A category with 30 products can produce thousands of URLs when color, size, material, price, and sort order are combined. The issue is not the filter control itself. It is the number of URL states a crawler can enter. Google's faceted-navigation guidance warns that these combinations can create effectively unbounded spaces, consume crawling, and slow discovery of useful pages.
Put every URL pattern in one of three buckets
Selected landing facets. A small set may answer a distinct buyer task, such as /chairs?material=oak, if it has a stable collection, useful text, and a reason to be reachable from site navigation. If a filtered URL is intended to be potentially indexed, Google advises normal parameter separators (&), a consistent order for path-encoded facets, and avoidance of duplicate filters. This is a deliberate editorial choice, not a rule that every facet deserves a landing page.
Visitor-only states. A price slider, arbitrary sort order, or a five-filter combination may help an active shopper without adding a useful search landing page. Google recommends preventing crawling when filtered URLs do not need to appear in Search, usually with scoped robots.txt rules or filter states encoded as fragments. Scope the rule carefully: a broad parameter pattern could also block a selected landing facet or individual item links. Google says canonical hints and nofollow can reduce some crawl activity but are generally less effective over time than preventing unwanted crawling. The full guidance explains the tradeoff.
Invalid or empty states. A duplicate filter, impossible combination, or page number beyond the result set should not silently become a valid category page. Google's guidance says an empty facet combination should return a 404 at that requested URL rather than redirect to one common error URL. The screen can still explain what happened and offer links back to useful categories; the zero-result experience guide covers that human recovery. Check framework and platform behavior before promising a particular server response from a single-page application.
Do not mix a crawl block with an index instruction
A noindex directive is read when a crawler can access the page. If robots.txt blocks that fetch, Google may never see the page's noindex instruction. Google documents this interaction directly. Choose a coherent policy per URL class and check existing indexed URLs before changing it. For a previously indexed filtered URL, the transition may need a different sequence from a new unwanted filter URL. Avoid mass rules until you have a sample of current URL patterns and Search Console inspection for important examples.
The key distinction is simple: pagination creates sequential URLs needed to reach more of one ordered collection; facets create alternative collections or views, and most combinations may not warrant crawlable URLs. In an ARCHV commerce or website scope, define the permitted patterns, direct-link behavior, error state, and audit sample together. The decision aims to conserve crawl effort and make important pages clearer; it does not guarantee ranking or indexing.