91 specification. It also added auditing multiple URLs with a single command in the CLI.
When comparing scores over time, check the Lighthouse version. Every PageSpeed Insights report shows it next to the emulated device information.
Frequently asked questions
Q.How is Lighthouse different from PageSpeed Insights?
Lighthouse is the auditing tool itself, while PageSpeed Insights is a Google service that runs Lighthouse in the browser. PageSpeed Insights also shows field data from real users from the CrUX report. Lighthouse can also be run locally, for example in Chrome DevTools, from the command line or as a Node module.
Q.Why does the Lighthouse score change with every test?
The score can change because it depends on test conditions and the tool version. According to the Lighthouse documentation, the default simulation has inherent inaccuracy because it predicts how the page loads. Scores also change after Lighthouse updates that alter the simulation or the metric weights. That is why it is worth comparing reports from the same version.
Q.How is the Lighthouse performance score calculated?
The performance score is a weighted average of five metrics. According to the documentation for Lighthouse 10, Total Blocking Time has the largest weight, 30%. Largest Contentful Paint and Cumulative Layout Shift each have 25%, and First Contentful Paint and Speed Index 10% each. A score of 90–100 means good performance.
Q.Does the Lighthouse score affect rankings in Google?
The Lighthouse score itself is lab data from a simulation, not data from real users. The Core Web Vitals report in Google Search Console is based on field data from users. According to Google's documentation, lab and field data can differ. Lighthouse is therefore mainly a tool for diagnosing and improving a page.
Let's talk growth
Send a brief or drop your contact. We'll reply within 24h.
Get a quote