Google says rel="next" and rel="prev" are unsupported and have no effect on indexing [1]. A paginated archive is crawled and indexed like any other set of pages: each page needs its own URL, a real link to the next page, and a canonical that points at itself. Filter and sort variants are where most of the remaining decisions are.

What changed, and what it means for old tags

For years, rel=next/prev was the markup for "these pages belong together." Google's documentation now lists it as unsupported [1]. That is a statement about Google. It is not an instruction to add the tags, and not a reason to rewrite templates that already carry them. If you are building a new template, leave them out. If you have them already, removing them is optional cleanup.

What Google recommends instead is ordinary web plumbing, covered below.

Google recommends sequential <a href> links from each page to the next page [3]. The link needs a real URL in the href attribute. A click handler on a <button> or a <span> does not count.

Before:

<button onclick="loadPage(3)">Next</button>

After:

<nav aria-label="Pagination">
  <a href="/guides/?page=1">1</a>
  <a href="/guides/?page=2">2</a>
  <a href="/guides/?page=3" aria-current="page">3</a>
  <a href="/guides/?page=4">4</a>
</nav>

Each page should be reachable through plain links in the HTML a crawler receives, not only after scripts run. Whether you use ?page=2 or /page/2/ does not matter. Pick one and use it everywhere.

Page 1 is the case to check in your CMS. Many systems serve it at both /guides/ and /guides/page/1/. Link to one version, and make the other redirect or canonical to it. How you do that depends on the platform, so test what yours emits.

2. Give each page a self-referential canonical

Google recommends a self-referential rel="canonical" on the canonical page itself [2]. For pagination, that means page 3 declares page 3:

<!-- on https://example.com/guides/?page=3 -->
<link rel="canonical" href="https://example.com/guides/?page=3">

Page 2 and page 3 list different items, so they are not duplicates of page 1. Pointing every page's canonical at page 1 is a common mistake in this area. Also check that the canonical URL exactly matches the URL you link to: same protocol, host, trailing slash, and parameter order. Our canonical tags guide covers the other ways canonicals go wrong and how to see which URL Google picked.

3. Keep filter and sort URLs out of the index

Google recommends avoiding indexing URLs with filters or alternative sort orders when you implement pagination [3]. A category with five sort options and ten filters can produce thousands of URLs that list the same items in a different order.

Decide per URL pattern:

URL patternExampleTreatment
Plain paginated page/guides/?page=3Indexable, self-canonical, linked
Sort order/guides/?sort=price&page=3Keep out of Search
Filter combination/guides/?color=red&page=2Keep out of Search unless it is a page you built to rank
Tracking parameters/guides/?utm_source=xCanonical to the clean URL

Be careful with noindex. Google says it is not recommended for preventing canonical selection within a single site, because it completely blocks the page from Search [2]. Use it only when you want a page out of Search entirely. It is not a way to say "this version, not that one." For choosing between duplicate versions, see canonical vs redirect vs noindex.

4. Give infinite scroll a paginated fallback

Google says its crawlers do not click buttons or trigger JavaScript functions that require user actions, so infinite scroll needs a paginated version if you want the content crawled [4]. A "Load more" button that only runs JavaScript falls under the same limit.

The pattern that works is progressive enhancement. Serve real paginated URLs, then let scripts upgrade the experience:

<ul id="results">
  <!-- items 1-20 rendered in HTML -->
</ul>

<a id="more" href="/guides/?page=2">Load more</a>
document.getElementById('more').addEventListener('click', async (e) => {
  e.preventDefault();
  const url = e.currentTarget.href;
  // fetch(url), append the new items, then update the link's href to page 3
  history.pushState({}, '', url);
});

Without JavaScript, the link goes to ?page=2, which is a normal crawlable page with its own canonical and its own link to ?page=3. With JavaScript, users get the scroll behavior. Check that the paginated URL, opened directly, renders the right items without any scroll.

5. Optional: describe the list with ItemList

Schema.org's ItemList documentation says numberOfItems should represent the entire list even when the list is split across multi-page pagination [5]. If you mark up a paginated list, the count is the total, not the number on the current page:

{
  "@context": "https://schema.org",
  "@type": "ItemList",
  "numberOfItems": 240,
  "itemListElement": [
    { "@type": "ListItem", "position": 41, "url": "https://example.com/guides/post-41/" }
  ]
}

Positions should continue across pages. Page 3 with 20 items per page starts at 41. This markup is optional, and the notes behind this article do not show that any engine requires it.

6. Sitemaps

Links are the primary path to deep pages, and a sitemap is a second one. List the canonical URLs you want indexed, and leave out the filter and sort variants you decided to exclude. How to build and submit an XML sitemap covers the format and validation.

How AI crawlers fit in

The documented points are narrow:

  • OpenAI says OAI-SearchBot is used to surface websites in ChatGPT search results, and that a site that disallows it will not be shown in ChatGPT search answers [6]. If your robots.txt has broad Disallow rules on archive or parameter paths, check they do not catch the paginated URLs you want crawled. Checking AI crawler access walks through the test.
  • Cloudflare's AI Search documentation says its crawler can discover URLs from links, sitemaps, or both. Its discover crawl can follow up to 100,000 link hops, with a default depth of 5 [7].

The second point describes one product's configuration, not a rule for every AI crawler. In our reading, the safe takeaway is that links and sitemaps are the discovery paths documented so far, which is another reason to keep pagination as plain links and to list canonical URLs in a sitemap. Other vendors' crawl depth is not documented in the sources we have.

How to test the setup

  1. Fetch page 2 without JavaScript: curl -s "https://example.com/guides/?page=2" | grep -E 'rel="canonical"|href="/guides'. Confirm the canonical is page 2 and the next-page link is present in the raw HTML.
  2. Follow the "next" links from page 1 to the last page. Every step should be a normal link returning 200.
  3. In Search Console's URL Inspection, compare the user-declared canonical with the Google-selected canonical for a deep page.
  4. Search for a sort or filter URL pattern in Search Console's indexing reports and see whether any have been indexed.

What to do next

  • Check that every paginated page has a self-referential canonical matching its own URL.
  • Replace any button-only "Next" or "Load more" with an <a href> to a real URL.
  • List the filter and sort parameters on your site and decide which stay out of Search.
  • Make sure robots.txt does not block the paginated paths you want crawled.