You’re right, and thanks just changed the year. Glad you enjoyed the updates!

Hi Rob, I appreciate your question. Generally, if parameter filters are what make the pages unique and paginated then keeping them in the canonical tag makes sense. However, if those paginated filters can be combined infinitely, then it probably makes sense to determine at what point the filter combinations don’t add any more value and stop using canonicals after a certain number of filters has been added.

I’m a little confused by the suggestion to exclude the filter arguments from canonical URLs. That logic seems to assume that the filters are superfluous to the pagination, when in fact that’s generally not the case, since filters are typically applied before pagination.
For example, let’s say we are displaying 25 widgets at a time, so widgets/?p=1 displays widgets 1-25, and widgets/?p=2 displays widgets 26-50.
Now we add a price filter to page two: widgets/?p=2&minPrice=200&maxPrice=499
If that just displayed the same “page two” items minus those that don’t match the filter, then we can conclude that crawlers need not be aware of the filter. But that’s not how web apps typically work. Most likely it would apply price filter on the entire inventory, and then display 26-50 (or whatever page) of that result set. The resulting widgets could easily be 100% unrelated to what was showing on “page 2” without the price filter.
But as you pointed out, including all search parameters in canonical URLs can severely bloat the crawl space. So I’m stumped on how to do this correctly.
Am I missing something?

Thanks, Jonathan. Regarding your question. If you’re putting noindex/follow tags on a page, then you’re telling search engines that you don’t want those pages indexed. Generally speaking, you do want to allow indexation of paginated pages. I would not recommend combining rel=prev/next tags, with the self-canonical and the noindex/follow tags. I do recommend combining the rel=prev/next tags with a self-referencing canonical tag on paginated pages, in most cases. If you know that your paginated pages will end up being low quality somehow (maybe they’re very duplicative in content) then you might consider using noindex/follow. Again though, usually you’d want to allow search engines to index your paginated pages and help them with the rel=prev/next tags too.

Hi Derek, thanks for sharing your experiences and those positive comments!

Thanks Rodrigo! To your question, each of your paginated pages should ideally have a self-referencing canonical tag.

Thanks, Joyce. The self-referencing canonical tag is setup on paginated pages so Google has the choice of ranking any page they decide is relevant to a query. If you canonical all paginated pages to the root, then only the root will end up ranking. However, you’re saying that you really just want the root page ranking and not any of the paginated pages.
I’d still say let Google decide that because there are likely queries you’re not thinking about (often because they’ve never been used before or are long tail) and put a self-referencing canonical on each paginated page. You could leave the canonical off the paginated pages, and just use a canonical on the root page. This would still allow Google to rank a paginated page if they wanted but would send a stronger signal for the root page.