When Google announced in March 2019 that it had not used rel=prev/next for indexing decisions “for some time,” the SEO industry assumed Google had developed a superior alternative signal for understanding paginated content. It had not. Google replaced an explicit signal with heuristic inference — using URL patterns, on-page navigation elements, and content similarity to guess which pages form a paginated series. This heuristic approach works reasonably well for standard pagination patterns but fails predictably for non-standard implementations, meaning the deprecation did not simplify pagination strategy — it made getting it right harder because you can no longer tell Google directly how your pages relate.
Google’s Heuristic Pagination Detection System
Without rel=prev/next, Google identifies paginated series through multiple heuristic signals that it pieces together during crawling and indexing. John Mueller confirmed that Google is “able to recognize common types of pagination setups on our own” and can “process those as normal internal links and understand the context from there” (Google Search Central, 2019).
The primary detection signals operate in a layered model. URL pattern recognition is the strongest signal. Sequential parameters like ?page=1, ?page=2, ?page=3 or path-based patterns like /page/1/, /page/2/, /page/3/ provide clear sequential indicators. Google’s systems parse these patterns to identify numbered series and infer the relationship between pages. URLs that use non-standard parameter names (like ?offset=20, ?start=50) are harder for Google to parse heuristically but can still be detected when combined with other signals.
On-page navigation elements provide the second signal layer. Numbered pagination links at the bottom of a page (1, 2, 3… 10, Next) serve as explicit structural indicators that Google can read from the HTML. These links create a sequential chain that Google follows during crawling, discovering each subsequent page through the navigation. The link text itself — numbers and directional labels like “Next” and “Previous” — provides semantic cues that reinforce the sequential relationship.
Content template similarity is the third signal. When multiple pages share the same template structure (identical header, navigation, footer, layout) but differ in the specific content items displayed (different product listings, different article excerpts), Google infers that these pages are variants of the same collection. The template matching is particularly effective for category listing pages and search result pages where the structural consistency is high.
Internal link patterns that create sequential chains — page 1 links to page 2, which links to page 3, without pages linking to non-adjacent pages in the sequence — form the fourth signal. This chain structure differs from hub-and-spoke patterns or random cross-linking, giving Google additional confidence that the pages form an ordered series.
The heuristic system works reliably when all four signals align. Standard implementations using ?page=N parameters with numbered pagination links on consistent templates are detected accurately. The system becomes unreliable when signals conflict: non-sequential URL parameters combined with AJAX-loaded pagination controls and inconsistent templates produce ambiguous signals that Google may resolve incorrectly.
How Paginated Pages Compete With and Cannibalize Each Other Post-Deprecation
Without explicit pagination markup, Google sometimes treats individual paginated pages as independent content rather than series members. Mueller has stated that Google treats paginated pages as “normal pages,” meaning each page in the series is evaluated independently for ranking purposes (Search Engine Journal, 2024). This treatment creates a cannibalization risk that did not exist when rel=prev/next explicitly told Google the pages formed a series.
The cannibalization manifests in Search Console data. A product category with 20 paginated pages may show impressions splitting across pages 1, 3, and 7 for the same category keyword. Google cannot determine from heuristic signals alone that page 1 is the preferred landing page for broad category queries. It sees 20 pages with similar template structure and partially overlapping content, and it tests multiple pages against the query to determine which performs best.
Glenn Gabe’s SEO case study documented a site where 67% of indexed URLs were pagination pages, creating massive cannibalization across category keywords (GSQi, 2024). The paginated pages competed with each other and with the base category page, fragmenting ranking signals across dozens of URLs that should have consolidated on page 1.
The post-deprecation cannibalization risk is highest for category pages where the title tag, meta description, and H1 remain identical across all paginated pages. When page 1 and page 5 share the same <title>Shoes - Category Page</title>, Google has no metadata-level signal to prefer one over the other. Differentiating metadata across paginated pages — <title>Shoes - Page 5 of 20</title> — gives Google a heuristic cue that the pages are numbered variants, but it also dilutes the keyword optimization that most sites want concentrated on page 1.
The Content Accessibility Problem for Items Beyond Page One
Google crawls page 1 of a paginated series with high frequency because page 1 typically sits at low click depth, receives the most internal links, and often has the strongest external backlink profile. Each subsequent page in the series sits at increasing click depth and receives fewer direct internal links. The crawl frequency drops proportionally.
In practical terms, items that appear only on page 5 or later of a product category may never receive adequate crawl attention. Google’s crawl budget allocates resources based on perceived importance, and a page at click depth 5 with no direct internal links except the sequential pagination chain is perceived as low-importance. Products listed on deep pagination pages may take weeks or months to be crawled and indexed, if they are indexed at all.
The accessibility problem compounds for sites with rapidly changing inventories. If new products are added to the beginning of a paginated list (pushing older products deeper), the deep pages change content frequently — but Googlebot visits them infrequently. The result is a persistent staleness gap where Google’s index of deep pagination pages lags significantly behind the actual content.
There is no explicit signal available to tell Google that page 5 is part of a critical series requiring complete crawl coverage. The heuristic system recognizes the sequential relationship but does not infer crawl priority from it. Google treats the pagination sequence as discovered content to be crawled as resources permit, not as an interconnected series requiring exhaustive coverage.
The practical mitigation is reducing pagination depth where possible. Increasing the items per page from 20 to 48 reduces a 500-product category from 25 pages to approximately 11 pages, keeping all content within a crawl depth that Google is more likely to visit regularly. For categories too large for this approach, supplementary internal links from other site sections directly to deep pagination pages (or to the products listed on those pages) provide alternative crawl paths that bypass the sequential chain bottleneck.
Signals You Can Still Send to Reinforce Pagination Relationships
While rel=prev/next is deprecated, several alternative signals help Google’s heuristic system identify pagination correctly and maintain crawl coverage across the series.
Self-referencing canonical tags on each paginated page are the most important remaining signal. Google’s updated e-commerce best practices documentation explicitly recommends that each paginated URL should be indexable and contain a self-referencing canonical tag (Google Search Central, 2024). Mueller reinforced this by warning against canonicalizing page 2+ to page 1, stating that “page 2 isn’t equivalent to page 1, so the rel=canonical like that would be incorrect.”
Consistent URL parameter naming reinforces the heuristic detection. Using ?page=2 consistently across all paginated sections is more reliably detected than mixing ?page=2 in some sections with ?p=2 or ?offset=20 in others. URL consistency gives Google’s pattern recognition a clean signal to parse.
XML sitemap inclusion of all paginated pages ensures Google discovers the full series regardless of crawl depth limitations. Including ?page=1 through ?page=20 in the sitemap with accurate <lastmod> dates tells Google the complete scope of the series and when each page was last updated. This does not guarantee crawling but provides a discovery pathway that supplements the sequential link chain.
Structured internal linking that goes beyond the basic sequential chain improves crawl coverage. Instead of only linking page 1 to page 2, page 2 to page 3 sequentially, include links from page 1 to pages 5, 10, and the last page. This reduces the effective click depth for deep pages and gives Google multiple entry points into the pagination sequence.
The signal that does not work despite common implementation is rel=prev/next itself. Some sites continue implementing it as a hedge, reasoning that it cannot hurt even if Google ignores it. Mueller has confirmed that Google does not use it, and no evidence suggests it provides any heuristic benefit. The implementation effort is better spent on the signals that Google’s heuristic system actually processes.
Does implementing rel=prev/next still provide any benefit as a secondary signal even though Google deprecated it?
No measurable benefit exists. Google confirmed it stopped using rel=prev/next for indexing decisions, and no evidence suggests these tags contribute to heuristic pagination detection. The implementation effort and maintenance cost produce zero return. That engineering time is better allocated to ensuring consistent URL parameter patterns, server-rendered pagination navigation, and proper canonical tag implementation across the paginated series.
Should paginated pages be included in XML sitemaps?
Yes. Including all paginated pages in the XML sitemap ensures Google discovers the full series even when crawl depth limits prevent Googlebot from reaching deep pages through sequential link chains. Each paginated URL in the sitemap should include an accurate lastmod date reflecting when the page content last changed. This provides a discovery pathway that supplements the on-page pagination navigation links.
Does increasing items per page to reduce pagination depth affect ranking performance?
Increasing items per page reduces the number of paginated URLs, which concentrates equity and engagement signals on fewer pages. A category with 500 products showing 48 per page creates 11 pages instead of 25 at 20 per page. The tradeoff is page load time: larger pages with more products load slower, and Core Web Vitals degradation can offset the consolidation benefit. Test page load performance at different item counts to find the threshold where consolidation gain exceeds load speed penalty.
Sources
- Search Engine Land. Google Hasn’t Supported rel=next/prev for a While. https://searchengineland.com/google-no-longer-supports-relnext-prev-314319
- Search Engine Journal. Google Shares Guidance on Pagination for SEO. https://www.searchenginejournal.com/pagination-seo/295627/
- GSQi (Glenn Gabe). What Happens to Crawling and Rankings When 67% of Indexed URLs Are Pagination. https://www.gsqi.com/marketing-blog/pagination-indexing-levels-seo-case-study/
- MarketingSyrup. Google Doesn’t Support rel=next/prev. What to Do. https://marketingsyrup.com/google-doesnt-support-rel-next-prev-for-paginated-content/
- Orit Mutznik. Pagination Best Practices 2024: eCommerce Case Study. https://www.oritmutznik.com/ecommerce-seo/pagination-best-practices-ecommerce-case-study-and-tips-for-search-friendly-pagination
