Google PageSpeed Insights is a free tool that measures how quickly a web page loads and how smoothly it behaves for visitors. It can help identify delays caused by hosting response time, images, scripts, fonts, caching, and other website components.
The report may combine real-user data with a controlled Lighthouse test. These sections measure performance under different conditions.
Table of Contents
How to test a page
- Open Google PageSpeed Insights.
- Enter the complete URL you want to test.
- Select Analyze.
- Wait for the test to finish.
- Review both the Mobile and Desktop tabs.
PageSpeed Insights tests one URL at a time. Testing the homepage does not automatically test product pages, blog posts, forms, or other sections.
For a representative overview, test several important pages, such as the homepage, a landing page, a product or service page, and a page containing forms or large images.
Mobile and desktop results
PageSpeed Insights creates separate reports for mobile and desktop devices.
The mobile test uses more restrictive device and network conditions, so its score is often lower. This does not automatically indicate a hosting problem.
Mobile performance can also be affected by large images, JavaScript, page builders, fonts, advertisements, and third-party services.
Real-user data and lab data
Real-user data
The Discover what your real users are experiencing section uses anonymised data from eligible Chrome visits.
It reflects performance across different devices, locations, networks, and browser conditions.
This section may not appear when the page does not have enough eligible traffic. Its absence does not indicate a website problem.
Lab data
The Diagnose performance issues section is generated by Lighthouse during the current test.
It provides a controlled snapshot and helps identify technical causes of slow loading or unstable behaviour.
Lab results can vary because of network conditions, cache state, server load, and third-party services.
Understanding the performance score
The performance score ranges from 0 to 100:
- 90–100: Good
- 50–89: Needs improvement
- 0–49: Poor
The score combines several Lighthouse metrics. A score of 100 is not required for a website to perform well.
The individual metrics and recommendations usually provide more useful information than the overall score alone.
Important performance metrics
Largest Contentful Paint
Largest Contentful Paint (LCP) measures how long it takes for the largest visible element to appear.
A high LCP can be caused by:
- Slow server response
- Large images
- Render-blocking CSS or JavaScript
- Delayed font loading
- Third-party content
Interaction to Next Paint
Interaction to Next Paint (INP) measures how quickly the page responds visually after a visitor interacts with it.
A high INP is commonly linked to JavaScript tasks that keep the browser busy.
INP is mainly based on real-user interactions, so it may be unavailable when there is not enough field data.
Cumulative Layout Shift
Cumulative Layout Shift (CLS) measures unexpected movement of visible elements while the page loads.
Layout shifts can happen when images have no defined dimensions, fonts change after loading, or banners and embedded content resize their containers.
First Contentful Paint
First Contentful Paint (FCP) measures when the browser first displays visible page content.
It can be affected by server response time, stylesheets, fonts, and JavaScript.
Total Blocking Time
Total Blocking Time (TBT) measures how long the browser is prevented from responding during the Lighthouse test.
Long JavaScript tasks are a common cause of high TBT.
Speed Index
Speed Index measures how quickly the visible parts of the page appear during loading.
It should be reviewed together with the other metrics.
Checking server response time
PageSpeed Insights may show a finding related to the initial document request, server response time, or Time to First Byte.
Time to First Byte (TTFB) measures how long it takes for the browser to receive the first byte of the page response.
TTFB can be affected by:
- Hosting resources
- Application processing
- Database queries
- WordPress plugins or themes
- Page caching
- External API requests
- Network distance
- Redirects
TTFB is useful when evaluating hosting performance, but it does not measure the complete page-loading process.
A server can respond quickly while images or scripts make the page slow. A delayed initial response can also occur even when the rest of the page is well optimised.
Using the findings and diagnostics
Below the metrics, PageSpeed Insights lists findings that may explain the result.
Common findings include:
- Large or poorly optimised images
- Unused JavaScript or CSS
- Render-blocking resources
- Large network payloads
- Long JavaScript tasks
- Third-party scripts
- Inefficient browser caching
- Layout shifts
These are diagnostic suggestions. Resolving every item may not be necessary.
Start with findings that show the largest estimated delay or affect a failing Core Web Vital.
How to compare results accurately
Performance tests are more useful when they are performed consistently.
- Test the same page URL.
- Use the same device tab.
- Run the test several times.
- Compare the median result rather than the fastest test.
- Avoid changing several settings at once.
- Record the active cache configuration, Add-ons, Boost Packs, CPU, and RAM.
- Use the same cache conditions for each comparison.
When testing a resource or caching change, measure the page before and after the change under similar conditions.
A single result may be affected by temporary conditions, so repeated tests provide a more reliable comparison.
What PageSpeed Insights cannot identify
PageSpeed Insights can show performance symptoms and the resources contributing to them. It cannot always identify the exact hosting setting, WordPress plugin, theme function, or external service responsible.
For example, a slow initial server response may be caused by limited resources, an uncached page, a slow database query, or an external API request.
For WordPress websites, PageSpeed Insights can be used together with Query Monitor. PageSpeed Insights tests the visitor experience, while Query Monitor shows what WordPress processes internally.
What to expect after making changes
Lab data can change as soon as the page is tested again.
Real-user data is based on historical visits, so improvements may take time to appear.
Caching can also affect results. The first request after clearing the cache may be slower because the page must be generated again.
Compare repeated results over time rather than treating one score as a permanent measurement.
When to investigate further
Further investigation may be useful when:
- TTFB remains high across repeated tests
- Several pages show a slow server response
- Performance becomes worse during traffic spikes
- Cached pages remain slow
- Results change significantly without a known update
- PageSpeed Insights cannot complete the test
- The website returns an HTTP error
Before contacting support, record the tested URL, date, approximate time, device mode, relevant metrics, and recent website or hosting changes.
Summary
Google PageSpeed Insights combines real-user information with a controlled Lighthouse test. Use the individual metrics, TTFB, and diagnostic findings to understand page performance.
Repeat tests under consistent conditions before comparing website or hosting changes.