Free website speed test tools (and how to read them)

You can test your website's speed for free in about 2 minutes, and you do not need to install anything or hand over a credit card. The tools worth using are Google PageSpeed Insights, GTmetrix, WebPageTest and the Lighthouse panel already built into your browser. The trick is not running the test - anyone can run the test. The trick is knowing which 3 numbers out of the dozens these tools throw at you actually predict whether your visitors leave. This page covers both, and it is the layer that sits after the tool: what to do once you have a number you do not understand.

The free tools worth using

PageSpeed Insights is the one to open first. Paste your URL into pagespeed.web.dev, wait 20 seconds, and you get a score out of 100 plus - and this is the part most people skip - real measurements from actual Chrome users who visited your site over the past 28 days. It is free, it needs no account, and it is the closest free tool to what Google itself uses to judge your pages. Its one weakness is that it tests a single URL at a time and gives you the result as a snapshot, so it is a diagnosis tool, not a monitoring one. Start here every time. We walk through running it in how to test website speed free.

GTmetrix is the friendlier read of the same underlying engine. It runs Lighthouse for you, draws a waterfall chart that shows every file your page loads and how long each one took, and lets you pick a test location and connection speed. The free tier tests from a limited set of servers (none in Australia without an account) and throttles how often you can run, but for seeing where the time goes - which image, which script, which font - its waterfall is easier to read than anything Google ships. Use it when PageSpeed tells you that you are slow and you want to see exactly what is eating the seconds.

WebPageTest is the serious instrument. webpagetest.org lets you choose a real device, a real location (including Sydney), a real connection profile, and runs the test multiple times so you get a median rather than one lucky or unlucky number. It produces a filmstrip showing your page painting frame by frame, which is the single most persuasive thing you can put in front of a web developer who insists the site "feels fine on my machine". It is more intimidating than the others and slower to run. Reach for it when the easy tools disagree, or when you need evidence rather than a vibe.

Chrome DevTools Lighthouse is the one already on your computer. Open your site in Chrome, press F12, click the Lighthouse tab, and run a report - same engine as PageSpeed Insights, no website to visit. Its advantage is that it tests the page as you are seeing it right now, including pages behind a login or on a staging site no public tool can reach. Its trap is that it runs on your machine, over your connection, with your browser extensions interfering, so the number is easy to flatter or wreck by accident. It is for spot checks during a fix, not for a verdict. More on what it is in what Lighthouse is.

The numbers that actually matter

Ignore the big score out of 100 for a moment. It is a weighted average of things you cannot act on directly, and chasing it is how people waste a weekend moving a number from 78 to 84 that no visitor will ever feel. Three measurements underneath it are the ones that matter, because Google calls them Core Web Vitals and uses them in ranking, and because they each map to a thing a human actually experiences.

Largest Contentful Paint (LCP) is how long until the biggest thing on screen - usually your hero image or headline - finishes loading. This is "how long until the page looks ready". Good is under 2.5 seconds. It is the number to fix first, because it is the one visitors feel as "this site is slow".

Interaction to Next Paint (INP) is how quickly the page responds when someone taps a button or opens a menu. This is "does it feel sluggish when I touch it". Good is under 200 milliseconds. It replaced the old First Input Delay metric in 2024, so any guide citing FID is out of date.

Cumulative Layout Shift (CLS) measures how much the page jumps around as it loads - the ad that pushes the text down, the button that moves just as you go to tap it. This is "does it stop me misclicking". Good is under 0.1.

The other thing to understand is lab data versus field data, because every tool shows both and they routinely disagree. Lab data is a single test run in a controlled simulation - repeatable, but artificial. Field data (PageSpeed labels it "Discover what your real users are experiencing") is the median of real visitors on real phones and real connections over 28 days. When the two conflict, the field data wins, every time. It is the only one that reflects the cheap Android phone on a 4G connection in a regional town that is actually loading your site.

Common variations

Why two tools disagree. GTmetrix says 91, PageSpeed says 64, and both are "right". They tested from different locations, on different simulated connections, possibly at different moments when your server was under different load. A speed score is a measurement, not a fact about your site - it always carries the conditions it was taken under. This is why you compare a tool against itself over time, never one tool's number against another's.

Mobile versus desktop. Always test mobile, and expect it to be the worse, often dramatically worse, score. Desktop runs on a fast machine over a fast wired connection; mobile simulates a mid-range phone on throttled mobile data, which is closer to how most Australians actually reach you. Google predominantly uses the mobile experience to rank you. A great desktop score sitting next to a poor mobile one is not a contradiction - it is the mobile number telling you the truth. We unpack the gap in mobile vs desktop PageSpeed score.

Why your score changes every run. Run the same test three times and you will get three different numbers. Lab tests simulate a connection, and that simulation has natural variance; your server may be busier on one run; a third-party script may load an ad faster one time than the next. This is normal and it is why WebPageTest runs multiple passes and reports a median. One run is an anecdote. Treat a single number as a rough band, not a precise grade - and if a number swings wildly between runs, that volatility is itself a finding, often a sign of third-party scripts loading unpredictably.

The small-business-appropriate way to use them

You do not have time to become a performance engineer, and you do not need to. There is a three-step version that gets you most of the value:

Test mobile, not desktop. Set the tool to mobile and read that number. It is the one your customers and Google both care about, and the one your developer is least likely to have checked.

Use the field data, not the lab score. Scroll past the big simulated number to the "real users" section in PageSpeed Insights. If that section says your data is good, you are fine regardless of what the lab score says. If it says poor, you have a real problem worth money to fix.

Fix the top item, then re-test. Both PageSpeed and GTmetrix list opportunities in priority order. Do the first one - usually "compress your images" or "this one script is blocking everything" - then run the test again and watch the LCP move. One real fix beats ten cosmetic ones. Most small-business sites are slower than their owners realise - well past Google's 2.5-second "good" mark - which means a single honest fix usually moves you ahead of most of your competitors. That is the whole reason our website-speed pillar exists, and the practical first-fix walkthrough is our first why-is-my-site-slow walkthrough.

When a test result isn't enough

A speed test tells you that a page is slow and roughly where the time goes. It does not tell you why in the way you need to act - whether the cause is your hosting, an oversized hero image, a booking widget loading 14 third-party scripts, or a theme nobody has updated since 2019. Diagnosing that means reading a waterfall chart properly, and one URL's result does not represent a whole site: your homepage can score well while every product page quietly fails.

It also stops at speed. A page can load in 1.8 seconds and still be missing the DNS, email and security basics that decide whether your site is trusted and reachable in the first place - a fast site that lands in spam folders is still losing you customers. That is the gap a single-metric tool leaves. Running your domain through our scan checks Speed alongside the Domain, Email and Security categories in one pass, and grades each against the real Australian small business baseline rather than an abstract 100. A speed test is the right place to start. It is rarely the right place to stop.