Faceted Navigation and Pagination SEO for Large Catalogs

The short answer

Filters in a large catalog generate an almost endless number of URLs: color, size, brand, price, sorting and every combination of them. If all of that reaches the index, search engines spend crawl budget on duplicates and important product pages get crawled less often.

A working setup looks like this:

• Landing pages only for combinations people actually search for ("men's Nike sneakers").
• All other filters stay out of the index and, ideally, out of crawling.
• Pagination uses separate indexable URLs, each with a canonical pointing to itself.

Which filter combinations deserve indexing

Decide with data, not with the structure of your filter panel. Check every combination:

1. Demand — people search for it (Google Keyword Planner, Yandex Wordstat).
2. Inventory — the page has enough products, not one or two.
3. Uniqueness — its own title, H1, meta tags and, where possible, a short text.
4. Reasonable depth — one or two dimensions are usually enough: category + brand, category + type. Three or more filters at once rarely have demand.

Keep a "combination pattern → decision" table:

Give indexable combinations static, clean URLs and link to them from menus or internal linking blocks, not only through a JavaScript filter.

How to control parameter crawling

Several tools exist for non-indexable combinations, and they do different jobs:

• robots.txt Disallow — the bot does not crawl the URL. Saves budget, but the URL can still be indexed without content if many links point to it.
• meta robots noindex — keeps the page out of the index, but the bot must fetch the page to see the tag. Never combine it with Disallow: a blocked page cannot be read.
• canonical — a hint, not a directive. Fine for near-identical pages such as sort orders, but it does not guarantee crawl savings.
• Clean-param in robots.txt — a Yandex directive for parameters that do not change content (utm, sessions, sorting).
• Filters without links — parameters change via forms or JavaScript with no <a href>, so bots never discover those URLs.

User-agent: *
Disallow: /*?*sort=
Disallow: /*?*price=

User-agent: Yandex
Clean-param: sort&utm_source&utm_medium /catalog/

Also:

• Enforce a fixed parameter order so ?color=red&size=42 and ?size=42&color=red do not become two URLs.
• Return 404 for filter combinations with no results, or do not link to them at all.

Pagination without rel=prev/next

Google stopped using rel="prev" and rel="next" as a signal long ago, and Yandex relies on its own mechanisms. Pagination has to work on its own:

• Every page has its own URL: /catalog/?page=2 or /catalog/page/2/.
• Self-referencing canonical, not a canonical to page 1. Otherwise products on deeper pages may be discovered less reliably.
• Plain <a href> links to adjacent pages and a few nearby ones.
• Infinite scroll and "Load more" are fine, as long as paginated URLs also exist and are reachable through links.
• Do not repeat the category text on every paginated page; keep it on page 1.
• Titles for pages 2+ can include the number: "Sneakers — page 2".

Common mistakes

• Every filter is indexable, and the index fills with thousands of near-empty pages.
• Disallow and noindex on the same URL.
• All paginated pages canonicalized to page 1.
• Indexable filters reachable only through JavaScript, with no HTML links.
• Links to filters that return zero results.