A free tool from Google that measures how fast a web page loads and how well it provides a smooth experience.
Website developers reading about web performance metrics and Core Web Vitals.
01What it is and how it works
PSI draws from two separate data sources. Field data comes from the Chrome User Experience Report, which collects measurements from actual Chrome users across millions of websites over a rolling 28-day window. This tells you what your real visitors are experiencing right now. Lab data comes from Lighthouse, an automated test that loads the page in a controlled browser with set network and device conditions. This gives you consistent, repeatable results you can use to debug specific issues.
When you enter a URL, PSI runs Lighthouse and pulls the latest CrUX data, then combines them into one report. It evaluates your page against the Core Web Vitals metrics: Largest Contentful Paint (how quickly the main content appears), First Input Delay (how quickly the page becomes interactive), and Cumulative Layout Shift (whether the page jumps around as it loads). It also checks metrics like Time to First Byte, render-blocking resources, and image optimization. Scores range from 0 to 100, with green at 90 or above, orange in the 50 to 89 range, and red below 50.
PSI is Googles tool for testing how quickly pages load. It shows you both what real visitors experience and what a controlled test finds, along with specific steps to make the page faster.
A retailer enters their product page URL. PSI shows field data of LCP at the 75th percentile of 3.2 seconds, rated poor, and a Lighthouse lab result of 2.8 seconds. Both flag the same hero image as oversized. The report also shows a layout shift of 0.14 from a lazy-loaded banner, rated needs improvement.
02What to do about it
Start by fixing the items PSI marks as failed or needs improvement in the Diagnostics section. These are ordered by estimated impact, so tackle the top ones first.
Image optimization is the most common win. Compress images with modern formats like WebP or AVIF, set explicit width and height attributes to prevent layout shift, and use responsive images with srcset so mobile users download smaller files. Lazy load images that appear below the fold.
For server response time, consider upgrading hosting, enabling compression, using a CDN, or implementing caching headers. If third-party scripts are blocking rendering, load them asynchronously or defer them until after the main content renders.
Remove any CSS or JavaScript the page does not actually use. For render-blocking CSS, inline critical styles and load the rest asynchronously. For JavaScript, defer parsing or break long tasks into smaller chunks.
Rerun PSI after each batch of changes to confirm improvement. The field data updates on a rolling 28-day cycle, so measurable gains in real user experience take time to appear even after you ship the code.
- Compress and resize images before uploading
- Set explicit dimensions on all image elements
- Defer non-critical JavaScript
- Enable browser caching headers
- Move third-party scripts out of the critical path
03How it is measured or noticed
The first thing to look at is the score at the top of the report. It is a weighted average that reflects how well your page performs across all the measured metrics. A score above 90 is good; below 50 is poor.
Below the score, PSI shows your Core Web Vitals status as pass, needs improvement, or poor for both mobile and desktop. These are the metrics Google uses for ranking, so they matter more than the raw score.
The field data section shows percentile distributions. Look at the 75th percentile for mobile, because Google uses this as the benchmark for page experience. If the 75th percentile is in the good range, your page passes that metric.
The Opportunities and Diagnostics sections list specific issues with estimated savings in seconds. Each item expands to show the affected resources and links to documentation explaining the technique.
To spot changes over time, use the PSI URL with the &str=1 parameter to include historical CrUX data, or compare snapshots manually by saving reports. Google Search Console also surfaces Core Web Vitals data for your entire origin, grouped by URL pattern.
04Common mistakes
- Chasing a perfect score instead of fixing failed Core Web Vitals — a score of 99 with a poor LCP still hurts ranking
05Limits and what it is often confused with
PSI measures a single URL at a moment in time. It does not track performance across sessions, and it does not capture interactions that happen after the initial page load unless those interactions block the main thread during the measurement window.
PSI is not a substitute for real user monitoring. Field data reflects what happened over the past 28 days and cannot tell you what caused a specific regression. If your CrUX score drops, you need application performance monitoring or custom dashboards to find the culprit.
PSI is often confused with Lighthouse as a standalone product. Lighthouse runs inside PSI, but you can also run it in Chrome DevTools, as a Node module, or from the command line. Those versions may produce slightly different results because they run in different environments.
PSI also does not test behind a login wall, single-page applications that load content dynamically after the initial shell renders, or performance under authenticated user conditions. For those cases, use custom lab tests with tools like WebPageTest or Web Vitals JavaScript library in your own monitoring setup.
06Worked example
An editor runs PSI on their homepage and sees a score of 62 on mobile, with LCP at 4.8 seconds rated poor and Cumulative Layout Shift at 0.18 rated poor. The Opportunities section flags an unoptimized hero image contributing 3.2 seconds of load time and three render-blocking stylesheets. The Diagnostics section notes that 420 kilobytes of unused CSS is loaded on every page view. The editor compresses the hero image to WebP, inlines critical CSS for the above-the-fold content, and defers the remaining stylesheets. A follow-up PSI run shows the score improved to 81, with LCP at 2.3 seconds rated needs improvement and layout shift at 0.09 rated needs improvement. The page still fails one Core Web Vitals threshold, but traffic from mobile devices increased noticeably within two weeks, before the CrUX data window refreshed.
Frequently asked questions
How does PageSpeed Insights combine real user data with lab tests?
PageSpeed Insights merges data from real Chrome users (Chrome User Experience Report) with synthetic lab tests (Lighthouse) to provide both field and lab performance metrics. This dual approach ensures insights reflect actual user experiences while identifying technical issues through automated testing.
What does the PSI score at the top of the report indicate?
The score represents the overall performance rating based on Core Web Vitals and other metrics, with higher values indicating better user experience. It combines lab and field data to prioritize actionable recommendations for improvement.
Why should I focus on the Diagnostics section in PSI?
The Diagnostics section lists specific issues that need fixing, such as render-blocking resources or image optimization opportunities. Addressing these items directly improves load times and user experience, as they are prioritized based on their impact.
How is PageSpeed Insights different from Lighthouse?
Lighthouse is a standalone lab tool that runs synthetic tests, while PSI integrates Lighthouse data with real-world user metrics from the Chrome User Experience Report. PSI provides a broader view by combining lab and field data for a complete performance assessment.
What are common mistakes when using PageSpeed Insights?
Users often ignore mobile performance, overlook the Diagnostics section, or focus solely on the score without addressing specific issues. They may also confuse PSI with other tools like Lighthouse or assume a high score guarantees optimal performance across all devices.
Does PageSpeed Insights measure multiple pages or just one URL?
PSI evaluates a single URL at a time, not entire sites or multiple pages. This limitation means it cannot aggregate performance across a site, so testing individual key pages is essential for a comprehensive analysis.
How long does it take for PSI results to reflect changes after fixing issues?
Lab test results update immediately after fixes are implemented, but field data (Core Web Vitals) may take weeks to reflect improvements, as it relies on real user experiences accumulating over time.
Asked out loud
spoken, not typedThe same term in the words somebody uses speaking to an assistant rather than typing into a box — written from the situation, which is why each one carries the situation it came from.
Focus on the Diagnostics section first, as it highlights the most impactful issues like image optimization, render-blocking resources, and server response times. These fixes can be implemented quickly and will show immediate improvements in lab test results.
Use the mobile-friendly test version of PSI to ensure you're seeing mobile-specific metrics. Focus on the Core Web Vitals scores for Largest Contentful Paint and Cumulative Layout Shift, as these directly impact mobile user experience.
Slow load times detected by PSI often correlate with higher bounce rates, especially on mobile devices. Check the Diagnostics section for issues like unoptimized images or slow server response times, which can significantly degrade user experience.