PRIYA: The owner is reading this as found money. The filters, color, material, price, size, style, spin up a URL for every combination, Google’s indexing some of them, so the logic goes index all of them and collect the rankings. I want to kill that framing before it spreads, because filters don’t add pages, they multiply them, and most of what comes out is the same category sliced thinner.
ELENA: Multiply is the right word and people underestimate it badly. Five filters, five options each, isn’t twenty-five pages.
PRIYA: It’s the product, not the sum.
ELENA: Thousands per category. And the owner sees a handful indexed and assumes the rest are waiting in line to help. They’re not waiting to help, they’re crawl traps, near-identical thin pages the crawler burns time wandering while the clean category page gets less attention.
HANNAH: Hold on, though, “they’re all traps” is going to over-rotate us, and I’ve watched teams nuke their whole filter system on exactly that fear. It’s not all-or-nothing. Some of these combinations are real.
PRIYA: Define real.
HANNAH: Search demand. “Leather sofas” is a thing people type. “Blue velvet armchairs” is a thing people type. Those filter states map to actual queries, and a clean page for them earns its place. What nobody types is “sofas, blue, under eight hundred, modern, two-seat, fabric.” That’s not a query, it’s a filter state someone clicked through. And you don’t have to guess which is which, a keyword tool shows search volume for a combination, and Search Console’s query report shows what people already arrive on, so the demand-backed set is measurable, not a feeling. The line isn’t filtered-versus-not, it’s demand-versus-no-demand.
MARCUS: Right, and I want to push on Elena’s “traps” because if we walk out of here saying filters are dangerous, the owner blocks everything and loses the leather-sectional-sofas page that was genuinely bringing people in. That’s the failure I’d actually bet on happening. The error in “index them all” is the word all. But “index none” is the same mistake pointed the other way, and it’s the one cautious teams talk themselves into.
ELENA: I didn’t say block everything, I said most are traps.
MARCUS: You said it in a tone that gets repeated as “filters bad,” that’s all I’m flagging.
ELENA: Fair.
SOFIA: Can I bring the searcher into this, because it settles the argument from a different direction. Land on “leather sofas,” you get a real selection, you browse, good. Land on the six-filter permutation, you get three products, or one, or an empty page with a “no results” shrug. That’s a terrible arrival, and it’s a terrible arrival on exactly the URLs Marcus and Hannah agree shouldn’t be indexed. So the demand line and the quality line are the same line. The pages worth indexing are the ones that don’t embarrass you when someone arrives.
NOAH: And notice nobody chose to create these pages. That’s the tell for me. The good pages, the leather-sofa page, someone could decide to build that. The permutations weren’t decided, they fell out of the filter system as a byproduct, and byproduct pages almost never carry intent. It’s the same reflex we keep seeing, more URLs read as more chances, except here the URLs showed up free so it feels like a gift instead of a liability.
THEO: So the rule writes itself from what you’ve all said. Pick the filter combinations with real demand, the ones Hannah and Sofia keep circling, make those a deliberate indexable set, canonical, linked, titled like pages you meant to build. Everything else, suppress, canonical to the base category, keep out of the sitemap, handle the parameters so crawlers don’t wander in. Curate the few, smother the many.
AIKO: Theo, “pick the combinations” is fine for ten of them, but this is furniture, the catalog churns weekly, you cannot hand-pick forever.
THEO: Agreed, I meant the criterion, not a manual list.
AIKO: Then say it as a default, because that’s what survives. Indexability is off for filtered views unless a combination is promoted into the maintained demand-backed set. Off by default, opt-in for the ones that earn it. And monitor for parameter URLs leaking into the index, because the day suppression quietly breaks is the day the flood comes back and nobody notices for a month.
PRIYA: That’s the version I can take to the owner. Not “filters are bad,” not “index everything,” but a small promoted set and a default that says no to the rest. The leather-sectional page lives. The six-filter ghost town never gets born.
DANA: Then here’s where we land, and it’s the line Hannah drew and Marcus defended from both sides. We don’t index every filtered URL, because the permutations are thin near-duplicates that waste crawling and greet searchers with empty pages. We don’t block filters wholesale either, because the demand-backed handful, leather sectional sofas and its cousins, are real landing pages. We promote that set into clean canonical indexable pages, and we suppress the rest by default, canonicalized to the base category, out of the sitemap, parameters handled, with monitoring for leaks. Opt-in indexability, enforced by the system, not by someone remembering. The owner’s instinct that some filter pages rank is correct. The leap from “some” to “all,” just because the URLs exist, is the trap, and most of those URLs are liabilities wearing the costume of opportunities.
SOFIA: And the test is dead simple, would this page survive a stranger arriving on it cold. The demand-backed ones survive. The permutations fall apart on contact. That single question sorts the whole catalog.
DANA: Filters hand you URLs for free, and free is the tell, not the gift. Index the ones real people search for, and let the rest stay out of the way.