“Matt Cutts: That’s a good question, and I am not 100 percent sure about the answer. I can certainly see how there could be some loss of PageRank. I am not 100 percent sure whether the crawling and indexing team has implemented that sort of natural PageRank decay, so I will have to go and check on that specific case. (Note: in a follow on email, Matt confirmed that this is in fact the case. There is some loss of PR through a 301).”

From http://www.stonetemple.com/articles/interview-matt-cutts-012510.shtml

2. I’d say this was likely a strong factor, but just one of many. Redirecting pages to what sound like orphan pages on the ccTLDs was probably the main issue.

3. Well if you have server logs that show Googlebot hitting all old URLs that’s a start, but how can you tell that Googlebot hit any and all external URLs linking to those pages?

4. Don’t get me wrong – this is a fascinating case study, but the problem is that there are so many dynamic variables in the mix that is is virtually impossible to test what you tried to test. If there wasn’t a full site on the ccTLDs then I’d have to say in my opinion you’re comparing apples with oranges.

I’d say that if you created a new .ie website with a full architecture, migrated the relevant pages over (301s etc), and then created strong interlinking between the .com and .ie that you’d likely rank well. The thing you cant transfer over is the domain authority which likely comes from links pointed at the root of your .com. Well, you can transfer some of this (via interlinking), but you’ll never be 100% sure how much.

I applaud you for taking the risk with this, and I think there’s still some tests you could/should consider that have a lower risk level. I’m thinking a parallel UK site might be a good test – I’d say any gains would far outstrip any duplication risk.

Of course all IMHO

1. We used 301′s – I was of the opinion that they pass on all link juice without the source pages having to be re-spidering.

2. This is our conclusion as well, that it comes down to domain authority

3. We proof positive that Google saw and recorded the redirects.

4. You are right that the test wasn’t complete but was the best we could do. It did have a home page etc, however the site structure was not complete. I am not sure if this invalidates anything as the redirected pages were getting the same link love from the .com domain as they were getting before the redirect.

Would love to chat about this sometime

Did you take into account that Google penalise a page if there’s a redirect?

That’s not correct AFAIK.

Caelen I think there’s a number of very serious issues with your test that make it virtually impossible to draw any conclusions from what happened:

1. Pages rank very often based on their backlinks. Even if only a small number of the original pages had backlinks, there’s no way to know if Google had recrawled all the external pages and re-applied their links;
2. As Leo touched on, domain authority is likely a big factor, and the ccTLDs likely had little domain authority;
3. Large sites and wholesale redirects dont work so well. Basically if you dont manage your redirects then there’s a strong chance that Google never actually saw deeper redirects. Say you have a site with 4 tiers, and you redirect tiers 2-4 to a new domain at the same time. It’s likely that Google will crawl tier 2 more frequently than tiers 3 and 4. They will see the redirects on tier 2 and move over to the new URL. But the problem is that many pages at tiers 3 and 4 no longer have internal links from the original domain and become orphans. Google may never recrawl those deeper tiers to discover their redirects as a result.
4. Large sites often depend on internal architecture to achieve high rankings. It sounds like you may not have built this test in a way that replicates a real move over (but forgive me if I’m wrong). Did the country sites have full navigation/homepages etc? If not then the internal architecture of the country sites may also have contributed to poor results.

There are other workarounds you could look at Caelen – you could consider running near-dupes on multiple domains. I know it might sound whacky, but sometimes this can work if you can mash things up enough.

Last thing – I’m not disputing your conclusion that ccTLDs aren’t the optimal solution for you, but I think you should take care relying on the results of this test to determine that. I don’t think the test is valid TBH.

Next time we meet we should sit down and discuss geotargeting – it’s an area I really enjoy tinkering in.

Rgds
Richard

No we haven’t. We are now of the opinion that country specific TLDs aren’t the way to go for us and we are going to abandon the approach. We think that building more authority into the .com domain is the best strategy for us.

This is partly because the advantage of country specific TLDs are unproven (in our case) and even if they were the strategy would be a long term one as we have historically found that it takes a very long time to build up authority in a domain and the priority that Google seems to apply are in flux.

Have you seen this work elsewhere?

Very good point about user’s behaviour. We didn’t take this into consideration and certainly it would make for a good study.

Regarding the redirects, we use them extensively and have seen no evidence of a Google penalty.