Test Your Website Speed for Free
Measure your website's performance, review available Core Web Vitals, and discover opportunities to improve the user experience.
Free and without an account. Up to 10 tests an hour from one connection, public addresses only. Results are kept for 24 hours and can be shared by link.
What this website speed test measures
The test opens the page you enter in a real Chromium browser, twice, each time with an empty cache: once as a desktop with a 1440 × 900 window and no throttling, and once as a phone with a 390 × 844 screen, a slowed network (150 ms latency, about 1.6 Mbit/s down and 750 kbit/s up) and a processor slowed four times. It then reads the timings the browser recorded while the page loaded.
Core Web Vitals from the lab: LCP and CLS
Largest Contentful Paint (LCP) is when the largest image or block of text in the first view finished drawing — usually the moment a visitor feels the page has arrived. Google counts 2.5 seconds or less as good. Cumulative Layout Shift (CLS) adds up how far visible content jumps while the page loads; 0.1 or less is good.
Loading milestones: FCP, TTFB and the load event
First Contentful Paint (FCP) is when anything first appears. Time to First Byte (TTFB) here is the time the server took to start answering the request for the page itself, so a high figure points at the server or its caching rather than at the page. The load event fires once the page and the files it needs at first have downloaded.
Responsiveness: Total Blocking Time
Total Blocking Time (TBT) adds up how long scripts kept the browser’s main thread busy after the first paint, counting only the part of each task beyond 50 ms. While the main thread is blocked, taps and key presses wait. The Core Web Vital for responsiveness, Interaction to Next Paint, needs real visitors and is explained below.
Page weight and requests
The test counts every file the page downloaded, its size as it crossed the network, and which files were the heaviest and the slowest. Those two lists are usually the quickest route to an improvement.
How to read your results
The Provelex lab score
Each profile gets a score from 0 to 100. Every scored metric is placed on the same log-normal curves Lighthouse version 10 uses, then weighted: Largest Contentful Paint 30%, Total Blocking Time 30%, Cumulative Layout Shift 25% and First Contentful Paint 15%. Lighthouse also weights Speed Index, which this test does not measure, so its share is split between the two paint metrics. That makes the score comparable in spirit to Lighthouse’s, but not identical to PageSpeed Insights.
The colours
A score of 90 or more is shown as good, 50 to 89 as needing improvement and below 50 as poor, as Lighthouse colours them. The four scored metrics are coloured by their own share of the score, so desktop is held to tighter targets than mobile. Server response time, load event and page weight use fixed thresholds, and request count is shown for reference only.
Mobile and desktop together
Read the two side by side. A page that is fine on desktop but poor on mobile usually sends too much to a slow connection or runs too much script for a slower processor. A page that is slow on both often has a slow server response or a very heavy first view.
Lab data vs field data
Lab data comes from loading the page once under fixed conditions, as this test does. Because the conditions are the same each time, it is the right tool for finding what makes a page slow and for checking whether a change helped.
Field data comes from real visitors’ browsers on their own devices and connections. Google collects it from Chrome users who have opted in and publishes it as the Chrome UX Report, summarised over the previous 28 days. Google’s Core Web Vitals assessment — LCP, INP and CLS — is based on field data, not on any lab score.
This test reports lab data only. It does not query the Chrome UX Report, so it never shows field figures, and it cannot report Interaction to Next Paint, which only exists when somebody interacts with the page. A good lab result is a strong sign, not a guarantee of a good field result.
How to improve website speed
Make images smaller
Images are often the heaviest files. Serve them at the size they are displayed, use modern formats such as WebP or AVIF, and load images below the first screen lazily. If the largest element is an image, do not lazy-load that one.
Defer scripts the first view does not need
Long scripts block the main thread and raise Total Blocking Time. Load non-essential scripts with defer or after the page is interactive, remove ones that are no longer used, and review third-party tags such as chat widgets and trackers.
Speed up the server response
A slow Time to First Byte delays everything after it. Cache pages that are the same for every visitor, use a content delivery network, and look for slow database queries behind the page.
Reserve space to stop layout shifts
Give images and video explicit width and height, reserve space for adverts, banners and embeds, and avoid inserting content above what is already on screen.
When to retest
Retest after every change you make for speed, and after releases that add scripts, images or third-party tags. Run the test two or three times before drawing conclusions: lab results vary a little between runs because networks and servers do. Field data changes more slowly, because it covers the previous 28 days of real visits.
Frequently asked questions
Is the website speed test free?
Yes. It needs no account and no credits. To keep it free for everyone, each connection can run up to 10 tests an hour and two tests can run at once across the service. Results are deleted after 24 hours.
Why is my score different from PageSpeed Insights?
The Provelex lab score uses Lighthouse’s scoring curves and weights, but it is a different measurement. PageSpeed Insights runs Lighthouse on Google’s servers and, by default, simulates a slow connection after loading the page quickly; this test slows the connection and processor while the page actually loads, from our server’s location. It also does not measure Speed Index, so that weight is shared between First and Largest Contentful Paint. Scores from either tool also vary a little from run to run.
Why is my mobile score lower than my desktop score?
The mobile profile loads the page on a phone-sized screen over a deliberately slow connection (150 ms latency, about 1.6 Mbit/s) with the processor slowed four times, and is scored against Lighthouse’s mobile curves. That is closer to a mid-range phone on a weak mobile signal than to a new phone on Wi-Fi, which is the point: it shows where slower devices struggle.
Why is Interaction to Next Paint (INP) not shown?
INP measures how quickly a page responds to real taps, clicks and key presses, so it can only come from real visitors (field data). This lab test loads the page without anybody using it and does not use Google’s Chrome UX Report, so there is no INP to report. Total Blocking Time is shown as the closest lab signal.
Does the test change anything on my website?
No. It loads the address you enter twice, the way a visitor’s browser would — once as a desktop and once as a phone. It does not click, sign in or submit anything, and private or internal addresses are refused.
Speed is one part of a healthy site
The free health check reads security headers, encryption, accessibility, search tags and the mobile view as well. A free account keeps an assessment history and adds AI-driven testing of your site’s journeys.