usdigitalmarketingindex.com SEO
How can sorting parameters produce duplicate ecommerce category URLs?
Updated November 15, 2026
Because a sort control that changes the URL creates a new address for the same set of products. A category of 300 boilers sorted by price contains the same 300 boilers as the category sorted by rating. Only the order differs. If each sort order has its own URL, and those URLs are linked from every category page, a crawler finds several addresses that all return one list. Google’s ecommerce guidance describes the case directly, with the example of supporting ?order=price “to return the same list of results ordered by price”, under the heading “Avoid indexing URLs with filters or alternative sort orders” (Google’s guidance on pagination and incremental page loading). Sorting is a view of a category, not a category.
Sorting state versus product set
The controls above a product grid sit side by side and behave differently.
| Control | Changes the set of products? | Changes the order? |
|---|---|---|
| Sort by price, rating, newest, name | No | Yes |
| Filter by brand, size, price range | Yes | No |
| Pagination | Shows a different slice of the same set | No |
A sort order answers no question that the default category does not. Someone looking for “combi boilers” is served by the category in any order. That is why a sorted URL is a duplicate in substance: the content is the same, rearranged.
Filters are a separate matter, because a filtered list is a different set and can sometimes correspond to something people search for.
URL patterns and duplicates
A category URL example. The site and category are placeholders.
Preferred: /boilers/combi/
Sort states: /boilers/combi/?sort=price-asc
/boilers/combi/?sort=price-desc
/boilers/combi/?sort=rating
/boilers/combi/?sort=newest
Combined: /boilers/combi/?sort=price-asc&page=2
/boilers/combi/?page=2&sort=price-asc
/boilers/combi/?sort=rating&sort=price-asc
How the count grows:
| Source of variation | Example | Effect |
|---|---|---|
| Each sort option | Four options | Four more URLs per category |
| Sort with pagination | 10 pages by four options | Forty more |
| Parameter order | <!–INLINECODE1–> before or after <!–INLINECODE2–> | Each combination twice |
| Repeated parameters | <!–INLINECODE3–> given twice | More still |
| Default written explicitly | <!–INLINECODE4–> | A copy of the preferred URL |
Multiply by the number of categories. A store with 500 categories and four sort options has 2,000 sorted first pages before pagination is counted.
Google’s URL guidance for ecommerce sites contains two recommendations that reduce this at the source: “Use ?key=value URL parameters rather than ?value, where possible”, and “Avoid using the same parameters twice”, noting that Googlebot may ignore one of the values (Google’s guidance on URL structure for ecommerce sites). Generating parameters in one fixed order removes the reordering duplicates.
Canonical and crawl choices
The pagination guidance names two ways to keep sort variations out of the index: “block unwanted URLs from being indexed with the noindex robots meta tag or discourage crawling of particular URL patterns with a robots.txt file”.
The options, with what each does:
| Method | Effect | Caution |
|---|---|---|
| Sorting without a URL change | No sorted URLs exist to crawl | The default order must be the one in the served HTML |
| Canonical tag on sorted URLs pointing to the unsorted URL | Signals the preferred URL | A hint that Google weighs with other signals; sorted pages are still crawled |
| <!–INLINECODE5–> on sorted URLs | Keeps them out of the index | Pages must stay crawlable for the rule to be read |
| robots.txt disallow for the sort parameter | Stops crawling of the pattern | A blocked page's canonical tag and <!–INLINECODE6–> cannot be read, so do not combine a block with either |
| Sort controls not rendered as crawlable links | Crawlers do not discover the sorted URLs through navigation | Shoppers must still be able to use the control |
Pick one approach and make every signal agree with it. Consistent preference signals for the example:
| Signal | Setting |
|---|---|
| Internal links to the category | <!–INLINECODE7–> only |
| Canonical tag on <!–INLINECODE8–> | Self-referencing |
| Canonical tag on <!–INLINECODE9–> and <!–INLINECODE10–> | <!–INLINECODE11–> |
| Sitemap | <!–INLINECODE12–> only |
| Parameter order | Fixed |
| Default sort | No parameter in the URL |
The same guidance asks sites to “Use the same URL in internal links, sitemap files, and” canonical tags. A sitemap that lists a sorted URL, or a menu that links to one, undoes the preference.
One interaction needs care: sorting combined with pagination. Page two of a category sorted by price is not the same slice as page two in the default order. The pagination guidance says each page in a sequence should have its own canonical URL and that the first page should not be used as the canonical for the rest. So do not point sorted pages two and beyond at page one. Keep sorted URLs out of the index with noindex, or do not generate them as URLs at all.
The products themselves must remain discoverable. Whatever is done with sorted URLs, every product should be reachable through crawlable links in the default listing and its pages.
Validating one preferred category URL
| Check | Method | Pass condition |
|---|---|---|
| Internal links | Crawl the site; list URLs containing the sort parameter | None linked from navigation, or only as the chosen method allows |
| Canonical tags | Inspect sorted URLs in the crawl | Point to the unsorted URL, where that method is used |
| Sitemap | Search the sitemap for the parameter | Absent |
| Index | Page indexing report in Search Console; URL Inspection on a sorted URL | Sorted URLs not indexed; Google's selected canonical is the preferred URL |
| Crawl activity | Server logs or the crawl stats report, filtered by the parameter | Sorted URLs a small and falling share of requests |
| Search appearance | Performance report filtered to URLs containing the parameter | No impressions on sorted URLs |
| Parameter order | Request both orders | One redirects to, or canonicalizes to, the other |
| Product reach | Crawl from the category | Every product found without using a sorted URL |
Is a sorted view ever worth indexing? A page titled for the cheapest items in a category may look like a search target. If a store wants such a page, it should be built as a page in its own right, with a stable URL and content suited to the purpose, not left as a by-product of the sort control.
Filters, which change the product set and can multiply combinations far faster than sorting does, are treated in the guide to how JavaScript faceted filters can expose too many crawlable URL states.
The rule: one preferred URL per category, sort orders treated as views of it, and every link, canonical tag, and sitemap entry pointing the same way.
Add your comment to this answer