Google's John Mueller said on Twitter that "anecdotally, it feels like we're already doing much better with getting languages & location right, even without hreflang or geotargeting settings."
This was in response to an excellent question from Glenn Gabe about can Google use MUM AI to figure out page language pairs without hreflang. In which, John added to his tweet that Google is getting better at this "with or without any specific ML model."
Here are those tweets:
Anecdotally, it feels like we're already doing much better with getting languages & location right, even without hreflang / geotargeting settings -- with or without any specific ML model. I could see that all getting better over time, but crystal balls are not my speciality :)
— 🐄 John 🐄 (@JohnMu) January 18, 2022
My feeling is equivalent content across languages is a non-issue. If you search for "fromage" you'll automatically find French content. It doesn't have to match English "cheese" content. Even with brands, if a user is 'french-seeking', it can match the French brand homepage.
— 🐄 John 🐄 (@JohnMu) January 18, 2022
I have no evidence, anecdotal or otherwise, to say either way. I also do not know if MUM can work at a whole page of content level versus a word or phrase level at this point. But what Glenn is suggesting is that since Google seems to be at a place where it can't understand if a piece of content is equivalent when in another language, that this might solve that?
Forum discussion at Twitter.
