Ecommerce faceted navigation

Long definition

A clothing site with five facets and ten values per facet generates 100,000+ filtered URLs from one PLP. None of those URLs deserves its own slot in the index, but most ecommerce platforms create them as crawlable links by default. The result is a crawl trap that drains crawl budget away from the PDPs that actually convert.

The decision is which facets to keep indexable and which to suppress. The framework that holds up across verticals:

Indexable facets — high search demand and clean intent. Brand on a multi-brand retailer ("Nike running shoes"), color on fashion ("black leather jacket"), size category on furniture ("queen bed frames"), spec on electronics ("4k monitor"). These get clean URLs (/running-shoes/nike/, not /running-shoes/?brand=nike), self-referential canonicals, custom H1s and intro copy, and inbound links from the parent PLP.

Blocked facets — low intent or pure refinement. Price ranges, ratings, sort orders (?sort=price-asc), in-stock toggles, "show 60 per page". These should be uncrawlable: either nofollow on the facet links, or parameter rules in Search Console deprecated in 2022 (so use robots.txt Disallow: patterns or meta robots noindex instead).

Combinatorial facets — multi-facet stacks like ?color=red&size=m&brand=nike&price=50-100. Almost always blocked. The few that get indexed are curated landing pages with editorial copy, treated like dedicated PLPs.

The implementation pattern: indexable facets render as clean static-looking paths (/black-leather-jackets/); refinement parameters use a query string and are blocked by Disallow: /*?*sort= or Disallow: /*?*price=. Use rel="canonical" on every parameterized URL pointing to the clean PLP equivalent. Combine with noindex, follow on combinatorial pages to keep the link graph passing while removing them from the index.

The signal that a faceted-nav strategy is working: log files show Googlebot spending time on PDPs and indexable PLPs, not chewing through ?ratingFilter=4&pricerange=50-75&sortby=newest&page=12 URLs that nobody searches for.

Common misconceptions

• "Block everything in robots.txt and you're safe." Robots.txt prevents fetching but Googlebot still spends discovery cycles on the disallow check. For deep facet trees, fixing the source (don't generate the links, or nofollow them) beats a blocking bandage downstream.
• "Canonical to the parent PLP solves duplicate content." Canonicals are hints. Google ignores them when the target URL serves visibly different content. A green-only filter page with green products doesn't canonicalize to the all-colors PLP — it just gets indexed anyway and dilutes.
• "Faceted nav is purely an SEO problem." It's also a crawl-budget and infrastructure problem. Sites with millions of facet URLs see Googlebot consume hundreds of GB of bandwidth per month on URLs nobody will ever rank.
• "Modern search engines figured this out." They didn't. Faceted nav remains the single most common ecommerce crawl issue across verticals as of 2026.