usdigitalmarketingindex.com SEO
How should canonical tags be configured across a paginated listing?
Updated November 15, 2026
Each page in the sequence should carry a canonical tag pointing to itself. Page two is canonical to page two, page three to page three. Pointing every page at page one is the configuration to avoid, because the later pages list different items and are not duplicates of the first. Google’s guidance says so in two sentences: “Give each page a unique URL”, and “Don’t use the first page of a paginated sequence as the canonical page. Instead, give each page its own canonical URL” (Google’s guidance on pagination and incremental page loading).
Distinct page-two content
A canonical tag says this page is a copy of that one. Page two of a category is not a copy of page one. It shows the next set of products or articles, which appear on no other page of the listing.
If page two declares page one as canonical and Google accepts the hint, page two is treated as a duplicate and set aside. The links on it to products 25 through 48 are then on a page Google has been asked to disregard, and those products depend on being found some other way.
Google’s guidance treats the pages as separate: it notes that “URLs in a paginated sequence are treated as separate pages by Google”.
Why canonical-to-page-one is the wrong default
The practice has an understandable motive, which is to concentrate attention on the first page. Three problems come with it.
- It misdescribes the pages. They are not duplicates, and Search Console’s documentation notes that a duplicate page must be similar to the canonical (Search Console’s page indexing report documentation).
- It can hide deeper items. Products and articles linked only from later pages lose a crawl path.
- It may simply be ignored. A canonical is a hint, and one that contradicts the content may not be followed, leaving the outcome unpredictable.
There is one case where a canonical to a single URL is right: when a view-all page exists that contains every item, loads acceptably, and is the page you want in search. Then the component pages are partial copies of it, and pointing them at the view-all page describes a true relationship. This applies to the view-all page, not to page one.
Variants of a paginated page are a separate matter. Page two sorted by price, or filtered by color, lists a different slice of items from the plain page two. Google’s guidance says to “Avoid indexing URLs with filters or alternative sort orders”, and names two tools for that: the noindex robots meta tag, or a robots.txt rule for the URL pattern. Keep those variants out of the index with one of the two. A canonical tag is the weaker tool here: it is a hint, and the pages it would join do not list the same items.
Pagination link discovery
Canonical tags describe the pages. Links are what let a crawler move through them. The same guidance asks site owners to “include links from each page to the following page” using anchor elements with an href, and adds that linking back to the first page from the others can hint that the first page is the better landing page.
A page-one and page-two illustration. The URLs are examples.
Page one:
<head>
<title>Trail running shoes</title>
<link rel="canonical" href="https://www.example.com/trail-running-shoes/">
</head>
<body>
<!-- products 1 to 24 -->
<nav aria-label="Pagination">
<a href="https://www.example.com/trail-running-shoes/?page=2">Next</a>
</nav>
</body>
Page two:
<head>
<title>Trail running shoes</title>
<link rel="canonical" href="https://www.example.com/trail-running-shoes/?page=2">
</head>
<body>
<!-- products 25 to 48 -->
<nav aria-label="Pagination">
<a href="https://www.example.com/trail-running-shoes/">First</a>
<a href="https://www.example.com/trail-running-shoes/">Previous</a>
<a href="https://www.example.com/trail-running-shoes/?page=3">Next</a>
</nav>
</body>
Points to note in the illustration:
- Each canonical names its own URL.
- Page one has a single address. The form with ?page=1 should redirect to it, or canonicalize to it, so the first page does not exist twice.
- The page parameter is a real query parameter. The guidance says not to use fragment identifiers for page numbers, because “Google ignores fragment identifiers”.
- The same title on both pages is acceptable. The guidance says pages in a sequence “don’t need to follow” the usual advice on distinct titles.
- The links are ordinary anchors, present in the HTML.
The older rel next and rel prev link elements are not required. Google’s guidance states that it “no longer uses these tags, although these links may still be used by other search engines”.
Validation
Check a paginated listing against this table.
| Check | Pass condition |
|---|---|
| Canonical on page one | The clean listing URL |
| Canonical on page two and later | That page's own URL, with its page parameter |
| The page-one variant with ?page=1 | Redirects to, or canonicalizes to, the clean URL |
| Sorted and filtered variants | Kept out of the index: noindex, or a robots.txt rule for the pattern, not both |
| Next and previous links | Anchor elements with href, present in rendered HTML |
| Page numbers | In the query string or path, never in a fragment |
| A page beyond the last | Returns a not-found status, not an empty 200 |
| URL Inspection on page two | User-declared canonical is page two |
| A crawl from page one | Reaches the last page and every item |
Run the last check after any change to the listing template. It is the one that shows whether deep items are still reachable.
A related caution about noindex: applying noindex to pages two and beyond is sometimes suggested as an alternative. It keeps those pages out of results, and it is not what Google’s guidance describes. Self-referencing canonicals and working links are.
How the pagination itself should be designed for a category with thousands of products, including page size and navigation, is covered in the guide to search-friendly pagination for a category with thousands of products.
The rule: each page in the sequence names itself, links to the next, and has one address.
Add your comment to this answer