Welcome to the Q&A Forum

I'm running an SEOMoz campaign for a small site to monitor some tweaks I've made and testing new things out.

Over a year or so the changes were great - organic traffic was rising, domain metrics were too.  Then in October 2012 domain trust, Moztrust and all other link metrics went to zero.  There's no data in OSE for this particular small site and as far as I can tell, no impact on search.  The limited number of inbound links all appear to be intact and there's no reason why the site would be hit by Panda/Penguin style updates.

Why has this sudden loss of data occurred?  Is the site no longer in SEOMoz's databases?  Without any link data it's difficult to tell why this has happened.

If it is no longer in the Moz database, how do I get it back - that data was useful!

As an update (and perhaps by asking reminding me of something I once read), IP delivered content isn't the way forward.  Spiders by definition originate with their own IP and would therefore be directed to content for that locale.

Strategically I think the solution works (from an SEO perspective the categorization element alone is important) but IP served content is not the way forward.

The core question is still brand vs product site then...  arsebiscuits!

I guess another element to consider if using auto-detecting an IP and serving up content is how links are made into the new site.

One excerpt (from the 'Art of SEO no less) indicates that links from the target country will serve to identify the site as being targetted for that country.  Makes perfect sense.

But- if you serve dynamic content, then an inbound link could conceivably just be made to the core domain rather then the absolute link with the country mirror parameters:

i.e.

link from German site -->  productA.brand.com  (and then the user gets re-directed to country mirror)

or

link from German site --> productA.brand.com/de/de  (absolute link)

In the first example I'm not sure where the link juice would ultimately be attributed to.

Hi Russ and thanks for the reply.

That is basically the current model but the products themselves are then split further across business streams which is an issue and could be improved with your suggestion.

However, the real issue is the weakness of the brand in many countries (compared to the brand value and awareness of the main products which are not unique to the company).  When this is coupled with one of 37 product lines being so much more well known then the others, each local site becomes 'categorized' according to this major product.

Looking at the link profile of a particular country site (or checking it on the Google Adplanner) you see that the site categories (i.e. how the site is viewed by search engines) are based around this one product and this hinders performance for the other products served on the domain.

This is why we want to create individual domains/sites (on sub-domains) for each product, so the content and link profile will be dedicated to each individual product.  In other words, a product rather then brand based site.

The worry with auto-geotargetting and serving up content based on IP via content mirrors is that this could be seen as cloaking and may affect search performance.

Although, that said, IBM seem to do a good job of serving localised content and using their ccTLD portfolio (ibm.de --> ibm.com/de/de ).  The only thing is they are so clearly a brand first site and don't need to rely on search to generate traffic.

I hope that is a bit clearer- this is a nightmare topic to articulate!

I read with interest the recent post on international SEO and the top level domain architecture approaches to local content:

http://www.seomoz.org/ugc/folders-vs-subdomains-vs-cctld-in-international-seo-an-overview#jtc135670

The issue I have is a little more complex:

1. The business sells a wide variety of products (37) but one is by far and away the biggest and most popular.  This means that due to the link profile of the various country sites and HQ site, search engines categorise the site according to this product (this is easily seen with the Google Adplanner) and the other product lines suffer as a result.

2. The current architecture is to have a .com site and then individual ccTLD country sites, again with all products on each site.  This creates an issue as in most countries the brand is not strong (compared to the keyword names and search volumes of the products) and so it is not that effective in generating organic traffic.  The .com hogs much of the inbound links and the country sites themselves are not that well optimised for a number of reasons.

A proposed solution has been to leverage the strength of the .com and the  search volume for the product names, and to produce thematic sites based on each product:

productA.brand.com
productB.brand.com
productC.brand.com

In this way, the sites, content and link profiles are aligned around the more desirable products and we can expect improved organic search performance as a result (or at least ensure relevant traffic finds the relevant content fast).

In terms of providing localised content, the plan was to use content mirroring and to then assign each content mirror to a specific geo-location using the webmaster tools console (and other SE equivilents).  This is shown I think in one of Rand's videos.

ProductA.brand.com/de/de  Germany site for product A with unique German content
ProductA.brand.com/fr/fr   French site for product A with unique French content

This makes economic sense to me as to utilise the ccTLDs would result in hundreds of separate sites with all the licence and server considerations that entails.  For example, for product A alone we would have to produce:

productA.brand.de
productA.brand.fr
productA.brand.cn
productA.brand.jp
ect ect ect

This just would not be sustainable in license/server costs alone across 37 products and 24 countries.

However, I saw in a recent presentation at SES London that (auto) geo-targeting is risky, often doesn't work well for SEO and can even be seen as cloaking.

I think the above strategy could still work, but perhaps we should avoid the use of auto-geotargetting altogether and hope the search engines alone do their job in getting users to the right content as we optimise the unique content for each country (and if they don't, ensure our desgn, UX and country selectors do the job instead).

SEO guru consensus is to use the ccTLD if you own it, but as described above, in the real world that just isn't possible or practical given the company's strategic position.

Which leads to the final question- we do own the brand ccTLDs- if they are directed back to the content mirror for the country on the .com, is there any SEO benefit in doing so aside from directing back any link juice associated with the domain)?