Website speed affects user experience, conversions and, indirectly, search visibility. Google PageSpeed Insights (PSI) is the free tool most people use to check it. But the numbers it gives can be confusing: a score out of 100, several metrics with abbreviations, and two different kinds of data.
In this guide we explain what PageSpeed Insights is, how it works, the difference between field and lab data, how to use it and read the score, what Core Web Vitals and the other metrics mean, what to do when data is not found, how to read recommendations, how to improve images, JavaScript, CSS, fonts and caching, the role of CDN, hosting and third-party code, and a four-week improvement plan. For the direct link to related topics, see technical SEO and, for a faster page format debate, what is Google AMP.
What Is Google PageSpeed Insights?
A free Google tool that analyzes the performance of a web page on mobile and desktop and gives recommendations for improvement. It combines real-user data (where available) with lab testing.
What Is PageSpeed Insights Used For?
To measure how fast and stable a page feels, to find the causes of slowness and to prioritize fixes.
Why Is Website Speed Important?
Slow pages frustrate visitors, lower conversions and can hurt search performance. Speed matters especially on mobile; see mobile SEO and responsive web design.
How Does PageSpeed Insights Work?
Real User Experience
It shows data from real Chrome users (field data) when your page has enough traffic.
Diagnosing Performance Problems
It runs a controlled test (lab data) with Lighthouse to point to specific issues.
What Is the Difference Between Field Data and Lab Data?
| Aspect | Field data | Lab data |
|---|---|---|
| Source | Real users over the past period | A simulated test now |
| Use | What users actually experience | Debugging and diagnosing |
| Variation | Reflects real devices and networks | Fixed conditions, repeatable |
Field data is the better indicator of what users experience, and lab data is better for finding and testing fixes.
How to Use PageSpeed Insights
- Decide which URL to analyze.
- Enter the URL in the analysis box.
- Start the analysis.
- Review mobile and desktop results separately.
What Do PageSpeed Insights Scores Mean?
The performance score from 0 to 100 summarizes lab metrics. In Lighthouse’s classification, 90 and above is considered good, 50-89 needs improvement and below 50 is poor.
Why Does the PageSpeed Score Change?
Test conditions, network, server load, dynamic content and third-party scripts can all make results vary between runs.
Does the PageSpeed Score Have to Be 100?
No. A perfect score isn’t necessary. Focus on the Core Web Vitals and on the real experience of your users.
What Are Core Web Vitals?
Three metrics Google uses to describe user experience:
Largest Contentful Paint (LCP)
How long it takes for the main content to appear. See Largest Contentful Paint.
- Why is LCP high? Large images, slow servers, render-blocking resources.
- How to improve it: optimize the main image, improve server response, reduce blocking CSS/JS, use caching and a CDN.
Interaction to Next Paint (INP)
How quickly the page responds to user interactions.
- Why is INP high? Heavy JavaScript, long tasks on the main thread.
- How to improve it: reduce and split JavaScript, remove unused code, defer non-critical scripts.
Cumulative Layout Shift (CLS)
How much the layout jumps during loading.
- Why is CLS high? Images without dimensions, late-loading ads and embeds, web fonts swapping.
- How to improve it: set width and height on images, reserve space for dynamic content, optimize fonts.
What Are the Other Metrics?
- First Contentful Paint (FCP): when the first content appears.
- Speed Index: how quickly content is visually displayed.
- Total Blocking Time (TBT): how long the main thread is blocked.
- Time to First Byte (TTFB): how quickly the server starts responding.
URL-Level Data vs. Origin-Level Data
URL-level data is for the specific page; origin-level data aggregates all pages of the site. A single page may have too little traffic for its own data, in which case the origin data is shown.
What Does “No Data Found” Mean in PageSpeed Insights?
It means there aren’t enough real-user samples for that URL or site. Rely on lab data and track the field data as traffic grows.
Why Are Mobile and Desktop Scores Different?
Mobile tests simulate slower devices and networks, so scores are usually lower. Big images, heavy JavaScript and desktop-first resources hurt mobile most.
How to Read the Recommendations
Opportunities
Changes that could improve load time, with estimated savings.
Diagnostics
Additional information about the page’s performance.
Passed Audits
Things already done well.
Should Recommendations Be Prioritized?
Yes. Start with changes with the biggest effect for the least work, and focus on the metrics that fail.
How to Optimize Images
Compress and resize images, serve them at the displayed size, use modern formats and set dimensions. You can use the image compression and image format converter tools, and read SEO optimization for images.
Should WebP and AVIF Be Used?
Yes, where supported; they are usually smaller than JPEG and PNG at similar quality, with a fallback if needed.
What Is Lazy Loading?
Loading images and iframes only when they are about to appear. Don’t lazy-load the main (LCP) image.
How to Improve JavaScript Performance
Remove unused code, split bundles, defer or delay non-critical scripts and avoid long tasks.
How to Improve CSS Performance
Reduce unused CSS, inline critical CSS and avoid render-blocking stylesheets.
How to Optimize Web Fonts
Limit the number of fonts and weights, use modern formats, preload critical fonts and use font-display so text shows immediately.
Why Is Caching Important?
Caching lets browsers and servers reuse files, so repeat visits and page loads are much faster.
Does a CDN Raise the PageSpeed Score?
It can, by serving files from locations closer to users. See what is a CDN.
Does Hosting Affect PageSpeed Results?
Yes. Slow servers raise TTFB and LCP. See what is hosting.
Why Do Third-Party Codes Cause Problems?
Analytics, ads, chat widgets and embeds add requests and JavaScript. Load only what you need and defer the rest.
DOM Size and Other Issues
Very large, deeply nested pages are heavy to render. Simplify the structure. On WordPress, review plugins; see WordPress SEO plugins.
Monitoring Speed with Search Console
Google Search Console has a Core Web Vitals report grouping URLs by status, which is useful for finding template-level issues. Combine with an SEO analysis and on-site work; see on-site SEO optimization.
A Four-Week PageSpeed Improvement Plan
Week 1: Measure.
- Test the home page
- Review service and blog pages
- Analyze product and category templates
- Record mobile and desktop results
- Note real-user data
Week 2: Server and images.
- Optimize large images
- Identify LCP elements
- Check hosting and TTFB
- Set up caching
- Consider a CDN
- Reduce redirects
Week 3: Code.
- Remove unused JavaScript
- Review third-party code
- Clean up CSS files
- Optimize fonts
- Reduce long main-thread tasks
Week 4: Stability and testing.
- Add dimensions to images
- Fix dynamic areas to avoid layout shift
- Test on real mobile devices
- Check forms and menus
- Test conversion steps
Frequently Asked Questions
Is Google PageSpeed Insights free?
Yes. The tool can be used without paying anything. You don’t have to create an account to analyze.
How do I use PageSpeed Insights?
Enter the URL of the page to analyze and start the test. Review the mobile and desktop results separately and evaluate the metrics and the optimization recommendations.
What should the PageSpeed score be?
In Lighthouse’s classification, 90 and above is considered good. But you should focus on Core Web Vitals and the real user experience, not only the score.
Why does the PageSpeed score keep changing?
The test environment, network conditions, server load, dynamic content and third-party scripts can cause results to change.
Why is the mobile PageSpeed score low?
The mobile test can represent more limited device and network conditions. Large images, heavy JavaScript and resources prepared for desktop can lower the mobile score.
