Google Lighthouse Explained: What It Is, What It Measures and How to Use It

If you have worked through the audit checklist on this blog, you will have come across references to Google PageSpeed Insights and performance scores throughout. Both are built on the same underlying tool: Google Lighthouse.

Lighthouse is one of the most useful, freely available tools for understanding how well a website is genuinely performing, not how it looks, but how it actually behaves for a real visitor. It is also widely misunderstood, often glanced at briefly for a single score and then ignored, even though it can provide a genuinely detailed diagnostic of what is and is not working on a website.

This article explains what Lighthouse actually is, what each part of its report measures, and how to use it properly rather than just checking the headline number.

What Google Lighthouse Actually Is

Lighthouse is an open-source, automated auditing tool built by Google. It loads a webpage in a simulated environment, runs a series of tests against it, and produces a report covering performance, accessibility, best practices and search engine optimisation. It is free, available to anyone, and requires no technical setup to use in its simplest form.

Lighthouse is built into Google Chrome’s developer tools and is available directly through Google PageSpeed Insights online. It can also be run from the command line or integrated into automated testing pipelines for more advanced use. For most small business owners, the version built into PageSpeed Insights is the most accessible starting point.

Lab data versus field data

It is worth understanding one important distinction early on. Lighthouse itself produces what is called lab data: results from a simulated test run in a controlled environment. PageSpeed Insights also shows field data where available, which is real-world performance data gathered from actual Chrome users who have visited the page, known as the Chrome User Experience Report, or CrUX.

Lab data is useful for diagnosing specific issues in a controlled, repeatable way. Field data reflects what real visitors on a range of devices and connections are actually experiencing. Both are valuable, and a healthy website should look good in both, but they are not always identical; understanding the difference helps avoid confusion when the two do not match exactly.

The Four Categories Lighthouse Measures

Performance

This is the category most people focus on, and it covers loading speed, responsiveness and visual stability, essentially the Core Web Vitals metrics covered in detail in the earlier article on this blog. The performance score is calculated from a weighted combination of several specific metrics, including Largest Contentful Paint, Total Blocking Time and Cumulative Layout Shift.

Accessibility

This category assesses whether a website is usable by people with disabilities, including those who use screen readers or navigate without a mouse. It checks whether images have alt text, whether colour contrast is sufficient, whether form fields have proper labels, and whether the page can be navigated using a keyboard alone.

Best Practices

This category covers a range of general web development standards, including whether the site uses HTTPS, avoids known browser console errors, displays images at an appropriate aspect ratio, and avoids deprecated or potentially insecure code patterns.

SEO

This category checks a set of fundamental, automatable SEO requirements: whether the page has a title tag and meta description, whether links have descriptive text rather than generic phrases like ‘click here’, whether the page is mobile-friendly, and whether search engines can crawl the page. It is worth being clear that a perfect SEO score in Lighthouse covers only the technical basics. It does not assess content quality, backlinks, or the GEO factors covered elsewhere on this blog.

Understanding the Performance Score

The performance score, shown as a number out of 100 with a colour code (red, orange, or green), is the part of the report that gets the most attention and is also the part most likely to be misunderstood.

It is a weighted composite, not a simple average

The performance score is not a straightforward average of the underlying metrics. Different metrics carry different weight in the final calculation, and the weighting has changed over time as Google has refined what it considers most representative of real user experience. This is one reason scores can shift slightly even when nothing on the website has changed.

Scores can vary between runs

Because Lighthouse simulates network and device conditions, and real-world conditions, such as server response time, can vary slightly between tests, it is normal to see small fluctuations in score between consecutive runs of the same page. A single test result should be treated as an indicative snapshot rather than an exact, immutable figure. Running a test three or four times and looking at the general pattern gives a more reliable picture than relying on a single run.

Mobile and desktop are tested separately

As covered in the article on achieving a perfect Core Web Vitals score, mobile and desktop conditions differ, and Lighthouse tests them separately, using different simulated network and device constraints. A site can score very differently on each, and both should be checked rather than assuming a good desktop score means mobile is equally strong.

How to Actually Use a Lighthouse Report

The most common mistake people make with Lighthouse is looking only at the headline scores and ignoring the detailed findings underneath. The real value of the tool is in the specific, itemised list of opportunities and diagnostics it provides below each score.

Running a report

• Visit pagespeed.web.dev and enter your website’s URL
• Wait for the report to generate, which typically takes a short while as the test runs
• Review the four category scores at the top, for both mobile and desktop
• Scroll down to the detailed findings beneath each score

Reading the detailed findings

Beneath each score, Lighthouse lists specific issues it has identified, grouped under headings such as Opportunities and Diagnostics. Each item names a specific issue, such as ‘Eliminate render-blocking resources’ or ‘Properly size images’, and clicking on it expands to show exactly which files or elements are responsible.

This is where the actionable detail lives. A score of 65 on its own tells you that something needs attention. The detailed findings tell you specifically what.

Prioritising what to fix

Lighthouse often estimates the potential time saving from addressing each specific issue, shown in seconds. Issues with larger estimated savings are generally worth prioritising first, though this should be balanced against how straightforward each fix is to implement. A small, simple fix is often worth doing before a larger one with a marginally bigger estimated saving.

Common Issues Lighthouse Flags on Small Business Websites

Certain issues recur across small-business websites, particularly those built on WordPress and with a number of plugins installed over time.

• Unoptimised images: images uploaded at full camera resolution and displayed at a much smaller size, without proper compression
• Render-blocking resources: CSS or JavaScript files that delay the page from displaying any content, often linked to the earlier article on clean CSS
• Excessive JavaScript: plugins or trackers loading scripts that are not genuinely necessary for the page to function
• Missing image dimensions: images without explicit width and height set in the code, causing layout shift as they load
• Poor colour contrast: text and background colour combinations that do not meet accessibility contrast thresholds
• Missing meta descriptions: pages without a meta description, flagged directly in the SEO category
• Non-descriptive link text: links using generic text such as ‘read more’ rather than text that describes the destination

A Practical Checklist for Using Lighthouse

☐  Run a report for your homepage on both mobile and desktop at pagespeed.web.dev

☐  Run reports for two or three other key pages, not just the homepage, since performance can vary across a site

☐  Run each test two or three times to check for consistency before drawing conclusions

☐  Read the detailed findings beneath each score, not just the headline number

☐  Note the estimated time savings for each flagged issue to help prioritise

☐  Address straightforward, high-impact issues first (commonly image optimisation)

☐  Re-test after each round of changes to confirm the fix has had the intended effect

☐  Check the accessibility and best practices scores, not only performance and SEO

Frequently Asked Questions

What is a good Lighthouse score to aim for?

Google generally considers a score of 90 or above good, 50 to 89 needing improvement, and below 50 poor. For most small business websites, a realistic and worthwhile target is 80 or above across categories, with mobile performance being the hardest and most valuable category to push higher, as covered in the earlier Core Web Vitals article on this blog.

Why does my score change every time I run the test?

Some variation between runs is normal, since Lighthouse simulates network and device conditions and real server response times can vary slightly. Significant or wild swings, however, can indicate genuine inconsistency in how the page loads, such as a server under variable load, and are worth investigating rather than dismissing.

Is the Lighthouse SEO score the same as good SEO?

No, and this is an important distinction. The Lighthouse SEO score checks a set of automatable technical basics, such as the presence of a title tag, mobile-friendliness and crawlability. It does not assess content quality, keyword relevance, backlinks, or the GEO factors covered elsewhere on this blog. A perfect Lighthouse SEO score does not mean a page will rank well; it only indicates that it meets the basic technical requirements to be considered.

Can I use Lighthouse to check a competitor’s website?

Yes. Lighthouse and PageSpeed Insights can be run against any public URL, not just your own website. This can be a useful way to understand how your site compares to competitors in terms of performance and technical fundamentals. However, it does not tell you anything about their content quality, backlink profile, or GEO readiness.

Do I need to fix every single issue Lighthouse flags?

Not necessarily. Some flagged issues have a very small estimated impact and may not be worth the development time required to address them, particularly on a smaller website with a limited budget. The goal is meaningful improvement in genuine user experience and search performance, not a perfect 100 score for its own sake. However, as covered in the earlier case study on this blog, a perfect score is achievable and does offer real benefits when achieved properly.

Want Help Interpreting Your Own Lighthouse Report?

If you have run a Lighthouse report and are unsure what the findings actually mean for your website, or which issues are worth prioritising, get in touch for a straightforward conversation with no obligation and no sales pitch.