Faceted filter URLs: noindex them or let canonicals handle it?

Take a furniture store with around 300 real products but about 4,000 indexable URLs once colour, size and price-sort filters multiply. Half the team wants noindex on the filter pages, the other half says canonicals are enough. I don't think canonicals alone solve it, because they're only a hint, and Googlebot keeps crawling those filter links anyway, so crawl budget still leaks into them. Robots.txt is the wrong fix too: if you block the crawl, Google can't see a noindex tag, and the URLs can sit in the index with no content.

Noindex does free up crawling for category and product pages, but it can also bury a filter page that has real demand, like "black leather sofa". So I'd noindex only the combinations with zero impressions in Search Console over a few months and keep the rest. How are others deciding when a filter page is actually earning clicks?

Agreed on the zero-impression cut, but be careful with that data. A filter URL can show zero impressions just because Google ranks the category page for "black leather sofa" and the filter page never gets a chance to compete. So what matters is whether the query has demand and whether the filter URL is the one that answers it. In Search Console, filter Performance by Page containing the parameter, then compare its queries with what the category page ranks for. If the category page already covers a term well, noindexing the filter loses you very little.

Do that check before you change anything, because once a URL is noindexed its impressions stop reporting, and you can't use the data afterwards to tell whether you made a mistake. Either way, keep the filter URLs out of the XML sitemap. A sitemap listing sends the same signal as internal links do.

The check makes sense, but I think the answer comes down to whether people actually search for that filter combination as a product in its own right. "Black leather sofa" is a real query with its own intent, so that page may be worth keeping indexable even if the category page sometimes ranks for it. Sort orders and price bands like "sofa price low to high" aren't things people type into Google, so you can noindex those with little risk. To sort your combinations into those two buckets, run the candidate terms through Keyword Planner and see whether anyone searches for them.

I wouldn't use Keyword Planner as the deciding step. It gives a rough Google-wide volume for a term, but it can't tell you whether your filter URL is the page that should answer it. Phrases like "black leather sofa" often show modest volume there even though they convert well, so you could noindex a page that's earning sales just because a tool showed a small number.

I'd look at the results page instead. Search the term and see what Google ranks. If it's showing your category page or product listings in the right colour, the filter URL adds little. If it's showing colour-specific landing pages from other retailers, the filter page has a fair shot and should stay indexable.

Fair point on the results page, but what Google shows today partly reflects what your site sends it. If the filter page is indexed with the same thin product grid as the category page, Google may just pick the category page because it's the stronger of the two. The SERP then makes the filter look unnecessary when it's really just not useful as a standalone page.

So when the category page wins, I wouldn't jump straight to noindex. Check whether the filter page has its own title, a heading that names the colour, and a sentence or two of copy that differs from the parent. If it doesn't, you have two real options: give it that or noindex it. The SERP alone won't tell you which one is better. Does your category page already put the colour in its title, or does it just say "Sofas"?