Google PageSpeed Insights: Complete Core Web Vitals Guide
Google PageSpeed Insights combines real-user performance information with a simulated Lighthouse test to help diagnose how a webpage loads, responds and remains visually stable. It is one of Google's most useful performance-analysis tools, but its scores are frequently misunderstood. This guide explains PageSpeed Insights, Core Web Vitals, CrUX, Lighthouse diagnostics and the practical fixes that matter, as part of our wider Google playbook.
What Is Google PageSpeed Insights?
PageSpeed Insights analyses a webpage using two complementary sources of performance evidence: real-user field data where available and Lighthouse laboratory diagnostics.
Real-User Experience
Where sufficient Chrome User Experience Report data exists, PSI shows how real Chrome users have experienced the page or its origin over the recent reporting window.
Simulated Testing
PSI also runs Lighthouse in a controlled environment. This provides an immediate diagnostic snapshot that can reveal bottlenecks even when a page has insufficient real-user data.
Performance Diagnostics
The report identifies technical opportunities and diagnostics related to images, JavaScript, CSS, server response, rendering and other resources that can affect page performance.
Field Data vs Lab Data in PageSpeed Insights
Understanding this difference explains why the green Core Web Vitals assessment and the Lighthouse performance score can appear to disagree.
Field Data
Field data represents real-user experiences collected through the Chrome User Experience Report, commonly abbreviated to CrUX. PageSpeed Insights reports this data over a trailing 28-day period.
- Based on eligible real Chrome user experiences.
- Represents actual devices and network conditions.
- Uses a rolling historical reporting period.
- Can be available at page or origin level.
- Requires sufficient real-user data.
- Used for the Core Web Vitals field assessment.
Lab Data
Lab data is generated by Lighthouse when PSI analyses the URL in a simulated environment. It is useful for debugging because it can reflect the current page immediately.
- Generated during the current test.
- Uses controlled testing conditions.
- Useful for identifying technical bottlenecks.
- Can change from one test run to another.
- Does not represent every real user's device.
- Produces the familiar Lighthouse performance score.
What Is CrUX in PageSpeed Insights?
CrUX is the real-user dataset behind PageSpeed Insights field reporting and several other Google Core Web Vitals tools.
Why Some URLs Have Field Data and Others Do Not
A page needs enough eligible real-user samples to appear at URL level in the CrUX dataset. A recently published page or a low-traffic URL may therefore have no page-level field data.
When page-level data is insufficient, PageSpeed Insights may show origin-level data instead. That represents experiences across pages on the wider origin, so make sure you know which level you are reading.
If neither the page nor origin has sufficient qualifying data, PSI can still run Lighthouse, but real-user field data may not be available.
LCP, INP and CLS Explained
The current Core Web Vitals measure loading performance, responsiveness and visual stability using real-user experience at the 75th percentile.
Largest Contentful Paint
Loading performanceLCP measures how long it takes for the page's largest relevant content element to be rendered within the viewport. On many pages this is a hero image, banner, prominent text block or another major above-the-fold element.
Interaction to Next Paint
Page responsivenessINP assesses responsiveness by observing the latency of qualifying user interactions throughout a page visit and evaluating how quickly the page provides the next visual update.
Cumulative Layout Shift
Visual stabilityCLS measures unexpected movement of visible page content. A page feels unstable when text, buttons, images or other elements suddenly shift while the visitor is reading or trying to interact.
FCP, TTFB and Other Performance Signals
Core Web Vitals receive the most attention, but supporting metrics often help reveal why a vital is failing.
First Contentful Paint
FCP measures when the browser first renders meaningful page content such as text or imagery. It can help identify delays before the user sees the first visible response.
Time to First Byte
TTFB measures the time between the navigation request and the first byte of the response arriving. Slow backend processing, network latency or caching problems can push TTFB higher.
Speed Index
Lighthouse uses Speed Index to estimate how quickly visible page content is displayed during the simulated load.
Total Blocking Time
TBT is a lab metric that helps identify periods where long main-thread tasks can prevent the page from responding promptly during loading.
What Does the PageSpeed Performance Score Mean?
The large 0–100 score comes from Lighthouse lab testing. It is useful for diagnosis and comparison, but it should not be confused with real-user Core Web Vitals.
Do You Need a Perfect 100?
No. Lighthouse considers scores from 90 to 100 good, but Chrome's own documentation notes that achieving a perfect 100 is extremely challenging and not expected.
Optimising a site from a genuinely poor experience to a fast and stable one usually matters far more than spending substantial development time moving a score from 98 to 100.
A strong Lighthouse performance result under the conditions used for that particular test.
Investigate the underlying metrics and diagnostics rather than focusing only on the number itself.
The simulated test indicates substantial performance problems deserve investigation.
Why Does My PageSpeed Score Change Between Tests?
A Lighthouse test is a single simulated run. Small differences between runs are normal and do not necessarily mean the website changed.
Network Variation
Resource delivery and network behaviour can vary slightly between tests, affecting the timing of important assets.
CPU Timing
Variations in processing and task scheduling can influence simulated performance metrics, especially on JavaScript-heavy pages.
Third-Party Resources
Advertising, analytics, chat widgets, video embeds, external fonts and other third-party services may respond differently between runs.
Server & Cache State
Backend load, cache warming, CDN behaviour and dynamic processing can all change the response time observed by an individual test.
How to Read PageSpeed Insights Recommendations
The report is most useful when you trace each recommendation back to the metric and user experience it can realistically improve.
JavaScript Execution
Large scripts and long tasks can occupy the main thread, delaying rendering and user interactions. Review whether every script needs to load immediately and whether unused functionality can be removed.
Render-Blocking Resources
Critical rendering can be delayed when the browser must process blocking stylesheets or scripts before displaying important above-the-fold content.
Image Delivery
Oversized images, poor compression, unnecessarily large intrinsic dimensions or late discovery of hero imagery can substantially increase LCP and transferred bytes.
Web Fonts
Font files can delay text rendering or contribute to layout changes when fallback and final font metrics differ substantially.
Main-Thread Work
Parsing, compiling and executing excessive JavaScript can leave less browser capacity available for rendering and responding to interactions.
Third-Party Code
Analytics, advertising, chat, tracking, social and embedded services can add network requests and JavaScript work outside your direct application code.
Server Response
Slow backend processing can delay every later stage of page loading. Investigate hosting, database work, uncached requests and application bottlenecks where TTFB is consistently poor.
Caching
Appropriate browser, page and edge caching can prevent visitors from re-downloading or regenerating resources unnecessarily.
Layout Shifts
Use the diagnostics to identify unstable page elements and then reserve appropriate dimensions or change how dynamic content is inserted.
Common Ways to Improve PageSpeed Insights Results
Do not apply optimisation techniques blindly. Identify the bottleneck first, then prioritise changes that improve the experience of actual users.
Optimise Images
Serve appropriately sized images, compress them efficiently and choose modern formats where they are suitable. Avoid delivering a huge desktop image to a small mobile viewport.
Prioritise the LCP Resource
The browser should discover important above-the-fold resources quickly. Avoid unnecessarily lazy-loading the main LCP image and remove avoidable request chains before it.
Lazy Load Below the Fold
Images and embeds far below the initial viewport may not need to compete with critical above-the-fold assets during the first load.
Reduce JavaScript
Remove unnecessary libraries and scripts, delay non-critical functionality and split heavy work where possible so the main thread remains responsive.
Reduce CSS Overhead
Remove unnecessary styles where practical and make sure critical rendering is not delayed by excessive stylesheet processing.
Improve Asset Delivery
CDNs, compression and effective caching can reduce transfer time and improve delivery for geographically distributed users.
Optimise Fonts
Reduce unnecessary font families and weights, consider local hosting where appropriate and avoid making visitors download large typography sets they never use.
Reserve Element Space
Provide stable dimensions for images, video, adverts, embeds and other dynamic blocks so surrounding content does not unexpectedly move.
Using PageSpeed Insights on WordPress Websites
WordPress itself is not automatically slow. Performance problems usually come from the complete stack of theme, plugins, media, hosting, third-party code and page design.
Do Not Install Five Performance Plugins to Fix One Problem
Start with the PageSpeed diagnosis and determine whether the dominant issue comes from the server, JavaScript, CSS, images, fonts, page-builder output or third-party resources.
Using several overlapping cache or optimisation plugins can introduce conflicting minification, delayed scripts, layout problems or cache behaviour. A controlled configuration is usually easier to diagnose.
For wider technical analysis, combine performance testing with a structured SEO audit rather than treating PageSpeed as a complete technical SEO assessment.
Does PageSpeed Insights Affect Google Rankings?
Page performance matters, but a PageSpeed score should never be mistaken for a direct ranking formula.
Good Core Web Vitals Matter — But They Do Not Guarantee Rankings
Google recommends achieving good Core Web Vitals for both users and Search, and page experience aligns with what its ranking systems seek to reward. That does not mean a page with perfect performance will outrank a more relevant, useful or authoritative page.
SEO should therefore balance performance with content quality, search intent, crawlability, internal linking and the wider signals explained in our Google ranking factors guide.
For mobile-heavy search audiences, performance work should also be considered as part of a broader mobile SEO strategy.
PageSpeed Insights vs Lighthouse vs Search Console
These tools overlap, but they are designed for different levels of performance diagnosis.
PageSpeed Insights
Best when you want a convenient page-level report combining available CrUX real-user data with a current Lighthouse lab analysis and actionable performance diagnostics.
Google Lighthouse
Lighthouse is the underlying automated auditing system used for PSI's lab diagnostics. It can also be run through Chrome DevTools and other developer workflows and covers more than performance alone. Read our Google Lighthouse guide.
Search Console Core Web Vitals
Search Console is better for seeing broader groups of URLs with real-user Core Web Vitals issues across a verified property. See our Google Search Console guide.
A Better Way to Use PageSpeed Insights
Avoid trying to turn every audit line green. Start with real-user impact and work backwards to the underlying technical cause.
Check Field Data
Determine whether URL-level or origin-level CrUX data is available.
Identify the Failing Vital
Determine whether the primary real-user problem is loading, responsiveness or layout stability.
Use Lab Diagnostics
Use Lighthouse evidence to reproduce and investigate likely technical causes.
Fix the Root Cause
Prioritise the resource, script, server or layout behaviour actually causing the issue.
Retest Immediately
Use repeated lab tests to confirm that the implementation changed the page as expected.
Monitor Field Recovery
Remember that CrUX represents rolling real-user history, so field improvements are not necessarily visible immediately.
PageSpeed Insights Is Not a Complete SEO Audit
A fast page can still have major crawling, indexing, content, canonical, internal-link or architecture problems.
On-Page SEO
PageSpeed does not tell you whether the page properly satisfies search intent, targets relevant queries or uses its headings and content effectively. See our on-page SEO guide .
Website Architecture
Technical performance does not replace good crawling paths, logical hierarchy and internal linking. Explore our website architecture guide .
Broader SEO Toolset
Combine PSI with crawlers, Search Console, analytics and other diagnostic platforms available through our SEO tools hub .
Common PageSpeed Insights Mistakes
Most PSI mistakes come from optimising the displayed number rather than understanding what the report says about real users and the current page.
Chasing a Perfect 100
Once a site is performing well, forcing the last few Lighthouse points can consume significant development effort with little practical improvement for users.
Confusing Field and Lab Data
A current Lighthouse test and a rolling real-user CrUX dataset answer different questions. They should not be expected to match exactly.
Expecting Field Data to Update Immediately
A technical fix can improve today's page while the field report continues to include user experiences from the previous weeks.
Trusting One Lighthouse Run
Lab tests naturally vary. Compare several controlled runs and focus on persistent bottlenecks rather than reacting to one unusually high or low score.
Installing Optimisation Plugins Blindly
Adding caching, minification and delay plugins without understanding the problem can create conflicts while failing to address the actual bottleneck.
Treating PSI as the Whole SEO Strategy
Excellent performance cannot compensate for irrelevant content, poor indexation, weak architecture or no authority. Performance is one part of technical and user-experience optimisation.
Common Questions About Google PageSpeed Insights
Quick answers covering Core Web Vitals, Lighthouse scores, field data, mobile performance and SEO.
01 What is Google PageSpeed Insights?
Google PageSpeed Insights is a web-performance analysis tool that combines real-user Chrome User Experience Report data where available with a Lighthouse lab test and performance diagnostics for a supplied URL.
02 What is a good PageSpeed score?
Lighthouse classifies a performance score of 90–100 as Good, 50–89 as Needs Improvement and 0–49 as Poor. The score should be used as diagnostic guidance rather than as the sole performance objective.
03 What are the current Core Web Vitals?
The current Core Web Vitals are Largest Contentful Paint for loading, Interaction to Next Paint for responsiveness and Cumulative Layout Shift for visual stability.
04 What is a good LCP score?
A good Largest Contentful Paint is 2.5 seconds or less at the 75th percentile of page loads.
05 What is a good INP score?
A good Interaction to Next Paint is 200 milliseconds or less at the 75th percentile.
06 What is a good CLS score?
A good Cumulative Layout Shift score is 0.1 or less at the 75th percentile.
07 Why does PageSpeed Insights say no data?
The page may not have enough eligible real-user samples in the CrUX dataset. PSI can sometimes fall back to origin-level data; if the origin also lacks sufficient data, the field-data section may not be available.
08 Why is my mobile PageSpeed score lower than desktop?
Mobile and desktop experiences differ because devices, processing power, network conditions, layouts and resources can differ. Diagnose the actual mobile bottleneck rather than assuming the desktop implementation behaves identically.
09 Is PageSpeed Insights the same as Lighthouse?
No. PageSpeed Insights uses Lighthouse for its lab analysis, but PSI also integrates available CrUX field data. Lighthouse itself is a broader automated auditing system. See our Google Lighthouse guide.
10 Does passing Core Web Vitals guarantee higher Google rankings?
No. Google recommends good Core Web Vitals and strong page experience, but passing the thresholds does not guarantee high rankings. Relevance, content, links and many other signals remain important.
Optimise the User Experience — Not Just the Green Score
PageSpeed Insights is most valuable when you use each part of the report for the question it can actually answer. CrUX shows how real eligible users have experienced the site over time, Lighthouse provides an immediate controlled diagnostic test, and Core Web Vitals give you a consistent framework for loading, responsiveness and visual stability. Identify the metric that is struggling, trace it to the underlying technical cause, fix that cause and then monitor both lab tests and real-user data. The goal is a faster website for people, not merely a prettier performance report.
