What are SEO best practices for Java Language Redirections?

[skrauss]

We would like to get some insight on what is the best practice of setting up canonical URLs in the below scenario.

CMS used: Liferay – we believe they are using java. The URL structure at this stage can not be changed to best practices (/en/ and /ar/). Currently the language redirections works like this:

English: https://www.website.com/page1?AF_language=en

Arabic: https://www.website.com/page1?AF_language=ar

Depending how you entered the website last time the root URL will show English or Arabic content without the ‘sufix’:

https://www.website.com/page1

All 3 different URL’s are being indexed on Google - which is causing duplication and confusion.

We have a few ideas:

Have 2 main URLS:

https://www.website.com/page1?AF_language=en and have the canonical set to https://www.website.com/page1?AF_language=en

https://www.website.com/page1?AF_language=ar and have canonical set to https://www.website.com/page1?AF_language=ar

However, how would you handle the root page which does not have a specific language attached. If we need to make a choice we would go with Arabic, as mainly Arabic pages are indexed on Google with the root domain. This way we would (hopefully) retain the rankings for this.

Question: did anybody had to deal with a similar situation? What would you do in a similar situation and why?

Thanks for all your input.

[effectdigital]

Quite an old style of architecture, it's a shame it cannot be changed. Just so others understand a bit more, what you refer to as a 'suffix' is actually called parameters. In a URL, anything following "?" is parameters

If the language on the root is dynamic (it changes) then it's very difficult for you to hreflang to it effectively as it will conflict the the parameter URLs (which are language based) AND additionally, you won't know for certain what language to hreflang to. That also makes canonicals to the root quite tricky IMO

I think what you are already doing, is the best of a bad situation. At least the parameter-based URLs set a designated language which you can rely on to be the same

If you look at this official URL from Google: https://support.google.com/webmasters/answer/182192?hl=en

Scroll down until you find the heading "Using locale-specific URLs" and look at the table underneath of that heading

Parameter-based geo-targeting, is actually the only one of of multiple architectures, which Google put in red text and explicitly warn people away from. Since the site you are looking at has crossed that red line, you may need to manage expectations about results. If they're going to pick the worst possible format and stick with it, without asking you as a consultant what is best, they've kind of shot themselves in the foot there

P.S: Regarding 'actual' redirects, not canonicals. For sites that have proper sub-folder structure, usually you redirect users based on their location, but allow them to flag select to 'escape' the redirections (which can sometimes go wrong). You also usually exempt Google's user-agent ('googlebot') from regional redirects, as they can only crawl from one location at once and otherwise they think areas of your site keep going up and down due to all your redirects. But with your structure, I'm not sure I would even touch redirects. It's in enough of a state as it is without rolling those dice