1364470 - Update Country Codes in Hreflang tags

Status: RESOLVED FIXED

(www.mozilla.org :: General, enhancement)

Description

[pgerman]

Attachment: Screen Shot 2017-05-12 at 11.33.18 AM.png

We're currently getting a 10k errors listed in SEM Rush for our HREFLANG tags saying that some countries are not recognized. This impacts our ability to rank in locations that would use these site locales. 

Specifically the following values are not recognized:
-ast
-cak
-dsb
-hsb
-kab
-lij
-son

Can we replace these with the ISO 639-1 values for these countries as this is what google uses?

https://en.wikipedia.org/wiki/List_of_ISO_639-1_codeshttps://en.wikipedia.org/wiki/List_of_ISO_639-1_codes

Comment 1

[Kohei Yoshino [:kohei]]

Those locales don't have the ISO 639-1 value; only -2 or -3:

ast: https://en.wikipedia.org/wiki/Asturian_language
cak: https://en.wikipedia.org/wiki/Kaqchikel_language
dsb: https://en.wikipedia.org/wiki/Lower_Sorbian_language
hsb: https://en.wikipedia.org/wiki/Upper_Sorbian
kab: https://en.wikipedia.org/wiki/Kabyle_language
lij: https://en.wikipedia.org/wiki/Ligurian_(Romance_language)
son: https://en.wikipedia.org/wiki/Songhay_languages

Comment 2

[pgerman]

https://support.google.com/webmasters/answer/189077?hl=en 

Here's Google's Page Identifying what they support.

I recommend that we remove HREFLang tags for locales that are not recognized by google. @Hoosteeno, would you agree?

These locales' pages have between 15 and 500 sessions landing on their pages per month.

Comment 3

[Justin Crawford [:hoosteeno] [:jcrawford]]

I agree with comment 2. Other articles seem to confirm that invalid HREFLang codes can confuse search algorithms.

Comment 4

[Kohei Yoshino [:kohei]]

Bug hreflang is not only for Google. I think we can just ignore those errors.

Comment 5

[Kohei Yoshino [:kohei]]

s/Bug/But/

Comment 6

[pgerman]

Yandex uses this same iso encoding stadard as well, and does bing however bing doesn't use hreflang tags. Duck duck go doesn't have a search engine for these languages. Additonally, Baidu does not use this tag either.

Since these tags are for search engines and the major search engines do not support these locales or don't use this tag, we should remove the Hreflang tags so as to not confuse them. They are the source of a significant portion of our downloads.

https://blogs.bing.com/webmaster/2011/03/01/how-to-tell-bing-your-websites-country-and-language
https://yandex.com/support/webmaster/yandex-indexing/localized-markup.xml
http://searchengineland.com/duckduckgo-adds-auto-search-suggest-to-34-new-international-regions-224372
http://martinkura.com/baidu-international-websites-hreflang-geographic-target-meta-language-tag/

Ignoring this errors will have an adverse effect.

Comment 7

[pgerman]

Hey Eric, is there something we need to do to unblock this?

Comment 8

[Eric Renaud [:erenaud]]

Peter - yes, let's move forward with the plan to remove those identified Hreflang tags so we can eliminate these errors.

Comment 9

[Kohei Yoshino [:kohei]]

Attachment: pull request

I'm still hesitate to remove specific locales from the hreflang list, because it may also be used by others. For example, mozilla.org's Translation Bar [1] actually uses the list. However, given the functionality is currently disabled, the removal should not have negative effects at this time.

I sent a GitHub pull request to make the change.

[1] https://github.com/mozilla/mozilla-infobar/

Comment 10

[Paul [:pmac] McLanahan]

I'm with Kohei on this. I'm failing to see how these meta tags are negatively impacting any actual search engines. I see that an SEM Dashboard is complaining, but it seems that it shouldn't be affecting any real users or search providers. Removing these from the pages would make list of languages for other tools to use on these pages incomplete, and I'm hesitant to do that. Do we have any evidence that these locales are being negatively impacted on any search provider? Our pages are translated into these languages by our fans in these regions. I would very much hate to put up any barrier to accessing their work.

Comment 11

[Kohei Yoshino [:kohei]]

Are there any errors on Google's official Search Console? (Even so, I think Google and other search engines will simply ignore those unknown locales.)

Comment 12

[pgerman]

Yes. There are errors listed in the Google Search console with regards to these tags.

https://pageshot.net/52aZzo1tabYAnkUv/www.google.com

We can see that the search console is reporting back this error for the use of these tags as a result we should resolve this. As a result we should expect this to negatively impact our rankings. 

Google has given us these tools to allow us to see and resolve any issues that may be affecting us. They would not give us the tools to identify and address these issues if they were inconsequential.

Comment 13

[[github robot]]

Commits pushed to master at https://github.com/mozilla/bedrock

https://github.com/mozilla/bedrock/commit/fb65378007d093f60d4334ae6b138bf8ed19f3de
Fix Bug 1364470 - Update Country Codes in Hreflang tags

https://github.com/mozilla/bedrock/commit/14eb52a6c2c9da85f04b6e4eb392e475a4e3c671
Merge pull request #4870 from kyoshino/bug-1364470-hreflang-errors

Fix Bug 1364470 - Update Country Codes in Hreflang tags

Comment 14

[Chris More [:cmore]]

Thanks, Kohei. I've denoted this changed in GA as tomorrow (6/9/2017) and Peter can watch the search console in the upcoming days to see if the errors go away.