Adam Hafez
Key takeaways
- Canonical tag usage rose to 68% of desktop pages and 67% of mobile pages in 2025, up from 65% on both in 2024.
- Pages that other pages treat as the canonical version, called canonicalized, rose to 7% on desktop and 9% on mobile.
- HTTP header canonicals remain rare, found on 0.58% of desktop sites and 0.50% of mobile sites, against 64.40% and 64.27% for raw HTML canonicals.
- Canonical mismatches were detected on 0.72% of desktop sites and 0.71% of mobile sites, a small drop from 2024.
- Rendered-change canonicals, where JavaScript alters the tag after load, affect 2.71% of desktop sites and 3.02% of mobile sites.
Canonical tags keep growing, and they are still mostly written the same way. The Web Almanac’s 2025 SEO chapter measured them in the July 2025 HTTP Archive crawl and found another rise in usage, while the failure rates it tracks stayed low. This report covers only the canonicalization section of that chapter.
How many pages use a canonical tag?
The chapter reports canonical usage at 68% of desktop pages and 67% of mobile pages in 2025, up from 65% on both in 2024. It also measures canonicalized pages, meaning a page whose canonical URL points to a different location: 7% on desktop and 9% on mobile, up from 6% and 8% in 2024.
The Almanac describes canonicals as “strong hints” rather than directives, unlike meta robots, so a search engine can still choose a different URL. See our technical SEO hub for related measurements.
Where do sites put the canonical?
Most sites place it in the head as rel=“canonical”, which the chapter says most CMS platforms standardize. HTTP header canonicals can also cover non-HTML files such as PDFs, but they are far rarer.
| Implementation | Desktop | Mobile |
|---|---|---|
| HTTP header canonical | 0.58% | 0.50% |
| Raw HTML canonical | 64.40% | 64.27% |
| Rendered canonical | 66.01% | 66.11% |
Header canonicals were 0.7% on desktop and 0.6% on mobile in 2024. Raw HTML canonicals were on par with 2024, while rendered canonicals rose to 66% on desktop and 66.1% on mobile from 64% and 65%. The chapter reads the raw and rendered gap of around 2% as sites mostly shipping the tag in raw HTML and not removing it at render.
How often do canonicals conflict?
The chapter advises a static, server-side tag because rendering it with JavaScript leaves more room for error. Its inconsistency figures show how much room that is in practice.
| Issue | Desktop | Mobile |
|---|---|---|
| Canonical mismatch | 0.72% | 0.71% |
| Rendered change canonical | 2.71% | 3.02% |
| HTTP header changed canonical | 0.25% | 0.21% |
Mismatches fell about 0.1 points from 2024, from 0.8% to 0.7% on both device types. The larger exposure is the rendered change, where the canonical differs after JavaScript runs, and it is worth checking on JavaScript-heavy sites because indexing depends on which version a crawler reads.
What the report does not show
The Almanac does not break these rates down by CMS or site popularity in this section, and the figures are shares of pages or sites crawled, not of all web pages. Read them as prevalence, not as evidence about ranking or indexing outcomes.
The evidence
- Period
- July 2025 crawl, vs 2024
- Sample
- 16,213,084 sites (12.2M desktop, 15.4M mobile)
Method: As stated on the Web Almanac methodology page, unless a chapter notes otherwise every 2025 metric was sourced from the July 2025 HTTP Archive crawl, drawn from the tables httparchive.crawl.* for date 2025-07-01. Site URLs come from the Chrome UX Report, and desktop and mobile are tested in separate WebPageTest configurations. The SEO chapter reports the share of pages or sites with each canonical signal, and compares raw HTML against the rendered DOM.
Sources
- 1.SEO | 2025 | The Web Almanac - HTTP ArchivePrimary
- 2.Methodology | 2025 | The Web Almanac - HTTP Archive
About the author
Founder
Founder, UpgradIQ, Inc.
Adam Hafez works on technical SEO and search measurement: how pages get crawled, indexed, ranked and now quoted by answer engines. He founded UpgradIQ, which reads Google Search Console and GA4 to tie ranking movement back to the changes that caused it. He publishes what the data supports and states the limits of it.
- Technical SEO
- Search Console and GA4 measurement
- Answer engine optimization
- Structured data
Related reading
Mobile-first indexing, from 2016 test to 2023 completion
Tracing Google's mobile-first indexing announcements from 2016 to the 2023 completion notice, and its documentation on mobile-desktop content parity.
How accurate are sitemap lastmod values? Bing's study
Bing's own study of sitemaps found 18% of lastmod values set incorrectly, usually to the generation date, and Google says it only trusts accurate ones.
Log analysis cut a jewelry site from 16M to 730K crawlable pages
A JetOctopus case study: log files showed Googlebot spending crawl on filter, sort and pagination URLs, and closing them shrank the site to 730K pages.
Hreflang XML sitemap generator
Build a sitemap with xhtml:link hreflang annotations for localized pages, with every alternate listed on every URL as Google's documentation requires.