GTmetrix vs PageSpeed Insights: Grades, Reports and Data

Your developer sends a GTmetrix report with a green A. Your SEO consultant sends a PageSpeed Insights screenshot with a red 41 on mobile. Same page, same day, and neither tool is broken: they measure different things under different conditions. This guide explains what GTmetrix is, how its grade is calculated and how to read a GTmetrix report, then breaks down how it differs from PageSpeed Insights in data, test location, throttling and scoring, so you know which number to act on.

What is GTmetrix?

GTmetrix is a website speed testing tool from an independent company of the same name. It loads your page in a real Chrome browser from a test server you choose, runs Lighthouse performance audits on that page load, and adds its own scoring, a request-by-request waterfall, optional video of the load and monitoring over time. You need an account to run tests; paid plans add more locations, tests and monitoring.

PageSpeed Insights (PSI) is Google's free tool. It runs one Lighthouse test from Google's infrastructure and, where enough Chrome users have visited, shows real-user Core Web Vitals data. Both use Lighthouse, which is why their numbers are related but rarely identical.

What is the GTmetrix grade?

The GTmetrix Grade is the letter from A to F at the top of every GTmetrix report. According to GTmetrix's own documentation, it is a weighted average of two percentage scores:

• Performance Score (60%): essentially the Lighthouse Performance score, measured on GTmetrix's own browser, hardware and the Analysis Options you chose (location, device, connection speed, ad blocking). It blends Total Blocking Time (30%), Largest Contentful Paint (25%), Cumulative Layout Shift (25%), First Contentful Paint (10%) and Speed Index (10%), the same weights as Lighthouse.
• Structure Score (40%): GTmetrix's own assessment of Lighthouse's Opportunity and Diagnostic audits, plus a few custom GTmetrix audits such as using a CDN, enabling keep-alive and avoiding CSS @import. Each audit is given points based on potential savings and importance, so this score reflects how well the page is built for speed rather than how fast one load happened to be.

The weighted result maps to a letter:

GTmetrix says it monitors and adjusts these weightings over time, so older articles quoting a different split (such as 70:30) may describe an earlier version. Two practical consequences follow. First, a page with a modest Performance Score can still earn a B if its Structure Score is high. Second, because the default test is desktop on an unthrottled connection (see below), the grade usually looks kinder than PSI's mobile score for the same page.

How to read a GTmetrix report

A GTmetrix report is organised into tabs. Read them in this order:

1. The header. The GTmetrix Grade, the Performance and Structure Scores, and three Web Vitals: LCP, Total Blocking Time (a lab stand-in for responsiveness) and CLS. Check the test details next to them: location, browser, device and connection. A result only means something once you know the conditions.
2. Summary tab. The Speed Visualization shows frames of the page loading, with markers where each metric fired. Top Issues lists the Structure audits that cost the most, and GTmetrix recommends starting there. The Page Details chart breaks the page down by request type and size, which shows at a glance whether JavaScript, images or fonts dominate.
3. Performance tab. Each Performance metric with its value, a colour rating and a short explanation, plus browser timings that don't affect the score but help with diagnosis.
4. Structure tab. Every audit sorted by impact, worst first. Expand one to see the exact files responsible, such as render-blocking stylesheets or oversized images.
5. CrUX tab. Real-user data from Google's Chrome User Experience Report for the latest 28-day period, with a Core Web Vitals pass or fail where enough data exists. GTmetrix notes this may differ slightly from what PSI or Search Console show.
6. Waterfall tab. Every request in load order, with DNS, connect, wait and receive times. This is where slow servers (long wait times), heavy third-party scripts and request chains become obvious.
7. Video and History tabs. A recording of the load if you enabled video, and graphs of past tests so you can see whether a change helped.

A simple routine: note the test conditions, fix the top one or two Structure issues, then re-run the test from the same location and device and compare. Run it a few times before drawing conclusions, as single runs vary.

At a glance

Difference 1: field data vs lab data

This is the most important difference, and the one most people miss.

PageSpeed Insights shows field data. The top section of a PSI report ("Discover what your real users are experiencing") comes from the Chrome User Experience Report: real visits from Chrome users over the previous 28 days. It reports LCP, INP and CLS, the three Core Web Vitals, plus FCP and TTFB. This is the same data source Google Search Console uses for its Core Web Vitals report. If a URL doesn't get enough traffic, PSI falls back to origin-level data, or shows none.

GTmetrix shows field data too, but leads with the lab. Its CrUX tab shows the same Chrome data, but the grade, scores and diagnostics all come from GTmetrix's own test.

Both tools show lab data. PSI's lower section, and GTmetrix's grade and diagnostics, come from a single synthetic page load in a controlled environment.

Why this matters:

• Only field data tells you whether you pass Core Web Vitals. A page can score 95 in the lab and still fail INP for real visitors on slower phones.
• INP can't be measured in a lab page load because it needs real user interactions. Lab tools use Total Blocking Time (TBT) as a proxy for responsiveness. GTmetrix reports LCP, TBT and CLS as its Web Vitals for this reason.
• Lab data is better for debugging. It's repeatable, it shows you the exact resources involved, and it works before a page has any traffic.

If you only remember one thing: use PSI's field data (or Search Console) to decide whether you have a problem, and lab tools to work out why. Our Core Web Vitals guide covers each metric.

Difference 2: test location

PSI runs its Lighthouse test from Google's infrastructure. According to Google's documentation, the test runs in a data center in North America, Europe or Asia, and the location used appears in the report's environment details. You can't choose it.

GTmetrix lets you pick where the test runs from. It currently lists 28 test locations across North America, Latin America, Europe, the Middle East, Asia-Pacific and Africa, with the number available depending on your plan. Its pricing page lists 5 locations on the Lite plan, 10 on Core and all 28 on Advanced and Expert.

Location affects every network-dependent metric: DNS lookup, connection setup, TTFB and download time. A site hosted in Frankfurt with no CDN will look much faster tested from London than from Sydney. If most of your customers are in one region, test from there. If they're everywhere, that's a sign you may need a CDN; see do you need a CDN.

Difference 3: device and throttling

This alone explains most "why is PSI so much worse?" questions.

PSI's mobile test uses Lighthouse's default simulated throttling: a "Slow 4G" network profile (150 ms round-trip time, about 1.6 Mbps download) and a 4x CPU slowdown, emulating a mid-range phone. That's deliberately pessimistic. PSI's desktop test uses a much faster profile.

GTmetrix loads the page in a real Chrome browser and, by default, tests as desktop on an unthrottled connection. You can switch to simulated mobile devices and throttled connections in Analysis Options, and save combinations as presets.

So a default GTmetrix run is roughly comparable to PSI's desktop result, not its mobile one. If you want to compare like with like, set GTmetrix to a mobile device and a throttled connection.

Difference 4: how the scores are built

PageSpeed Insights: Lighthouse Performance score

PSI's score is Lighthouse's Performance score, a weighted blend of five lab metrics. In current Lighthouse versions the weights are:

Scores of 90-100 are rated good, 50-89 needs improvement, and 0-49 poor. Note that the field data at the top of the page doesn't feed into this number at all. See PageSpeed Insights score explained for more.

GTmetrix: the GTmetrix Grade

GTmetrix's Performance Score uses the same Lighthouse metric weights as PSI, but on GTmetrix's hardware and test settings. The Grade then adds the Structure Score at 40%, as explained in the grade section above. Because Structure rewards good build practices regardless of conditions, a well-built page on a fast, unthrottled desktop test can earn an A while PSI's throttled mobile score sits in the red.

Difference 5: diagnostics and workflow

GTmetrix is built for investigating:

• Waterfall chart of every request, with timing broken into DNS, connect, wait and receive.
• Video and filmstrip of the page loading.
• Testing behind HTTP basic auth, useful for staging sites.
• Monitoring and alerts, history graphs and comparisons between reports.
• API access on paid plans.

PSI is built for a quick verdict and a prioritized list of opportunities, with no login and no setup. It also has a free API, which many other tools use under the hood, including our page speed checker, which shows the PSI score, Core Web Vitals and the top fixes for any URL.

Why the numbers change between runs

Both tools show some variation from one run to the next. Google's Lighthouse documentation lists common causes: A/B tests or rotating ads, changes in internet routing, different hardware and, for local runs, browser extensions or antivirus software. Treat any single lab score as one sample. Run three to five tests and look at the median before deciding a change helped or hurt.

Which should you use?

• "Does my site pass Core Web Vitals?" PSI field data or Search Console.
• "Why is this page slow?" GTmetrix (waterfall) or Lighthouse in DevTools.
• "How fast is it for customers in Australia?" GTmetrix with a Sydney location.
• "Did last week's plugin update slow things down?" Any tool with history: GTmetrix monitoring, or scheduled re-scans in a broader audit tool.
• "Is speed my biggest problem at all?" A full audit. Performance is often fine while DNS, email or security issues go unnoticed.

Run a free audit to see your PageSpeed Insights score and Core Web Vitals alongside SEO, DNS, email and security checks in one report. For a wider comparison of free tools, see free website audit tools compared.

Frequently asked questions

Why is my GTmetrix grade A but my PageSpeed score low?

Most often because GTmetrix tested on desktop with an unthrottled connection while PSI's headline is a throttled mobile test. GTmetrix's grade also includes its Structure score. Compare GTmetrix's mobile, throttled result with PSI's mobile score, or PSI desktop with GTmetrix's default.

Does Google use my GTmetrix score for rankings?

No. Google uses real-user Core Web Vitals data from Chrome as part of its page experience signals. Neither GTmetrix's grade nor PSI's lab score is a ranking input in itself.

What is a good GTmetrix grade?

An A means the weighted score is 90% or higher, and a B covers 80-89%. For most business sites, a B or better on a realistic test (a location near your customers, and a throttled mobile device if most visitors use phones) is a sound target. Check PSI field data or Search Console as well, because passing Core Web Vitals for real visitors matters more than the letter.

Does GTmetrix show real-user data?

Yes. The CrUX tab shows Chrome User Experience Report data for the latest 28-day period, with a Core Web Vitals assessment, where Google has enough data for the page. The grade itself is based only on GTmetrix's lab test.

Is GTmetrix still free?

You need an account to run tests, and the free tier is limited. GTmetrix's site currently describes a free trial, with paid PRO plans starting at $4.99 a month billed yearly. Check GTmetrix's pricing page for the current limits, since they've changed several times.

Which tool is more accurate?

Neither is more accurate; they answer different questions. PSI's field data is the most accurate picture of real visitors. GTmetrix's lab tests are better for reproducing and diagnosing a specific load.

What to do next

Open PSI and check whether your field data passes all three Core Web Vitals on mobile. If it doesn't, use GTmetrix or Lighthouse to find the slow resources, fix the biggest one, and re-test. If it does pass, stop chasing a lab score of 100 and look at the rest of your site's health instead.

Related reading

• How to Check Website Speed: Free Tools and What the Numbers Mean: How to check website speed with a free page speed test, read LCP, INP and CLS correctly, and test the whole site, not just one page. Free tools compared.

Sources

• Google: About PageSpeed Insights
• Chrome for Developers: Lighthouse performance scoring
• Lighthouse docs: Network throttling
• GTmetrix: Everything you need to know about GTmetrix reports
• GTmetrix: Test server locations
• GTmetrix: Pricing
• Search Console Help: Core Web Vitals report