Paginated category pages: canonical to page 1 or self-reference each page?

The old advice was to canonicalize pages 2, 3 and 4 back to page 1, but Google has said it can handle pagination on its own, so a lot of people now self-reference each page instead. The trouble with pointing everything at page 1 is that products linked only from deeper pages can fall out of the index and lose their signals. The case for page 1 is that it keeps near-duplicate pages from competing with each other.

I lean toward self-referencing canonicals, as long as each page shows a real product set. For thin, filter-style pages where page 2 is just more of the same, I'd noindex them instead.

To judge which approach is working, compare index coverage in Search Console against organic landing sessions in GA4, not just rankings. Which signal are you using?

Self-referencing makes sense for pages with real product sets. The rule I'd go by: if a page could work as a landing page on its own, self-reference it. If it just re-slices the same list, like sort orders or a filter that leaves three items, noindex it and let the main category page carry the ranking. Google picks the most useful, distinct page, so near-copies only split the signal. Also make sure products on deeper pages get other internal links, like related-product modules or breadcrumbs, because self-referencing won't help a product that only exists on page 4.

I'd skip noindex for sort orders. They're just re-slices of the same list, so a canonical to the default view says exactly that, and the internal links on those pages still pass value to the main category. Noindex is better saved for filter combinations that have no search demand at all.

Agreed on sort orders. The canonical only works if the default view is indexable and linked from the category nav, because Google can ignore a canonical that points at a URL it doesn't treat as the main version. Check Search Console to see which URL is actually getting the impressions.

I don't think the nav link is the deciding factor. The canonical is only a hint, and Google often honours it when the page is reachable through sitemaps or normal internal links, even if the category nav doesn't point there. What matters more is that the canonical, the sitemap and the internal links all agree. When they don't, Google picks its own version, and the coverage report shows that as "Duplicate, Google chose different canonical." That report is a better test than guessing about the nav.

Agreed on the coverage report, but the status is worth getting exactly right. "Duplicate, Google chose different canonical than user" only fires when your declared canonical and Google's pick disagree. The sitemap is the other place to look, because if a sort-order URL slips in there, you're sending two versions at once and Google tends to pick whichever one has more links. Keep the sitemap to canonical URLs only, then let the report confirm the choice.

Agreed on the sitemap. Lastmod is a signal too, and sort-order URLs sometimes get pulled into feeds or internal tools that stamp them as recently updated. So keep the sitemap to canonicals only, and make sure lastmod moves only when the actual product list changes.

The simplest test I'd suggest is whether someone would plausibly search for that exact URL. If yes, it gets a self-referencing canonical and a sitemap entry. If it's just a different sort order or a filter that barely changes the list, it canonicals to the default view and stays out of the sitemap. Search demand you can actually check in Keyword Planner or Search Console, while arguments about link equity tend to stay guesses. Run the test per URL rather than per template, because one filter might have real demand and its sibling won't.

Search Console has a blind spot when you use it to judge demand. A filter URL that isn't indexed, or doesn't rank yet, gets zero impressions, and that looks exactly like zero demand. So the impressions column can make a real filter look dead just because it never got a chance.

For e-commerce I'd start with Keyword Planner and look at the modifier itself ("waterproof", "under 50", "kids"), since it shows demand regardless of what your site currently ranks for. Then use Search Console to confirm the URL picks up impressions once it's indexed.

Are the filters you're weighing mostly modifiers like those, or more about sort order and stock status?

It depends on whether people actually type the modifier into a search. "Waterproof hiking boots" is a real phrase with volume, so that filter earns its own URL. "Under 50" usually doesn't. People search "cheap hiking boots" or "hiking boots under £50", and the wording shifts by market, so a single price-band page can look dead in Keyword Planner even when the intent is there. Size and colour filters work the same way: plenty of shoppers use them, but few search them as standalone phrases.

So check the full phrase, modifier plus category, instead of the modifier alone. Then look at what ranks for it. If the top spots are filtered listing pages from other retailers, there's demand you can serve. If it's mostly guides and reviews, a canonical to the main category is probably the better call.