How Does Google Choose Canonical URLs? Indexing Explained

In a video published three days ago, Google clarified canonical-related issues in indexing. Gary Illyes explained the signals Google uses to choose canonical pages and why duplicate pages can matter for SEO. Below you can watch the video, to which I added Turkish subtitles using vizard.ai. Unfortunately, the audio could not be processed due to technical issues. (The original video is at the bottom of this article.)

In the Google Search Central video, Google explained the part of web page indexing that involves choosing canonicals. It also covered what canonicals mean for Google, gave a short explanation of web page signals and talked about a page’s centerpiece. It described the actions taken on duplicate pages, which points to a new way of thinking about duplicates. Let’s watch and analyze the video together.

What Is a Canonical Web Page?

There are several ways to think about what “canonical” means and what it means to Google. In essence, it is about which version of the content you publish is the strongest. In other words, among similar pages on your website, the strongest one wins…

Canonicalization from Google’s point of view is quite different from what publishers and SEOs usually think. That’s why it is very valuable to hear this explanation from a Googler like Gary Illyes.

Google’s official documentation refers to the process of choosing a canonical as “duplicate reduction” and lists five typical reasons why a site may have duplicate pages.

Five Reasons Duplicate Pages Are Created:

Regional variations: for example, content accessible from different URLs for the US and the UK but essentially the same,
Device variations: a website using different pages for its mobile and desktop versions, making the pages duplicates,
Protocol variations: a site having both HTTP and HTTPS versions,
Site functions: the results of sorting and filtering functions on a category page,
Accidental variations: a demo version of the site accidentally left accessible to crawlers.

Beyond these five points, Gary Illyes gives us two more perspectives.

First Perspective: The Signals a Page Sends to Search Engines

Offering another definition of canonical pages, Gary explains the signals involved in canonical issues during indexing:

“Google determines whether the page is a duplicate of another already known page and which version should be kept in the index — that is, the canonical version.

The canonical version is the duplicate that best represents the group, based on the signals we’ve collected about each version.”

To explain canonical clustering, Gary says this about signals:

“For the most part, only canonical pages appear in search results. So how do we know which page is canonical?

When Google takes the content of your page — or more specifically, the main content or centerpiece of the page — it groups it with one or more pages with similar content, if there are any. This is called duplicate clustering.”

In his explanation, Gary refers to the main content as “the centerpiece of the page.” That’s interesting, because Google has a concept called Centerpiece Annotation, introduced by Google’s Martin Splitt. Although Martin Splitt didn’t explain the concept in detail, I see Gary Illyes’s statement as a clue for us.

In the part of the video where he explains what “signals” are, he continues:

“The search engine picks a canonical version by comparing a handful of signals already calculated for each page.

Signals are pieces of information the search engine collects about pages and websites and uses for further processing.

Some signals are very straightforward, such as site owner annotations in HTML like rel=”canonical”, while others, like the importance of a page on the internet, are more complex.”

For more detailed information on how search engines work, take a look at my article “How Do Search Engines Work?“.

Second Perspective: Alternate Versions of Web Pages

This part is really interesting and noteworthy, because it is an important topic for ranking for multiple variations of a keyword, especially for e-commerce sites.

Sometimes the content management system (CMS) creates duplicate pages to account for variations such as a product’s size or color, which can affect the description. Google may choose to rank these variant pages in search results when they are a better match for the search query.

This is an important consideration, because it can be tempting to noindex or redirect variant pages out of fear of a (non-existent) keyword cannibalization problem. Adding noindex to a page’s variants can sometimes backfire in scenarios where those variant pages could be the best-ranking pages for a more nuanced query that includes different colors, sizes or version numbers.

Original video: