PageSpeed Insights vs GTmetrix vs Google Search Console: what’s the difference?
You run your site through three different tools and get three completely different scores. PageSpeed Insights says 43. GTmetrix says B. Search Console says “Needs improvement.” They’re all looking at the same website — so why don’t they agree? Because they’re measuring different things.
What is the difference between lab data and field data?
This is the single most important thing to understand. Performance tools either simulate a page load in a controlled environment (called “lab data”), or they report on what real visitors actually experienced (called “field data” or “real-user data”).
- Lab data is consistent and actionable — the same tool testing the same page under the same conditions. Useful for catching problems and measuring the impact of changes.
- Field data is what actually happened for real visitors. It accounts for different devices, network speeds, and locations. It’s what Google uses for ranking.
GTmetrix and PageSpeed Insights both primarily run lab tests. Google Search Console reports field data from real users of your site.
What does Google PageSpeed Insights measure?
PageSpeed Insights runs a simulated page load using Lighthouse — Google’s open-source auditing tool — and produces a performance score from 0 to 100. The score is calculated from a weighted combination of six metrics, with Largest Contentful Paint (LCP) and Total Blocking Time (TBT) carrying the most weight.
PSI also shows field data when it’s available: real measurements from Chrome users who have visited your page recently, pulled from the Chrome User Experience Report (CrUX). If your page has enough traffic, you’ll see both a lab score and real-user Core Web Vitals at the top of the results.
PageWeight is built on the PageSpeed Insights API, so the scores it reports are the same as running PSI directly.
What does GTmetrix measure?
GTmetrix also uses Lighthouse for its performance analysis, so the underlying data is similar to PSI. The difference is in how it presents results. GTmetrix has its own “GTmetrix Grade” (A through F) which combines performance with a “structure” score based on best practices. This means you can have a 90 performance score but a B grade if some technical checks fail.
GTmetrix also offers desktop testing, historical tracking, scheduled monitoring, and waterfall charts that show every request the page makes. These are useful for technical audits but less relevant for most site owners. The free plan tests from Seattle by default; paid plans let you test from other locations.
Because both tools use Lighthouse, you’d expect the scores to be close — but they can still differ by 10–20 points due to server load at test time, slightly different Lighthouse versions, or different test conditions. This is normal.
What does Google Search Console show?
Search Console is different in kind, not just degree. It does not run performance tests. Instead, it aggregates real-user Core Web Vitals data from Chrome users who visited your site over the past 28 days, and classifies each URL as Good, Needs Improvement, or Poor for LCP, CLS, and INP.
Because it reflects real users across all their devices and connections, it’s the most representative view of actual performance — and it’s the data Google uses as a ranking signal. A page that scores 90 in a lab test can still show “Poor” in Search Console if real visitors on slow mobile connections experience it as slow.
The limitation: Search Console only shows data for pages with enough recent traffic to populate. New pages, low-traffic pages, or recently launched sites may show no data at all.
Why the numbers don’t agree
Given all of the above, the scores will almost always be different, and that’s expected:
- Lab tests simulate a single test run; real-user data averages thousands of visits across different conditions.
- Lighthouse scores vary run-to-run by 5–10 points due to server load and test machine variance.
- GTmetrix and PSI may run slightly different Lighthouse versions or use different throttling profiles.
- Search Console uses the 75th percentile of real user visits — it’s designed to capture the experience of slower-than-average users, not the best case.
Which one should you trust?
For understanding what Google actually uses for rankings: Search Console. It’s real-user data from the browser Google controls.
For diagnosing what’s causing problems and checking whether your fixes worked: PageSpeed Insights or PageWeight. Lab tests are consistent enough to detect meaningful changes, and they give you specific recommendations.
For a detailed technical audit with waterfall charts: GTmetrix. More useful if you’re a developer or working with one.
In practice: use a lab tool like PageWeight to identify and fix problems, then watch Search Console to confirm the real-world impact over the following weeks.