In countries where the "old" image search is still used (like Germany) - it is counted as Referral traffic rather than search traffic (not sure why). You could check in Webmastertools what your % of image search is (and compare January & February)

You could also try to check the landing pages report for this referral traffic and compare this with the landingpages in Acquisition >Search Engine Optimisation > Landingpages with secondary dimension "Google Property"

rgds,

Dirk

Hi Michael

The first thing you should do is to define the geography you are targeting in webmaster tools (If you have a generic tld).

if you would block visitors from Brazil on your site the bounce rate measured by analytics will go down. However, Google is not using your analytics data to measure the bounce rate (at least that is what they claim). As al these people get an error message when they try to visit your site,the real bounce-rate will increase rather than go down, making the situation even worse, you just would not be aware of it.

What you could do is set a custom filter in analytics, showing only the traffic from your target country and apply this filter to a new view. This gives you better insights on the behavior of your target audience.

rgds

Dirk

Not sure if what Dmitrii is stating is correct.
If you check the comments here https://www.distilled.net/blog/distilled/distilledlive-london-a-few-thoughts-on-hreflang/ they state:
" hreflang anotations do not consolidate link equity." (source: Maile Ohye (Google's Developer Programs Tech Lead) at SES London) "Hreflang was not designed to consolidate link authority" (source John Mu - chat with David Sottimano)

Also on Moz - Gianluca seems to be convinced of the same - https://moz.com/community/q/will-website-with-tag-hreflang-pass-link-juice-to-other-country-language-version-of-website -

Dirk

In that case it will reinforce the domain (like any external link to any page on the domain).

It's just that a link to domain.com/es/page will not count as a link to domain.com/en/page even when they are "linked" via the hreflang tag. Idem where the domains are different. ex domain.es/page & domain.co.uk/page - a link to .es page will not count for the .co.uk page (and domain) even when they are connected via the hreflang.

Dirk

Google is indexing both - the url which is referring to the image & the image of the url itself. So yes - you can use two different domains. The domain that only contains the images will normally not be indexed (there is nothing to crawl there - the images are only accessible through the main domain).

Personally, if image search is important for you I would never put the images on a domain that you don't own (like e.g. an amazon s3 url). This isn't the case for Booking - the are also owner of the bstatic domain. The reason why they use this domain is probably performance related. Note that they also redirect the root of bstatic.com to booking.com.

Dirk

Are you getting a lot of traffic from image search? Are these specific images worth finding (not just generic stock photo's)?

If not - it probably doesn't really matter. Is there a specific reason why they are doing this? Seems a bit odd for normal images (as it's mixing styling with HTML) - most of the time this is done for background images.

Dirk

Hi,

Did you properly configure your Analytics - normally someone who goes from the main site to the subdomain should not be measured as a bounced visit. If you are using Google Analytics, you can find here how you should change your trackingcode (if you use the async version) - here if you use Universal Analytics.

As google is not using your analytics tool to determine the bounce rate but uses it's own sources, this will have no impact on your rankings.

Apart from this issue, in general it is better to have everything on the main domain. You could check this presentation of Rand on this topic. You should check with your technical people if it wouldn't be possible to move the Magento to a directory of the main domain.

Plenty of sites are using different technology on different parts of the server, without it being visible to the visitors. As long as the user experience is not affected this should not be a problem - so there is no reason to put everything on one technical platform.

rgds,

Dirk

Not sure if you could really call it a best practice - but in Belgium (3 different languages) the normal configuration is not to determine the default language automatically but rather to present a first time visitor a "choose language" page and store the choice in a cookie for future visits. This is mainly for direct visits.

People coming in via search engines use queries in one of the languages, so normally Google will direct them to pages in that language. Again, on first time visit, the implicit choice of language is stored in a cookie.

All pages contain a link to switch to the other language(s) - which also changes the choice stored in the cookie.

Disadvantage of this system is that you add an additional layer to the site (choice of language) - advantage is that error margin is zero.

Systems which are based ip, browser language, ...etc are not 100% reliable - which could lead to unwanted results (in Belgium quite a sensitive issue if you server a Dutch page to a French speaking person - idem for French & Dutch speaking).

Hope this helps,

Dirk

The solution you propose is not going to work. Hreflang is used to identify different language versions of the same page (with minor variations). This implies that each page has a one to one relation with it's equivalent in the other languages.

This isn't the case on your site - the category pages on your US site do not correspond with the category pages on the en-gb site - you can't use the hreflang here (1 page US = 3 pages in en-gb).

You could do this on homepage, product pages and all pages that are more or less equivalent in all languages.

Please also note that hreflang is reciprocal - if you have following hreflang's

you have to put this code on both pages - on http://www.brandname.com/en-gb/product-a & https://www.BRANDNAMEUSA.com/product-a  - it's no use to only put it on the en-gb version as you currently do.

I also have doubts about "es-mx" for a site which is in entirely in English.

Dirk

You can't track this. You can only track the outgoing links. If they just enter google.com (or any other site) in the browser you are unable to track them.

rgds

Dirk

If you have the list of url's you want to check  in csv you could use Screaming Frog: put the string in a  custom filter & crawl the urls based on the list. Applying the filter will only include the url's that contain the string in the source code.

rgds,

Dirk