I have a question about using link[rel="canonical"]
elements.
Consider a company with an international site com-domain
and a site located inside China cn-domain
(actual domains trigger Stack Overflow spam filters). Both sites have identical content, available in English, French, and Chinese. Although the same content is available on both sites, the canonical address for the English and French pages should be the com-domain
address, and for Chinese it should be the cn-domain
site. I believe that the advice we have been given concerning canonical addresses might be incorrect.
The only difference between the two domains is the default language, where there is no language code in the URL path. On com-domain
it is English, and on cn-domain
it is Chinese.
Consider the following six URLs:
com-domain/path/to/product
(English default)com-domain/fr/path/to/product
(French)com-domain/zh/path/to/product
(Chinese)cn-domain/path/to/product
(Chinese default)cn-domain/en/path/to/product
(English)cn-domain/fr/path/to/product
(French)
I believe that the correct way to advise which addresses are canonical is to add link[rel="canonical"]
elements like so (https
changed to proto
due to spam filters):
For com-domain/path/to/product
and cn-domain/en/path/to/product
:
<link rel="canonical" href="proto://com-domain/path/to/product">
For com-domain/fr/path/to/product
and cn-domain/fr/path/to/product
:
<link rel="canonical" href="proto://com-domain/fr/path/to/product">
For com-domain/zh/path/to/product
and cn-domain/path/to/product
:
<link rel="canonical" href="proto://cn-domain/path/to/product">
However, a third-party is advising us to change the link[rel="alternate"]
elements instead. They advised that all six pags should just have these three alternates:
<link rel="alternate" href="proto://com-domain/path/to/product" hreflang="en">
<link rel="alternate" href="proto://com-domain/fr/path/to/product" hreflang="fr">
<link rel="alternate" href="proto://cn-domain/zh/path/to/product" hreflang="zh">
This strikes me as incorrect, both a misunderstanding of the alternate
relation, and because it breaks the bi-directional relationship that is supposed to exist between alternate pages.
However, I do confess to not being expert in these matters whereas the third-party do claim such expertise.
Could someone clarify which is correct?