Page Speed Deep Dive
Measure Core Web Vitals in a real Chrome browser and get optimization recommendations
In 30 seconds
- Good is LCP under 2.5s, CLS under 0.1, and Total Blocking Time under 200ms.
- Images and JavaScript are behind most slow pages.
- This is a lab test. Google ranks on real-user data, so use both.
How fast does it really load?
Enter a URL and we load the page in a real Chrome browser on our server, record how it loads, and report:
- Largest Contentful Paint (LCP): when the biggest image or text block appeared.
- Cumulative Layout Shift (CLS): how much the layout jumped around.
- Total Blocking Time (TBT): how long long-running JavaScript blocked the page.
- First Contentful Paint (FCP): when the first text or image appeared.
- Time to First Byte (TTFB): how long the server took to start answering.
- Full load time: when the browser’s load event finished.
- Page stats: total size, number of requests, and a breakdown by file type.
- Opportunities: suggestions based on what we measured, plus a screenshot.
The performance score out of 100 is an estimate built from these numbers, using Google’s good and poor thresholds for each one.
Why speed matters
Core Web Vitals are Google’s measures of real-world experience: loading (LCP), visual stability (CLS) and responsiveness (INP). They’re part of how Google ranks pages. Relevance matters far more, but between two equally useful pages, speed can tip it.
The bigger effect is on people. Slow pages lose visitors before they see anything, and layout shifts cause mis-taps. Faster pages tend to keep more visitors and convert more of them.
INP needs a real person clicking around, so no lab test can measure it. Total Blocking Time is the usual stand-in: lots of blocking JavaScript during load usually means sluggish clicks later.
Reading your results
Largest Contentful Paint
Good is 2.5 seconds or less; over 4 is poor. The LCP element is usually a hero image or the main heading.
Quick fix
Don’t lazy load the hero image. Compress it, serve WebP or AVIF, preload it if it’s set in CSS, and cut render-blocking CSS and JavaScript.
Cumulative Layout Shift
Good is 0.1 or less; over 0.25 is poor.
Quick fix
Give images and videos width and height, reserve space for ads and embeds, and don’t push banners in above existing content.
Total Blocking Time
Good is 200 milliseconds or less; over 600 is poor.
Quick fix
Ship less JavaScript, split big bundles, defer anything not needed straight away, and question every third-party tag: chat widgets, A/B tools and trackers add up fast.
Time to First Byte
A slow TTFB delays everything after it. Caching, a CDN, and faster database queries and server rendering all help.
Watch out
We test from one server location, so TTFB includes the network distance to your host. A distant host will look slower here than it does for nearby visitors.
Page weight and requests
Heavy pages hurt most on mobile networks, and images and JavaScript are usually the culprits. The page size analyzer breaks the weight down file by file.
Lab data and field data
This is a lab test: one load, from one place, on a fast machine. Perfect for diagnosing problems and checking a fix worked, but it varies between runs and won’t match every visitor’s experience.
Google ranks on field data from real Chrome users, which you can see in the Core Web Vitals report in Search Console. Use field data to find which pages are slow for real people, and lab tests to work out why.
Quick wins for most sites
- Compress images and serve WebP or AVIF. The image SEO analyzer finds the worst offenders.
- Lazy load images below the fold, never the hero.
- Remove plugins, scripts and tags you don’t use.
- Turn on gzip or Brotli, and long cache lifetimes for static files.
- Use a CDN so files come from near your visitors.
Quick answers
Why is my score different from PageSpeed Insights?
PageSpeed Insights runs Lighthouse with simulated slow mobile and its own scoring. Ours uses unthrottled measurements and simpler thresholds. Compare results from the same tool over time.
Why do results change between runs?
Server load, caching, third-party scripts and the network all vary. Run it two or three times and go with the typical result.