Google Lighthouse: Complete Website Audit & SEO Guide
Google Lighthouse is an open-source automated auditing tool for testing the quality of webpages. It can analyse performance, accessibility, web best practices, basic SEO fundamentals and emerging machine-interaction readiness. Lighthouse can be run directly in Chrome DevTools or integrated into development and continuous-testing workflows. This guide forms part of our wider Google playbook.
What Is Google Lighthouse?
Lighthouse runs a series of automated audits against a page and produces a report showing which tests passed, which failed and which areas deserve further investigation.
Automated Website Auditing
Lighthouse checks measurable technical conditions in a reproducible way. Each failed audit can act as a diagnostic starting point rather than forcing developers to investigate a page blindly.
Public, Local & Authenticated Pages
The Chrome DevTools workflow can be used on public websites, local development pages and pages that require authentication, making Lighthouse useful before as well as after deployment.
Developer & SEO Workflows
Lighthouse can be run manually during diagnosis or integrated into development workflows so regressions can be detected before poor code reaches production.
Four Ways to Run a Lighthouse Audit
The best workflow depends on whether you are analysing one page manually or building repeatable performance and quality checks into development.
Chrome DevTools
Open the page in Chrome, launch DevTools, select the Lighthouse panel, configure the audit options and generate a report. This is normally the easiest workflow for individual page diagnosis.
Command Line
Developers can install Lighthouse through Node and run audits from the terminal, making repeatable tests and scripted workflows possible.
npm install -g lighthouselighthouse https://example.com
Node Module
Lighthouse can be used programmatically as a Node module, allowing custom systems and continuous-integration pipelines to run and process audit results automatically.
Lighthouse CI
Lighthouse CI can help teams detect regressions by running audits repeatedly as part of development, testing and deployment processes rather than waiting for problems to appear after launch.
What Does Lighthouse Test?
Each category answers a different website-quality question. High scores in one category do not compensate for serious problems in another.
Performance
Measures the page in a controlled lab environment using loading, rendering, blocking and visual-stability metrics.
Accessibility
Checks automated accessibility conditions including names, labels, ARIA relationships, contrast and document semantics.
Best Practices
Tests browser, security, compatibility, user-experience and modern web-development practices.
SEO
Checks technical fundamentals that can affect search-engine understanding, crawlability, indexing and mobile usability.
Agentic Browsing
Experimental audits evaluate how reliably machine agents can discover, understand and interact with page functionality.
How Lighthouse Measures Performance
The Performance category is a controlled laboratory assessment rather than a record of what every real visitor experienced.
Performance Is a Weighted Metric Score
The overall Lighthouse Performance score is calculated from its scored performance metrics. Opportunities and Diagnostics normally do not add or subtract points directly; instead, fixing the problems they identify can improve the metrics that feed the score.
For real-user Chrome performance data and Core Web Vitals, use Google PageSpeed Insights alongside Lighthouse rather than treating one simulated run as evidence of every visitor's experience.
How to Read the Lighthouse Performance Score
The score is useful for diagnosis and regression testing, but the individual metrics usually provide more actionable information than the headline number.
Good
A strong result under the conditions used for the current Lighthouse test. A perfect 100 is not necessary and can require disproportionate effort once performance is already excellent.
Needs Improvement
Inspect the individual metrics, Insights and Diagnostics to determine which resources or tasks are creating the largest user-perceived performance cost.
Poor
The laboratory test indicates substantial performance problems. Prioritise large bottlenecks before trying to polish minor audit warnings.
What Lighthouse Accessibility Testing Checks
Lighthouse can identify many common accessibility failures automatically, but automated testing covers only part of a complete accessibility review.
Automated Accessibility Signals
The Accessibility score is calculated from weighted pass/fail audits. The weighting reflects the likely user impact of the accessibility issue rather than simply counting every failed check equally.
What the Lighthouse Best Practices Category Covers
Best Practices looks beyond speed to identify technical patterns that can affect security, browser behaviour, reliability and user experience.
Security
Audits can identify selected insecure or risky browser practices and front-end conditions that may expose users or applications to avoidable security problems.
Browser Errors
Console errors and problematic resource behaviour can indicate broken code or unexpected conditions that deserve investigation.
Deprecated Technologies
Lighthouse can warn about selected outdated browser APIs and practices that should be modernised before support becomes unreliable.
User Experience
Best Practices also includes selected technical checks designed to encourage reliable, secure and predictable interaction with modern webpages.
What Does the Lighthouse SEO Score Actually Mean?
The SEO category checks technical fundamentals that automated software can reasonably verify. It is not a complete organic-search assessment.
A Lighthouse SEO Score of 100 Does Not Mean “Perfect SEO”
Lighthouse can check whether selected technical conditions help search engines access and understand the page, but it cannot determine whether your keyword strategy is correct, whether the content fully satisfies intent or whether you have enough authority to compete.
For that reason, Lighthouse should sit inside a broader SEO audit that examines indexing, search performance, competitors, content, architecture, links and business goals.
Use Google Search Console to investigate what Google actually crawls and indexes, and use the Rich Results Test for supported structured-data eligibility.
Lighthouse Agentic Browsing Audits
Lighthouse is beginning to evaluate whether websites are constructed in ways that allow software agents to discover, understand and interact with page functionality reliably.
Website Quality Is Expanding Beyond Human Browsing
The experimental Agentic Browsing category evaluates machine-interaction readiness using deterministic audits. Current testing requires Chrome 150 or later, and individual WebMCP functionality may depend on experimental browser support.
Unlike the traditional Lighthouse categories, Agentic Browsing does not currently generate a conventional weighted score from 0 to 100. The report focuses on fractional progress, pass/fail results, warnings and actionable signals while standards continue to develop.
Why No 0–100 Score?
Machine-interaction standards are still emerging. A single definitive “agent readiness score” would imply more certainty than the underlying standards currently support, so Lighthouse exposes individual signals instead.
Accessibility Tree
Agents can use the accessibility tree as a machine-readable model of controls, labels, roles and relationships. Missing programmatic names can therefore hurt both accessibility and agent interaction.
Layout Stability
Agents that rely on screenshots or element coordinates can fail when buttons, forms or other controls move unexpectedly between identification and interaction.
WebMCP
Experimental WebMCP audits examine whether suitable website functionality is exposed to agents through emerging machine-interaction mechanisms.
Discoverability
The category includes machine-discoverability signals designed to help agents understand what functionality or information is available on a site.
Why Lighthouse Results Can Change
A Lighthouse audit is a laboratory test. Changes in the tested page, environment or browser can change the result even when the website has not been deliberately redesigned.
Device & Environment
Testing under different hardware or desktop/mobile conditions can produce different performance results because the scoring and simulated environment are not identical.
Dynamic Content
A/B tests, advertisements, personalisation and other dynamically selected resources can change what the browser loads during an audit.
Browser Extensions
Extensions can inject JavaScript or create additional requests, making a local DevTools audit less representative of a clean browser session.
Underlying Conditions
Routing, server performance and third-party responses can vary between runs. Judge persistent patterns rather than reacting to a single isolated score.
Use Lighthouse Before Problems Reach Production
Lighthouse becomes more powerful when it moves from an occasional audit tool into a repeatable development-quality process.
Test During Development
Use Chrome DevTools on local or authenticated environments to identify problems before launch rather than discovering them only after Search Console or users report the consequences.
Detect Regressions
Lighthouse CI can make repeated audits part of the development pipeline. Teams can detect when a new template, script or dependency damages an important quality metric.
Compare Against a Baseline
A stable baseline can be more useful than obsessing over universal perfect scores. The important question is whether a change made the target page or template materially worse.
A Better Lighthouse Audit Process
Use Lighthouse as a diagnostic system rather than trying to make every line of the report green without understanding its importance.
Open the Correct Page
Test the actual template or URL where the suspected problem exists.
Choose Audit Categories
Select the quality areas relevant to the current investigation.
Run the Audit
Generate the report under a controlled environment whenever possible.
Inspect Failed Audits
Understand why each important audit failed before attempting a fix.
Fix the Root Cause
Prioritise meaningful user, accessibility, security and SEO problems.
Retest & Monitor
Run the audit again and validate production behaviour with appropriate real-user or search tools.
Lighthouse vs PageSpeed Insights vs Search Console
The tools overlap, but each one operates at a different level of website analysis.
Google Lighthouse
Best for controlled automated auditing across Performance, Accessibility, Best Practices, SEO and emerging Agentic Browsing checks. It can be used manually or integrated into development workflows.
PageSpeed Insights
Best for page-performance analysis combining Lighthouse lab diagnostics with available real-user Chrome performance data. Read our PageSpeed Insights guide.
Google Search Console
Best for Google Search performance, crawling, indexing, Core Web Vitals URL groups and other property-level search diagnostics. See our Search Console guide.
What Lighthouse Cannot Tell You About SEO
A technically clean Lighthouse report can still belong to a page that has almost no chance of ranking for its target search intent.
Automated SEO Audits Are Only One Layer
Lighthouse can verify selected technical fundamentals, but it cannot tell you whether you chose the correct keyword, whether your article is more useful than competitors or whether your site has enough authority to compete.
Use Lighthouse alongside on-page SEO, website architecture, ranking-factor analysis and broader competitive research.
For an additional automated overview, you can also use the GuestPost.UK SEO Checker, while the SEO tools hub covers specialist platforms for deeper analysis.
Common Google Lighthouse Mistakes
The most common problem is treating a Lighthouse score as an objective rather than using the audits to improve the actual website.
Chasing Perfect Scores
Once a page performs well, moving from a very high score to 100 may require disproportionate work with little meaningful improvement for visitors.
Thinking SEO 100 Means Perfect SEO
Lighthouse only checks selected technical fundamentals. Keyword strategy, content usefulness, links, authority, competition and intent remain outside its scope.
Thinking Accessibility 100 Is Enough
Automated accessibility testing cannot replace keyboard, screen-reader and human usability testing.
Trusting One Performance Run
Performance scores can fluctuate. Look for repeatable problems and compare several controlled tests rather than reacting to one isolated result.
Fixing Every Audit Equally
Not every warning has the same business or user impact. Prioritise critical failures and the bottlenecks connected to real user problems.
Confusing Lab With Field Performance
Lighthouse is a controlled lab audit. For real-user Core Web Vitals, use CrUX-based tools such as PageSpeed Insights and Search Console.
Common Questions About Google Lighthouse
Quick answers covering Lighthouse scoring, SEO, accessibility, PageSpeed Insights, DevTools and Agentic Browsing.
01 What is Google Lighthouse?
Google Lighthouse is an open-source automated webpage auditing tool that can analyse performance, accessibility, web best practices, SEO fundamentals and additional website-quality signals.
02 How do I run Lighthouse in Chrome?
Open the page in Chrome, launch Chrome DevTools, select the Lighthouse panel, choose the required options and run the audit.
03 What is a good Lighthouse Performance score?
Lighthouse classifies 90–100 as Good, 50–89 as Needs Improvement and 0–49 as Poor. The individual metrics and failed audits are normally more useful for diagnosis than the headline score alone.
04 Is Lighthouse the same as PageSpeed Insights?
No. PageSpeed Insights uses Lighthouse for laboratory performance analysis, but PSI also includes available real-user Chrome performance data. Lighthouse is a broader automated audit framework. See our PageSpeed Insights guide.
05 Does Lighthouse test accessibility?
Yes. Lighthouse runs automated accessibility audits and calculates an Accessibility score. However, automated testing cannot confirm complete accessibility, so manual testing remains necessary.
06 Does a Lighthouse SEO score of 100 mean my SEO is perfect?
No. Lighthouse checks only selected technical SEO fundamentals. It does not fully assess keywords, search intent, content quality, backlinks, competition, topical authority or your complete indexing strategy.
07 Can Lighthouse test a local website?
Yes. Running Lighthouse through Chrome DevTools is useful for testing local development environments as well as public and authenticated pages.
08 Why does my Lighthouse score change?
Performance results can vary due to the test environment, hardware, browser extensions, dynamic content, advertising, network routing, server behaviour and third-party resources.
09 What is Lighthouse Agentic Browsing?
Agentic Browsing is an experimental Lighthouse category that evaluates signals related to machine interaction, including accessibility-tree quality, layout stability, discoverability and emerging WebMCP integration.
10 Can Lighthouse be automated?
Yes. Lighthouse can be run from the command line, used as a Node module and integrated into continuous-development workflows through Lighthouse CI.
Use Lighthouse to Find Problems — Not to Chase Perfect Numbers
Lighthouse is most useful when its audit categories are treated as diagnostic systems rather than badges. Performance identifies laboratory bottlenecks, Accessibility catches many machine-detectable barriers, Best Practices highlights technical risks, SEO confirms selected search fundamentals and the emerging Agentic Browsing category begins to examine how machine agents interact with websites. Run the right audit, understand why it failed, fix the root cause and then validate the result with the specialist tool that matches the problem.
