February 17, 2005
Google Sandbox: Age of site and Allegra
The Google Sandbox continues to be a phenomenon that has no hard information either on how it is used and what exact parameters have been applied.
While some content developers continue to assign to "myth" this link building issue, for many webmasters the unfortunate effects are clearly in play.
The short and simple agreement is that newer sites effectively suffer some form of "probation", where Google's algorithm is stacked against pages from newer sites ranking for search terms - especially for anchor text pointing to a site - such as from directory listings and link advertising.
It can even affect redirects - if you develop old Domain A, then after a time redirect to new Domain B, then all previous rankings held by Domain A will probably not be passed on to Domain B during a sandboxing period.
A new site is still able to rank on its content and internal link anchor.
However, as the internet is built upon links, and so long as new sites are sandboxed in this manner, they are effectively isolated from the benefits of anchor text - the key architecture of the internet itself for document interrelationships.
While Google has tried to separate organic search from paid listings, the insidious reality is that if you want your new website found in Google, then you'll probably have to use AdWords to develop your initial traffic streams from Google results.
Age of Site
It is uncertain whether the age of links themselves are involved in the sandboxing process - John Scott originally reported that Google was "grandfathering" links, and this would certainly make sense to some degree - after all, sometimes sites can rank very well on very few backlinks from established sites, as opposed to other sites with thousands of backlinks spread all over the internet.
The trouble with the "grandfathering of links" idea is that it is very difficult to test - it would be extremely difficult to ascertain whether other factors were at play amongst a string of other sometimes constantly changing variables that form Google's ranking algorithm.
However, one apparently clear factor is age of site.
Example - There are two sites I took on relatively recently, that for all intents and purposes are the same in SEO terms.
They have approximately the same number of links, and these links are spread over a similar IP range, including the same set of sites in certain instances. They also deal in roughly similar topic areas, and even compete for some of the same keywords.
Here is how they are ranking for targeted search terms (single keywords and two-keyword strings) on Google.com at present from the DigitalPoint keyword ranking tool:
Site A has all but 2 targeted searches outside of Google's top 10 rankings - whereas Site B has barely anything in the top 1000 - even for the same search terms.
All prior factors taken into consideration, the single biggest different is the date that the domain was registered. Site A is nearly 6 years old. Site B is only 18 months old. Coincidence?
Although it's possible that other factors are involved - perhaps Site A has accrued a few exceptional "authoritative" links that Site B lacks - the suggestion here is that age of domain is a factor in sandboxing.
On February 5th Google began an update which was named Allegra. This update has heralded the biggest change to Google rankings since November 2003.
There has been a lot of suggestion that the Allegra update has freed sites from sandboxing. However, the Allegra Updeate is still not completed - Google's string of datacenters are still reporting different results between them. Until they all line up, the update is not complete. Sites gaining a boost in traffic during this period may well lose that traffic again once the update has settled.
So what now?
The Google Sandbox remains an enigmatic phenomenon, but appears to remain a feature for restricting the influence of new sites in Google's search results.
It's as yet impossible to tell when the cut-off point between "sandboxed" and "un-sandboxed" exists - presuming there even is a distinctive boundary as once reported.
It also seems that the length of term for sandboxing is increasing - when originally reported in March 2004, sandboxing was effectively a 3 month delay. Now it is being reported on SEO forums that it may take as long as 6-8 months.
Another interesting possibility is that sandboxing may be related to the number of links accrued, with new sites that gain links too quickly effectively sandboxing themselves longer.
The idea has been particularly mooted in discussions such as Google and the Golden Ratio.
Even if incorrect, the idea that new domains can less "sandbox" themselves as much as "quicksand" themselves with high-degrees of link acquisition is a very appealing one.
I have a couple of big projects I intend to start work on, but I plan to leave then with only rudimentary linkage for their first six months of post-development, before I even considering opening up link advertising for them. Even then, the extent will probably be modest at first. I'll try to set up a control and see how both fare in terms of traffic generation.
Posted by at February 17, 2005 06:39 PM
> Discuss this in the Platinax Internet business forums
For comprehensive internet
All content © Copyright 2004 Brian Turner. All rights reserved.