A catalog filter helps someone narrow a set of items. That does not mean every possible combination needs its own search landing page. Decide which combinations support a useful, stable customer task and which exist only to adjust the current browsing session.
Start with a small decision sheet before writing crawl rules. It should connect each URL pattern to its audience, content, maintenance owner, and intended treatment. A broad technical rule can otherwise suppress a useful category along with thousands of low-value combinations.
Google's faceted-navigation guidance, checked September 9, 2026, distinguishes filters that need potential indexing from those whose crawling can be prevented. It notes that large filter spaces consume resources and can slow discovery. Canonical signals and crawl controls serve different purposes; a preferred URL declaration is not a guaranteed crawl barrier.
Map a fictional training catalog
Imagine a publication with 40 startup courses. A visitor can filter by career stage, format, duration, and topic. Selecting one option from each of four groups can create many combinations, including views with no courses and views that differ only by sort order.
The team identifies a durable customer need for “first startup job courses.” It prepares an actual collection with selection criteria, relevant courses, and guidance for that audience. The team also allows visitors to sort the collection by duration, but that sorting choice does not create a new educational topic.
This distinction is editorial before it is technical. A stable collection can explain why its items belong together. A raw combination might simply show the same cards in a different order. Give the former an owner and a clear place in navigation; keep the latter useful as a control without pretending it is another guide.
Write an explicit treatment for each pattern
The paths below are hypothetical. They show decisions to review, not rules to paste into an unrelated site. An engineer must test the actual route implementation and its existing search state before applying controls.
| Example | Reader task | Proposed treatment |
|---|---|---|
| /courses/first-startup-job | Choose training for an entry decision | Maintain a substantive collection as a discoverable landing page |
| ?sort=duration | Reorder the current collection | Avoid treating each sort order as a distinct content destination |
| ?format=video&duration=short | Narrow a browsing session | Evaluate whether any stable audience need justifies discovery |
| An impossible filter combination | No valid matching collection | Handle the invalid or empty URL deliberately |
For crawlable filter URLs, Google's guidance calls for consistent ordering and proper handling of empty or nonsensical combinations, including HTTP 404 responses in the cases it describes. Keep that technical behavior separate from an in-page message shown while someone changes a filter.
Test the edges of the URL space
Review a normal category, a selected valuable combination, a sort-only variation, a duplicate parameter, an impossible combination, and a nonexistent pagination page. Compare actual responses with the sheet. Record any route that produces a plausible page for a nonsensical request.
Inspect how combinations are generated. If two different selection orders create different URLs for the same view, decide how the application will keep its representation consistent. Test existing bookmarks and shared links so a cleanup does not unnecessarily break useful visitor behavior.
Avoid adding more collection pages simply because the filters can generate them. Require a short reader brief and a maintenance owner before promoting a combination. A catalog with ten strong collections can serve its audience better than hundreds of nearly identical landing pages.
Revisit the sheet when the catalog changes
New items can make a previously empty collection useful; removing items can leave a once-helpful page with little substance. Review the actual customer task and available content when those changes occur. Do not preserve a landing page solely because it contributes to the page count.
Use the overlapping-content diagnosis when several collections answer the same question. The decision sheet should leave the team with a manageable set of destinations and a clear explanation for every route family it keeps or constrains.