Every store has a set of URLs that could plausibly be either a real category or a filtered view of one. Getting the split right determines how much of your catalogue is indexable and how much of it is crawl waste.
The test is simple: a category earns a page when people search for it. A filter is a convenience for people already browsing.
Applying the test
Take “waterproof hiking boots”. If that phrase has real search demand and competitors have built pages for it, it deserves a category — a crawlable, indexable URL with its own title, description and content. If it exists only because your site happens to have a waterproof attribute, it is a filter.
The way to check is to search the phrase. If dedicated pages rank for it, the market has already decided it is a category. See when a keyword needs its own category page.
What each should look like
A category page
- A clean, static URL without query parameters.
- Crawlable and indexable, linked from navigation or a parent category.
- Its own title, meta description and introductory content.
- Self-canonical.
- Enough products to justify existing.
A filtered view
- Typically a parameter URL.
- Not indexed, and usually not worth crawling.
- No unique content of its own.
- Canonicalised to the unfiltered category, or blocked, depending on volume.
The mechanics of keeping filters under control are covered in faceted navigation SEO.
Where stores get it wrong
Indexing everything
Leaving all filter combinations crawlable generates thousands of near-identical URLs, consumes crawl capacity and dilutes the real categories. See index bloat.
Indexing nothing
The opposite error, and more costly than it appears. Blanket-blocking every filtered URL means genuine demand — colour, size, material, brand combinations people really search — has no page to satisfy it. Those are lost categories.
Promoting filters without content
Making a filtered view indexable without giving it a distinct title and copy produces a page that duplicates its parent. Promotion means building a real page, not flipping a switch.
Categories with three products
A category needs enough inventory to be worth landing on. Splitting a small range across many categories produces thin pages that compete with each other. See category vs subcategory.
Promoting a filter to a category
- Confirm demand exists and that the combination has enough products.
- Create a static URL rather than indexing the parameter version.
- Write a genuine title, description and introduction.
- Link to it from the parent category, not only from the filter interface.
- Make it self-canonical and confirm the parameter version still canonicalises to the parent.
Step four is the one most often skipped. A promoted category reachable only by clicking a filter is effectively an orphan. See ecommerce internal linking.
Frequently asked questions
How many filter-derived categories should I have?
As many as have genuine demand and enough stock, which is usually a handful per department rather than dozens. Each one is a page you must maintain.
Do I lose anything by not indexing a filter?
Only if people search for it. That is exactly the test, and it is worth checking rather than assuming in either direction.
Should filter URLs be blocked in robots.txt?
Where volume is very large, blocking saves meaningful crawl capacity. Where it is modest, canonical tags are usually enough and are less risky. See canonical vs noindex.
What about categories that overlap?
Overlapping categories targeting the same search compete with each other. Check for it directly. See keyword cannibalisation audit.
Consulting CTA
If you are unsure which filtered views deserve real pages, book an SEO consultation.