|Wrong cache showing in Google. Will this resolve itself?|
|It isn't at all unusual for Google to be confused when it finds a new site on the same IP, have been seeing this for many years. It usually sorts things out in 2 or 3 weeks. [webmasterworld.com...] |
We relaunched a batch of sites with different ccTLD's with the same or similar content, that were previously cached correctly. They sit on the same virtual server [ Amazon Elastic ], so I assume they have the same IP address, but I'll double check.
They are not caching correctly now.
.com and all other ccTLD sites are now caching as .com.au . In the previous month, they were all caching as .co.uk. It's quite a dance and the traffic had previously tanked in the cut over due to duplicate content [ canonicalization issues , which are now resolved ].
A search around the forums brings up many theories, and many instances similar to this, but the above post, is what I want to believe. That is, that it will resolve in time.
Most folks don't come back to tell the tale after things resolve, so it would be good to get some testimony from members who have experienced this phenomenon and how they believed it resolved, or that they fixed it.
Is anyone able to throw some light on this to help others going through a similar experience.
How long has it been so far, Whitey?
There have been 3 phases :
1. 12-5 weeks ago. Relaunched sites on multiple IPs , same server. Wrong cache showing and canonicalisation issues. .co.uk was shown on all sites
2. 5-1 week ago. Moved to new server, 1 IP address for all sites. Canonicalization issue fixed on 1 site; .com . Sites caching normally to their respective content, throughout the group of sites, but indexed pages on .com tank, and crawl rate remains low.
3. 18 June / yesterday ; all sites home pages are cached and show .com.au as the website incorrectly for 4 out of the group of 5; .com.au being normal.
I would think this is a bug, because Google filtering should take care of what shows in the SERPs
Stuff like this is why all my sites and all my clients sites are set to NOARCHIVE.
Google's cache display is handled separately from the search display, and caches and serps are often served from different data centers.
Several years back, when I watched site changes a lot more closely than I do now, I'd noticed that organic search results respond to onsite changes slightly faster than serps display does, and that cache display is slower yet. As I remember it, cache display might lag several days behind serps display. Caches also might experience glitches which have no relationship to rankings.
Keep in mind that Google splits some of its databases into multiple areas which are then managed by other databases... and the order of changes is necessarily predetermined in a way that preserves the integrity of data.
Layman's description here... this means if you've just missed an early part of a sequence of cycles, it might take a while for everything to catch up. Within this scheme, overlapping changes can create confusion and may slow things down overall.
Your situation also involves DNS propagation issues and latency, server setup issues, internationalization, etc. I can imagine that each of these involves its own cycle of assessment.
Here's a thread... interestingly also involving ccTLDs... that might have some parallels to your situation. The Google infrastructure at the time of the thread was probably different than it is now, but the discussion might provide some clues....
Google Caching Wrong Site?
Sep 6, 2011
|The obvious "explanation" would be that the two domains share the same IP address (do they?) and something in the server set-up is technically incorrect. Alternately, there could be a technical bug on Google's end - they used to happen years ago with shared IP addresses. |
If the sites are not on the same IP address, then I'm out of ideas except for a total Google bug. If that's what it is, Google usually fixes them pretty quickly. They seem to be related to corrupted areas when they move data around their servers.
Sorry for the off thread post but,
Dang I miss Ted and his witty humor. I was reading one of his post yesterday...:(
Thanks for your comment, bwnbwn. I miss him too. :(
|.com and all other ccTLD sites are now caching as .com.au . In the previous month, they were all caching as .co.uk |
Sheesh.... Just an update to this, the sites are now all caching as .com.
My gut feel is that Google is working out which one of the group of sites is the main site for global SERP results, because they are not sufficiently differentiated or authoritative on the separate ccTLD's. Search traffic remains a fraction of the pre migration to the new platform levels.
It's not perfect, but as an interim step to de-confuse Google we're going to add the ccTLD extension to each site's branding in the title tag :
e.g. blah blah blah - sitename.co.uk
and so on.
Also, differentiate the [home and inner] landing pages to be specific to each ccTLD market. At this stage it will still need to be scripted content as these are large sites [ not huge ].
|Google's cache display is handled separately from the search display, and caches and serps are often served from different data centers. |
A subtle variance of thought on my part here. I just wonder if this cache confusion throws up an insight into how Google treats duplicate content in it's search algorithm, even if they work off of different data infrastructures to produce and rank results.
Interestingly, the URL's and sub folders are all showing correctly in the search results.
Any more thoughts out there ?
I'm assuming that these are all English language sites. What differentiates the sites from each other, beyond the accent you'd use if you read them aloud? ;)
Do they have any individualized content? Distinctive spelling? Place names? Currency values? Local news?
Do they contain any different information for their different locations? Are the inbound links from independent sources and appropriately geo-local?
Any other reasons for their being on separate ccTLDs?
Are they heavily interlinked?
are you using any link rel elements?
Some good questions / and points raised :
|Do they have any individualized content? Distinctive spelling? Place names? Currency values? Local news? |
Not sufficiently. This is an ecommerce site in EN and only US/EN varies. However the .com site is intended for a global facing audience, except for uk , au and nz. Without full knowledge of the .ca language differentials, I have lumped this in with the US. We don't cater for PANIBS/Southern Asia [ Pakistan / India / Sri Lanka etc ] . We could create IP specific content, but the nuances wouldn't be sufficient. However, this does raise some good points so I'll put my thinking hat on to try and think outside the square.
|Do they contain any different information for their different locations? Are the inbound links from independent sources and appropriately geo-local? |
That's worth thinking about. Essentially, the pages are less about information and more about marketing. My concern about embellishing the content is that it will detract from the user experience - but again, I take your point and need to think outside the square.
One way of doing this is probably to benchmark local competitors in each market and see how they differentiate themselves. For example Public Holiday related offers only associated with that market [ this one doesn't actually fit us ].
Links - no the link quality is not good at all. A handful of links to each sites home pages exist, indeed I haven't analysed them, but some may be from the same sources - so that's worth looking into. Since needing to support multiple sites with links is likely inefficient, the emphasis will be on the .com site. Great point though to look into.
|Any other reasons for their being on separate ccTLDs? |
Again, a great point. The intention was to market to these specific regions. So the above points require some hard thinking as in Google's eye's they would ask "what's the point".
|Are they heavily interlinked? |
Not at all.
are you using any link rel elements?
Yes ( here's some excerpts with interim code left out between the elements ):
<link rel="profile" href="http://gmpg.org/xfn/11">
<link rel="pingback" href="http://www.website.com/xmlrpc.php">
<link rel="alternate" type="application/rss+xml"
<link rel="alternate" type="application/rss+xml" title="Website – blah blah blah »
<link rel='stylesheet' id='twentythirteen-style-css' href='http://www.website.com/wp-content/themes/website/style.css?ver=2013-07-20' type='text/css' media='all' />
<!--[if lt IE 9]>
<link rel='stylesheet' id='twentythirteen-ie-css' href='http://www.website.com/wp-content/themes/website/css/ie.css?ver=2013-07-18' type='text/css' media='all' />
I don't understand html [ beyond basic recognition ], so can you explain how your question may have relevance to the dupe content / caching issues, and what to potentially do about it?
in your case i would be interested in the possible usage of other link elements such as link rel canonical or link rel alternate hreflang (e.g. <link rel="alternate" href="http://example.com.au/" hreflang="en-au" />):
@phranque- good points, I'll get them ticked off.
Some more symptoms of Google's confusion is :
- entering the domain name.tld into Google search returns the .com version in the results
- inlinks under the domain name .com only show. Previously only .com.au has inlinks, now they're gone , so there's a bit of churn going on
- impressions and clicks are continuing at a fraction of the pre migration levels
Has anyone seen this type of behaviour surrounding duplicate content issues ?