Hreflang & duplicate content on a multi-language website

Question

I have a multi-language website like this:

example.com/de/about
example.com/fr/about
example.com/about (English)

On this website there are about 3000 users. Many of them have a profile picture, a name but not much written on their profile page. This means

example.com/de/sarah
example.com/fr/sarah
example.com/sarah (English)

have the same content and only differ in the translated header/footer/bread-crumbs.

I have a similar issue with my blog posts. They are only in English, and the language versions only differ in the translated header/footer/bread-crumbs.

My first thought was to put on all the language versions hreflang annotations and a canonical tag leading to the English version. However, I found an article from Search VIU which says that I should not do this:

If you are using a canonical tag solution on your website, make sure that URLs that have a canonical tag pointing to another URL do not receive hreflang annotations. hreflang annotations are okay for URLs that point to themselves via canonical tag and for URLs that do not have canonical tags.

I found another extreme case here: Canonical tag for untranslated content on a multilingual site where the language sites did not differ at all (not even in the navigation). The solution here was to use 'noindex' for the other language versions. Is this the best solution here as well? It has the downside that people from Germany & France will find a URL to a profile or blog with an English navigation only.

Answer 1

With the rel=canonical across languages (eg, de -> canonical -> en), you're saying that these pages are equivalent and that you have a preference regarding which URL to index. If that's the case, if you think the translated boilerplate doesn't add more value (which might be the case), then that (I'd call it "canonical language") seems like a good setup.

If, on the other hand, the translated boilerplate makes a big difference, then you'd want to use rel=canonical within the language pages (self-canonical per language), and use hreflang between the language versions. By doing that Google can pick the right language version and show that to the user in the search results.

The advantage of using a canonical language setup is that you have fewer pages indexed, which makes it a bit easier to get more content indexed. The downside is you need to pick a canonical language (either globally, or per content piece, it doesn't need to be the same across the site), and that users from search might end up on a page that's hard for them to understand. For profile pages, and user generated content (UGC) in general (you can't easily translate the content), I think there are arguments to be made for either of these setups, so my recommendation would be to look at your content (is it purely UGC, or do the profiles tie in with other functionalities?), at the capabilities you have available (is the rest of the site localized already? is your core functionality well-localized?), and at your users (would they mind landing on a page with other-language boilerplate content, or would they be immediately lost?). For anything non-trivial, I'd also do user-studies, and ideally test with a part of your site to try to get more objective feedback on both of these approaches.