In the latest edition of the Web Almanac, 45% of WordPress sites passed all three Core Web Vitals on mobile, against 74% of Wix sites, 69% of Squarespace sites and 85% of Duda sites. The HTTP Archive's Web Almanac 2025 matched real Chrome user data from July 2025 with the platform each site's home page was built on.
We planned this post as a different one: Core Web Vitals by CMS, measured from getReport's own reports. We can't publish that yet, and we would rather say why than print a chart that rests on too few sites. So this post does three things: it shows the public numbers and what they really measure, it explains how getReport works out which CMS a site runs, and it sets out the rules our own numbers will follow when we publish them.
Source: HTTP Archive, Web Almanac 2025, CMS chapter, chart "Mobile year-over-year Core Web Vitals performance per CMS"; Chrome UX Report data for July 2025, mobile, home pages of origins with enough Chrome traffic.
What these numbers measure, and what they don't
The Web Almanac's CMS chapter joins two public datasets. The HTTP Archive crawl of July 2025 loads millions of home pages and records which technologies each one uses, including its CMS. The Chrome UX Report (CrUX) for the same month holds what real Chrome users experienced on those sites. A site counts as passing when, at the 75th percentile of its visits, it has a good Largest Contentful Paint (LCP), a good Cumulative Layout Shift (CLS), and a good Interaction to Next Paint (INP) when there is enough INP data. That is the same pass mark Google's tools use.
Four things to keep in mind before you read the chart as a verdict on a platform:
- It is mobile only. Desktop results are usually better. Across the whole web, 48% of sites passed on mobile in 2025 and 56% on desktop, according to the Almanac's performance chapter.
- It covers sites with real traffic. CrUX only reports on sites with enough Chrome visitors, so the smallest sites are not in the sample at all.
- It counts sites, not visits. A busy shop and a hobby blog weigh the same.
- It is a correlation. WordPress runs everything from a static brochure on cheap shared hosting to a heavily customised news site. Wix and Duda host every site themselves on one stack. The chart shows the sites people build on each platform, not what the platform would do with the same site.
The Almanac's authors draw the same line. Their conclusion is that platforms with tightly managed environments tend to produce steadier results, while extensible systems such as WordPress "allow for both excellent and poor implementations". A fast WordPress site and a slow Wix site are both easy to find.
Shopify is the platform people ask about most and the one missing here: the Almanac's Core Web Vitals chart shows ten platforms, and Shopify is not among them. For more recent and monthly figures, including Shopify, the HTTP Archive publishes a Core Web Vitals Technology Report you can filter by platform yourself. We could not verify a current figure from it while writing this, so we quote none.
Loading is where the gap is
Split into the three metrics, the same data shows something more useful than the ranking. Almost every platform does well on responsiveness. Loading is what separates them.
Source: HTTP Archive, Web Almanac 2025, CMS chapter, chart "Mobile year-over-year CMS LCP performance"; Chrome UX Report data for July 2025, mobile.
Source: HTTP Archive, Web Almanac 2025, CMS chapter, chart "Mobile year-over-year CMS INP performance" (its axis label reads FID; the chapter text and query measure INP); Chrome UX Report data for July 2025, mobile.
WordPress sites are about as responsive as Wix sites (87% against 88% with a good INP) but far behind on loading (53% against 81% with a good LCP). Joomla and Drupal show the same pattern in a milder form. For a site owner that is good news: LCP is the metric with the most direct fixes. Server response time, the size and priority of the main image, and render-blocking CSS and JavaScript decide most of it, and you control all three on a self-hosted CMS.
If you run one of these platforms, start here:
- WordPress: Core Web Vitals on WordPress, then slow server response time, because a slow first byte delays everything after it.
- Shopify: Shopify speed optimisation covers apps, images and the theme, the parts Shopify leaves to you.
- Wix and Squarespace: the platform handles most of the delivery, so the levers are images, embeds and fonts; see Wix speed and SEO limits and the Squarespace checklist.
- Any CMS: how to fix Largest Contentful Paint works through the causes in order.
To see where your own site stands in the same Chrome data, run the Core Web Vitals check. It reads your page's CrUX record, or your origin's when the page has too few visits, so the result is directly comparable with the charts above.
How getReport works out which CMS a site runs
Every report detects the platform, and so does our free CMS detector. Detection is passive: we only read what the page itself reveals and never probe paths or guess.
There are two detectors, because they answer different questions:
- The technology detector runs when a report loads the page in Chromium. A hand-written table of signatures looks at the HTML, the response headers, the script and stylesheet URLs, cookie names and a short list of JavaScript globals. A CMS is recognised by evidence such as a
generatormeta tag (WordPress, Drupal, Joomla, TYPO3, Squarespace), platform headers (x-shopify-stageorx-shopidon Shopify,x-wix-request-idon Wix), platform asset hosts (cdn.shopify.com,static.parastorage.com) or globals likeShopifyandDrupal. Each hit carries a confidence and the evidence string, which the report shows behind the technical toggle. - The WordPress detector feeds the WordPress Doctor panel. It reads the fetched HTML for
/wp-content/and/wp-includes/paths, a WordPress generator tag, thewp-jsonAPI link in the HTML or theLinkheader, and the emoji script. Plugins and the theme come from their asset paths. Only once it says "WordPress" does the report look at a fixed list of at most 15 well-known paths.
A site that hides all of these signals, or runs a custom-built system, shows no CMS rather than a guess. That matters for statistics: a CMS breakdown built on guesses would put sites in the wrong bars.
When we will publish our own numbers
Our reports already collect the two halves we need: the detected CMS and the same Chrome UX Report field data the Almanac uses, read through Google's free API. What is missing is the sample. On 1 October 2026 our anonymous rollup covered 18 domains, which is why an earlier data post was postponed. A chart of 18 sites split across ten platforms would be noise with a source line.
When we do publish, the numbers will follow the rules every getReport aggregate already follows:
- Aggregate only, never a site name. The nightly rollup stores counts, never a URL, domain, report id or IP.
- One domain counts once. We take the latest finished report per domain from the last 90 days, so a site checked 200 times weighs the same as one checked once.
- At least 50 domains per bar. A CMS with fewer is shown as "not enough sites yet" with its count, not as a percentage. The same threshold already gates the hosting comparison in reports.
- Sample size and dates on every chart, as
nand a date range in the source line. - Our bias stated up front. People often check their site when something feels wrong, so our sample is not a random slice of the web. Our numbers will describe sites that were checked on getReport, not the web, and we will put them next to the public numbers rather than instead of them.
Before any of that, the rollup needs a CMS dimension, which it does not have today. Once it exists and the first platforms cross 50 domains, the post with our own numbers goes on the blog. Until then, the Web Almanac is the best public source we know, and we would rather link to it than compete with it on too little data.
Sources
- HTTP Archive, Web Almanac 2025: CMS, Core Web Vitals, LCP and INP charts; Chrome UX Report data for July 2025.
- HTTP Archive, Web Almanac 2025: Performance, share of sites passing Core Web Vitals by device.
- HTTP Archive, Core Web Vitals Technology Report, monthly figures by technology.
- Google, Chrome UX Report documentation, how the field data is collected.