What this website speed test measures
A slow page loses visitors before they read a word. This website speed test shows where the time and the bytes go. Our server requests the page like a first-time visitor, follows any redirects and splits the load time of the HTML into its parts: the DNS lookup, the connection, the TLS handshake, the time the server needs to build the page, and the download. The time to first byte (TTFB) is the moment the first byte of the HTML arrives, and it tells you how fast your hosting really is.
Then it reads the HTML and requests the stylesheets, scripts, images and fonts it names, up to 60 files, to add up the page weight and the number of requests. For each file it records the size sent over the network, whether it came compressed with gzip or Brotli, how long browsers may cache it, and for images the format and whether width and height are set. It also reports the HTTP version (HTTP/1.1, HTTP/2, and whether the server offers HTTP/3) and any CDN in front of the site.
Everything is measured from our server in Germany, without a browser. That is why the result shows load time and page size on the network, not Core Web Vitals. See the questions below for how it compares to Lighthouse.
How the speed grade is worked out
The score starts at zero and collects up to 100 points in eight rules. 90 points or more is an A, 75 a B, 60 a C, 45 a D, anything below an F. The list of fixes is sorted by the points each fix would win back, so the top item is the one worth doing first.
| Rule | Points | How it is scored |
|---|---|---|
| Server response | 20 | Time to first byte of the HTML (connect, TLS handshake and server wait). Full points up to 300 ms, 16 up to 500 ms, 11 up to 800 ms, 6 up to 1.2 s, 3 up to 1.8 s. |
| Redirects | 5 | No redirect before the page: 5. One: 3. Two: 1. More: 0. |
| Page weight | 20 | HTML plus every measured file, as sent over the network. Full points up to 500 KB, 17 up to 1 MB, 12 up to 2 MB, 8 up to 3 MB, 4 up to 5 MB. |
| Requests | 10 | The HTML, redirects and every file the page refers to. Full points up to 20, 8 up to 40, 5 up to 70, 3 up to 100, 1 up to 150. |
| Render-blocking files | 15 | Scripts in the head without async, defer or type module, plus stylesheets in the head beyond the first. None: 15, one: 12, two: 9, up to four: 6, up to seven: 3. |
| Compression | 10 | 6 points when the HTML comes compressed with gzip, Brotli or zstd, 4 points shared by your own CSS, JavaScript and SVG files over 1 KB. |
| Caching | 10 | Share of your own static files that browsers may keep for at least 7 days (Cache-Control max-age or Expires). |
| Images | 10 | 4 points for modern formats (share of image bytes not in JPEG, PNG or GIF), 3 minus one per image over 300 KB, 3 for width and height on every image. |
The fixes that help most
- Slow time to first byte: use page caching (a cache plugin on WordPress, or a CDN that caches HTML), and check slow plugins and database queries. If most of the time is connect and TLS, the server is simply far away from your visitors.
- Render-blocking scripts: add defer to scripts in the head, or async to independent ones like analytics, so the browser can paint the page before they run.
- Heavy images: resize photos to the size they are shown at and compress them with the image compressor, or turn JPEG and PNG files into WebP with the WebP converter. Add width and height to every image so the layout does not jump.
- No compression or short caching: both are server settings. Turn on Brotli or gzip, and send a long Cache-Control max-age for CSS, JavaScript, fonts and images. The HTTP header checker shows the headers your server sends and every redirect on the way.
- Fonts: use WOFF2 files, add font-display: swap so text shows while fonts load, and preload the one or two fonts at the top of the page.
How to test your page speed
- 1Enter the page address, ideally the final one your visitors land on, and press Test speed. Big pages take up to half a minute.
- 2Start with the first item under What to fix first, then check the timing bar to see whether the server or the network is slow.
- 3Use the file table to find the heaviest images and scripts, fix them, and run the test again after 10 minutes to compare.
Frequently asked questions
Is this the same as Google PageSpeed Insights or Lighthouse?
No. PageSpeed Insights runs Lighthouse, which loads the page in a real browser, runs its JavaScript and measures Core Web Vitals like Largest Contentful Paint. This tool does not run a browser. It requests the HTML and the files the page refers to from our server and measures the network side: server response, weight, requests, compression, caching and render-blocking files. That makes it fast and shows exactly which files to fix. For lab data in a browser, also run your page through Google PageSpeed Insights at pagespeed.web.dev.
Where is the test run from?
From our server in Germany, over a fast data center connection. Visitors in Europe see similar connect and TLS times; visitors in America or Asia wait longer for every round trip, and visitors on a phone network longer still. Run the test a few times: the first run can be slower when a cache on the site is cold.
Why is the page weight lower than in my browser?
The test counts the HTML and the stylesheets, scripts, images and fonts named in it, up to 60 files per run. Files that scripts load later, such as ads, tracking pixels, chat widgets or images in a slider, are not seen, and neither are images set as CSS backgrounds. Treat the weight as a lower limit. A browser also loads only the font files a page really uses, so at most six declared font files are counted.
What counts as my own file?
Files on the same site as the page, on a name that starts like it (such as a CDN domain like example-static.com for example.com), or on a host that serves at least a quarter of the page’s files. Only your own files count for caching and compression, because you cannot change how a third-party service sends its scripts.
Why did my score change between two runs?
Server response and the time for each file depend on the load on the site and on the network at that moment. A test is cached for 10 minutes, so a shared link shows the same result. After that, a new run measures again. To keep the tool fair, each visitor can run 6 tests in 10 minutes and 30 a day.
Does a better score mean better Google rankings?
Speed is one of many ranking signals, and Google measures it from real Chrome visitors (Core Web Vitals), not from any test score. A fast server, a light page and few render-blocking files do improve those real numbers, which helps visitors first and rankings second.