Ship a self-referencing canonical on every indexable page, point variant and filter URLs at the main product or category URL, and check URL Inspection for pages where Google chose a different canonical than you declared. It is a hint, and hints get overruled.
A hint, not a command
The rel=canonical link element names the URL you consider the primary version when several addresses show the same or similar content. Google treats it as a strong suggestion, weighed against internal links, sitemap entries, redirects, and how similar the pages actually are. That distinction drives everything else in this guide: a canonical that contradicts your other signals loses the argument, so the job is keeping all the signals telling the same story.
Self-referencing canonicals as the default
Every indexable page should carry a canonical pointing at its own clean URL. This looks pointless until you remember what URLs accumulate in the wild: ?utm_source= from your email campaigns, ?fbclid= from social clicks, session parameters from the menu platform. Each of those is technically a separate URL serving identical content. The self-referencing tag folds them all back into the clean address without you tracking down a single one.
Variants, filters, and sort orders
Store functionality mints URLs constantly. A flower category sortable by price and THC produces /flower?sort=price, /flower?sort=thc, and friends, all near-copies of /flower. Weight options can spawn separate URLs for the same strain. Point every one of these at the base URL. The variations keep working for shoppers; their ranking signals pool on the page you actually want in the results. What you should not canonicalize is distinct products. Two different strains are different pages, and a lazy canonical between them hides one from search entirely.
Canonical, noindex, or 301
Three tools, three jobs. A 301 redirect is for URLs that should not exist anymore; the visitor never sees the old address again. A canonical is for URLs that should keep working but not rank separately, which is exactly the variant-and-filter situation. Noindex is for pages that are not duplicates of anything but have no business in results, like an order-confirmation page. The one combination to never use is noindex plus a canonical to another page: one tag says ignore me, the other says credit that page over there, and Google gets a contradiction instead of an instruction.
Worked example: the sort parameter that ate a category
A store's flower category kept losing its ranking to its own sorted variant. Search results showed /flower?sort=price-asc instead of /flower, complete with an ugly parameter in the visible URL. The cause: the platform emitted no canonical on sorted views, and enough internal "sort by price" links had accumulated that Google picked the parameter URL as primary. Adding a canonical from every sorted view back to /flower, then confirming in URL Inspection that "Google-selected canonical" matched the declared one two weeks later, put the clean URL back in the results. The lesson is the checking step — a canonical you never verify is a canonical you are guessing about.
Key takeaways
- Canonicals are hints; conflicting internal links or sitemaps can override them.
- Self-referencing canonicals neutralize tracking-parameter duplicates for free.
- Filters, sorts, and weight variants point to the base URL; distinct products never do.
- Noindex plus a cross-page canonical is a contradiction, never a strategy.
- URL Inspection shows which canonical Google actually chose. Check it.
Frequently asked questions
What is a self-referencing canonical?
A canonical tag that points at the page's own clean URL. It looks redundant, but it mops up tracking parameters and duplicate paths automatically, which is why every indexable page should ship one by default.
Canonical or noindex for filtered menu pages?
Canonical when the filtered view is essentially the same content as the main page. Noindex when it is a page you never want surfacing that is not a duplicate of anything. Never combine noindex with a canonical to another page; the two signals fight.
Does Google always obey the canonical?
No. It is a hint, not a directive. If internal links, sitemap entries, or content similarity point elsewhere, Google may pick a different canonical, and it tells you which one in Search Console's URL Inspection tool.
Should product variants share one canonical?
Weights and quantities of the same product, yes; canonicalize them to the main product URL. Distinct products such as different strains, no. The test is whether a shopper would consider the two pages interchangeable.
Do canonicals pass link value?
When honored, a canonical consolidates the duplicates' signals onto the target, which is the point of using it. Leaving variants uncanonicalized quietly splits authority your main product pages could be pooling.
Want this done for your brand?
We build cannabis & CBD search visibility with full transparency, for dispensaries, brands, and MSOs.
Get a free cannabis SEO audit ›