Canonicals, hreflang, sitemaps and structured data

Canonicals, hreflang and structured data are signals Google checks, not orders it follows.

Before you read: predict

This item runs on this lesson's Google model, whose rules quote Google's pages. ExamHall's AWS exam page has three URLs: (1) /en/exams/aws-saa, (2) /ar/exams/aws-saa and (3) /exams/aws-saa, where visitors pick a language. Each page's <head> holds hreflang links with full URLs. Page 1 names en → page 1, ar → page 2 and x-default → page 3. Page 2 names ar → page 2 and en → page 1. Page 3 names only x-default → page 3.

Which pages does Google treat as versions of one another?

The 30-second version

Google groups duplicate URLs by content and shows one, the canonical. Redirects and rel="canonical" are strong hints, sitemaps weak; Google still chooses. Google uses an hreflang link only when both pages name each other; a missing return link drops just that pair. JSON-LD earns a rich result only for a listed feature.

Canonicals, hreflang and structured data are signals Google checks, not orders it follows.

How it works

Step 1 of 13

Duplicates. One page often answers at several URLs: tracking parameters, http and https, a sort order. Google shows one URL per group, the canonical.

In the kitchen

a film shown on three channels this week gets one full entry; the other showings point to it → one canonical per group of duplicates (step 1)

Analogy: The editor of a weekly TV listings magazine

In the listings magazine	In Search
a film shown on three channels this week gets one full entry; the other showings point to it	one canonical per group of duplicates (step 1)
a channel's press note naming its showing as the main one is weighed against the others	`rel="canonical"` as a strong hint (step 2)
the Welsh and English broadcasts of a series are cross-referenced only where each channel's listing names the other	hreflang counting both ways (step 8)
readers whose language has no broadcast are pointed to the "all languages" index page	`x-default` (step 10)
a press note's typed fields (cast, running time) print as a boxed panel only for programme types the magazine has a panel for	rich results only for listed features (step 13)

In real terms: each tag is a claim Google checks against the pages themselves: the content, the other versions' tags, its own list of features.

Breaks down when: an editor can phone a channel to ask what it meant; Google never asks, so a tag it cannot read, or a link that is not returned, simply counts for nothing.

Contrasting case: a 301 redirect is not a claim to weigh: the old URL stops serving the page, so the target is the only copy left to show.

Examples, run for real

4.1 ExamHall EXH-4630 → MR !4930: campaign links split one exam page

Campaign emails linked /en/exams/aws-saa?utm_source=…, and ExamHall's pages declared no canonical (EXH-4630). MR !4930 adds <link rel="canonical"> to every page's server HTML, built from the path with the query dropped.

4.1 EXH-4630, no canonical · /en/exams/aws-saa and its 2 utm_ URLs: 1 group of 3 · shown: Google's own choice
4.1 MR !4930, canonical from the path · the same 3 URLs: 1 group · shown: https://examhall.example/en/exams/aws-saa
4.1 MR !4930 · /ar/exams/aws-saa and its utm_ URL: their own group of 2 (translated content) · shown: https://examhall.example/ar/exams/aws-saa

The three English URLs form one group, and the canonical names the plain path; the Arabic URLs form their own group, since their content is translated (steps 1–3, 6).

4.2 Proof domain, chat: a sitemap as the only signal

A chat app's public channel archive answers at /c/general and at /c/general?highlight=42, which shows the same messages. Neither page declares a canonical.

4.2 /c/general and /c/general?highlight=42, no canonical, no sitemap · 1 group of 2 · shown: Google's own choice
4.2 sitemap.xml lists /c/general · strong signals 0, weak 1 · shown: https://chat.example/c/general

With no strong signal, the sitemap's weak one decides between the two duplicates (steps 2, 11).

Try it

Answer first, then check. Nothing is saved.

Question 1 of 6

This item runs on this lesson's Google model, whose rules quote Google's pages. ExamHall's catalog template was rewritten four ways. Each version serves /exams/aws-saa?utm_source=forum and is meant to tell Google that https://examhall.example/exams/aws-saa is the canonical URL.

Which version's canonical does Google not use?

Question 2 of 6

This item runs on this lesson's Google model, whose rules quote Google's pages. A chat app publishes its public channels as read-only archive pages. Google crawls four URLs, all answering 200: (1) /c/general, (2) /c/general?utm_source=newsletter, (3) /c/general?ref=invite and (4) /c/random. URLs 1 to 3 show the same messages; URL 4 shows other messages. A template bug gives every page, URL 4 included, <link rel="canonical" href="https://chat.example/c/general"> in its <head>.

Which URLs can appear in Google's results?

Question 3 of 6

This item runs on this lesson's Google model, whose rules quote Google's pages. ExamHall's exam page has an English version, an Arabic version and a language picker. All three pages name one another in their <head> with full URLs: en → /en/exams/aws-saa, x-default → /exams/aws-saa, and one value, the same on every page, for /ar/exams/aws-saa. The Arabic page is for Arabic speakers in every country. Searchers whose browsers are set to ar-EG, ar-SA and ar-MA should all be sent to it.

Which value fills hreflang="____" in the links naming the Arabic page?

Question 4 of 6

This item runs on this lesson's Google model, whose rules quote Google's pages. A chat app's site adds JSON-LD to four pages, each block with every property its feature's guide requires: (1) the help centre's questions page, "@type": "FAQPage", in the HTML the server sends; (2) a guide to setting up notifications, "@type": "HowTo", in the HTML the server sends; (3) the next "ask me anything" livestream, "@type": "Event", added by a script after the page loads; (4) the livestream after that, "@type": "Event", in the HTML the server sends, with a comma after its last property. As of 2026-10-16, Google's list of structured data features holds Event and no FAQ or How-to feature.

How many of the four pages are eligible for a rich result?

Question 5 of 6

This item runs on this lesson's Google model, whose rules quote Google's pages. ExamHall moves its hreflang links out of the page heads and into a sitemap; /en/exams/aws-saa and /ar/exams/aws-saa now carry none in their HTML. Each <url> entry below holds the page's full URL in <loc> and xhtml:link entries with hreflang="en" and hreflang="ar". A searcher whose browser is set to ar-SA should be sent to the Arabic page.

Which layout does that?

Question 6 of 6

This item runs on this lesson's Google model, whose rules quote Google's pages. A chat app's help article has five versions whose <head> tags all name one another, with full URLs and these values: en, en-GB, de, es-419 and x-default (a language picker). Four searchers' browsers are set to en-AU, es-419, de-AT and fr-FR.

How many of the four does hreflang send to the picker?

Go deeper

Pro tip

When asked how to stop tracking and filter parameters from splitting a page in search, answer in signal order: redirect the URLs that should not exist, put a self-referencing canonical built from the path in the server's HTML, list only canonical URLs in the sitemap, then say Google still makes the final choice. When: the interviewer asks how to avoid duplicate-content problems on parameter URLs. Status: SELF; not run.

Source

Valid as of 2026-10-16. Every rule the model adds is quoted from these pages in lib/google.mjs.

Google Search Central, How to specify a canonical URL with rel=canonical and other methods (named in the row; the page title's quotation marks dropped here): https://developers.google.com/search/docs/crawling-indexing/consolidate-duplicate-urls · google-consolidate-duplicates.txt. "Redirects: A strong signal that the target of the redirect should become canonical."; "Sitemap inclusion: A weak signal that helps the URLs that are included in a sitemap become canonical."; "Google will decide which pages (if any) are duplicates, based on similarity of content."; "link element is only accepted if it appears in the <head> section of the HTML"; "Even though relative paths are supported by Google"; "Don't use the robots.txt file for canonicalization purposes."; "make sure to specify a canonical page in the same language". Supports steps 1–6.
Google Search Central, Tell Google about localized versions of your page (named in the row): https://developers.google.com/search/docs/specialty/international/localized-versions · google-localized-versions.txt. "Each language version must list itself as well as all other language versions."; "If two pages don't both point to each other, the tags will be ignored."; "Google will still process the ones that point to each other."; "other codes that aren't listed in those standards, such as es-419, aren't supported."; "The reserved x-default value is used when no other language/region matches the user's browser setting." Supports steps 6–11.
Google Search Central, Introduction to structured data markup in Google Search (named in the row): https://developers.google.com/search/docs/appearance/structured-data/intro-structured-data · google-structured-data-intro.txt. "all 3 formats are equally fine for Google, as long as the markup is valid"; "Google can read JSON-LD data when it is dynamically injected into the page's contents, such as by JavaScript code"; "You must include all the required properties for an object to be eligible for appearance in Google Search with enhanced display." Supports steps 12, 13.
Google Search Central, Structured data markup that Google Search supports: https://developers.google.com/search/docs/appearance/structured-data/search-gallery · google-search-gallery.txt. "To make your site eligible for appearance as one of these rich results, follow the guide to learn how to implement structured data on your site."; "The actual appearance in search results might be different." Its feature list, as read, includes Article and Event and no FAQ or How-to feature. Supports step 13.

Checked against the tools and docs as of 2026-10-16.

Predict first