跳至正文
技术 SEO 与抓取索引 · Google SEO

Product Variants Not Indexed? Check URLs, Canonicals and Page Models

// / / 光算科技

Having many colors and sizes does not mean every SKU needs its own indexed Google result. First make sure a stable URL opens the correct selection. Then decide whether the product family belongs on one page or on several genuinely useful variant pages. Variant discovery, recognition and independent organic indexing are different outcomes.

Five colors with eight available sizes each produce 40 SKUs only if all combinations exist. That arithmetic neither requires 40 independently indexed pages nor predicts an indexing loss.

Choose the page model before setting canonicals

Google's product-variant documentation distinguishes single-page and multi-page implementations. The following paths are teaching examples, not a requirement to migrate working URLs.

ModelHow the shopper arrivesCanonical approach
One page with color and size selectors/coat opens the family; /coat?color=green&size=s preselects green, size SOne canonical URL represents the overall ProductGroup, normally the unselected base URL
Independent color pages with size selection/coat/green and /coat/blue provide their respective contentEach useful independent page has a corresponding preferred URL; do not force the entire group onto one canonical
A filtered category listing/coats?color=green&sort=price returns a collection of productsManage this as faceted navigation, separately from individual product variants

In the single-page model, a variant URL should preselect the correct image, price and availability and let the shopper add that variant to the cart. In the multi-page model, each page needs complete, self-contained markup for the entities it defines. Google explicitly notes that no single canonical URL represents the entire ProductGroup in that model.

A query string is not a sign of a content farm. Google's ecommerce URL guidance describes both path and query-parameter forms and recommends clear key-value parameters. Avoid unnecessary, constantly changing session or time values. A fragment such as #green is not a separate page for Google indexing.

Audit one product family from URL to cart

Create a working sheet with the URL, selected SKU, image, price and currency, availability, cart SKU and canonical. Open each sampled URL in a fresh session without previous selections, then refresh it. If every address falls back to the default color, the deep links are not reliably selecting variants.

  1. Check that important product and color pages have ordinary crawlable links, rather than navigation available only through click handlers.
  2. Compare the initial HTML and rendered output for the product information, canonical and structured data. A backend field is not proof of what the page delivers.
  3. Keep the internal links, sitemap and canonical consistent for pages intended to be independently indexed. A deep link used to preselect a size does not thereby need its own indexed result.
  4. When stock temporarily reaches zero, preserve useful product and replenishment information where appropriate and show the actual availability. Do not redirect based on an invented stock or sales threshold, or show fake stock to crawlers.

Use ProductGroup to describe relationships

Map the actual group identifier, varying properties and variant relationships using the supported ProductGroup properties, with a unique SKU or applicable identifier for each variant. Include the corresponding Product information and meet the relevant product-snippet or merchant-listing requirements. A ProductGroup type alone is not a complete implementation.

Price, currency, availability, image and the landing-page selection must refer to the same variant. On a multi-page site, off-page entities cannot substitute for the information needed to understand the entities defined on the current page. Google recommends Product markup in the initial HTML where possible; dynamically generated markup can make Shopping crawls less frequent or reliable, especially when price and availability change quickly.

Valid markup does not guarantee a rich result, indexing or sales. Check the output and eligibility separately from the business outcome.

Do not stack robots, noindex and canonical indiscriminately

  • Canonical indicates the preferred version of duplicate or very similar pages. It is a signal, not an instruction Google must follow.
  • Noindex asks supporting search engines to exclude the page from their index. It is not a substitute for canonicalization.
  • Robots.txt controls crawling. Blocking a URL can stop the crawler from seeing the canonical or noindex inside it.

Google's noindex documentation requires the page to be accessible so the rule can be read. Its canonicalization documentation advises against using noindex to select a canonical within a site. Blanket-blocking every color and size parameter can therefore interfere with useful variant discovery.

For sorting, price ranges and combinations that create a large space of low-value listing URLs, use the separate faceted-navigation guidance. Distinguish useful product entrances from unnecessary combinations before applying controls. A robots block alone does not establish that an already known URL has left the index; verify the result and consider robots caching and refetching.

The Search Console URL Parameters tool is no longer available. Google's March 2022 retirement notice announced its removal a month later. Instructions to open that tool and configure color or size are obsolete.

Diagnose indexing by stage

  1. Not discovered: inspect crawlable internal links, intended sitemap entries and the actual URLs.
  2. Cannot be fetched: inspect status codes, authentication, robots and server or CDN behavior.
  3. The wrong selection is fetched: fix parameter handling, rendering, images, prices and cart selection before debating rankings.
  4. Another canonical is selected: compare the user-declared and Google-selected canonical in the indexed data of URL Inspection. A noncanonical size URL not being independently indexed can match the intended design.
  5. Fetched but not indexed: examine usefulness, similar pages and indexing reports. A successful live test neither guarantees inclusion nor predicts the canonical Google will choose.

For a larger catalog, use a controlled sample with Screaming Frog's crawl and verification workflow, then assess duplicate versions against the page models above. Connect long-tail query research to a page plan rather than making a page for every keyword variation.

The useful developer handover is a set of URLs, expected variants, observed output, intended canonicals and retest evidence—not an instruction to index all SKUs. 中文版:商品颜色尺码变体的URL、canonical与索引排查。