The weeks leading up to this post were marked by observed changes in ranks impacting a large number of sites. Traffic losses on the order of 15%-30% were observed in some cases in addition to a major reordering on target terms, especially long tail terms in some markets. The widespread nature of these observations clearly indicate something major just happened. This usually suggests an algorithm change, or a change in the protocols governing how sites are getting indexed. It must also be said that we observed a similar traffic loss in the stats when Google was down for a few hours on 26 April 2009 (also on 6-8 Jan 2009), but we don't think that was in play here, since there were no other reports of an outage.
For the past couple of years, Google has been encouraging webmasters to use the nofollow attribute on links pointing to pages that do not contribute to the search. At the enterprise level, these suggestions were taken very seriously, because it seemed as if Google was providing a way for us to 'sculpt' the PR flow across our sites, enabling us to limit the wasteful passing of PR to pages that do not contribute to the search. Here's an early quote from Matt Cutts that clearly indicates how nofollow was being promoted:
The idea that you could "save" your PR for more valuable pages is clearly implied in this statement, and this was taken as fact by the seo community: You should preserve or 'sculpt' your on page PR to advantage pages that contribute to rank, and nofollow all that do not serve this purpose.
But yesterday, at SMX, Cutts made a staggering claim - that this technique no longer passes MORE PR to pages not blocked by the nofollow attribute. We're reading this to mean that if you are blocking bots from accessing pages, the PR saved from that effort is NO LONGER passed to the other pages. Basically, if you block pages, you're just throwing that PR away. This is a complete reversal of the previous policy, and one that will have a much larger impact on sites that use higher level optimization tactics - in other words this could harm sites who have been following Google's suggestions.
To address the flood of questions that will likely flow from this announcement, and because Google has never posted information providing real guidance on nofollow, they just posted some basic information in their webmasters area to explain what nofollow means:
http://www.google.com/support/webmasters/bin/answer.py?hl=en&answer=96569
http://www.google.com/support/webmasters/bin/answer.py?hl=en&answer=33582
Noticeably lacking from these documents is guidance on the specific issue of how PR is distributed when the nofollow attribute is used. So this is not contributing useful knowledge to all those businesses who implemented Matt's previous suggestions. And although he made the announcement at SMX, the WebProNews video just released at SMX includes a discussion by Cutts on nofollow but he makes no mention of the change. So as usual, we're left somewhat in limbo - no documentation, only verbal claims by Google's spokesperson that contradict a previous verbal claim.
Who Gets Hurt
We doubt that penalties will ensue from the overuse of nofollow, but if it is true that the PR is no longer "saved" for followed links, then those sites that are making heavy use of this technique may be harmed, as the PR connected to these nofollow'ed links is now discarded. And if your site was using this technique in the extreme - if all your links were blocked except one or two, you may be seeing a huge rank loss on your big terms, and because all your other links were intentionally blocked from indexing, you could be in a position where you have no productive ranks.
But most sites that have structural strength are not really dependent on PR sculpting and will probably not experience any issues with their ranks. Sites that are using semantically organized power centers, or are structured so that PR is groomed across related content, should not see any consequence of this new change unless the numbers of nofollow'ed links are very large with respect to the total number of links.
The real annoyance is that, once again, Google is changing the rules during the game, potentially harming sites whose intention was only to follow best practices. More information clearly needs to be forthcoming, especially from Google. This post is intended as a heads up for our clients and other power users.
And if you're using nofollow, a higher level strategy readjustment is in order. We had been advising clients to consider blocking all internal links, except for those that specifically contribute to rank, in an effort to conserve valuable PR. If this change is real, and we think it is, that strategy no longer applies. We're not recommending that everyone remove all their nofollows, but unless you're selling links, there is no longer any incentive to apply a nofollow strategy broadly across sites.
The real question for this moment is, "Why did this change occur?" We suspect 2 reasons:
1) Because of the ability to sculpt internal PR the nofollow attribute was being used (gamed) to advantage sites that were blocking all but their biggest targets. Many businesses could thrive by ranking one term in the top 5, even if it meant holding no other rank.
2) Widespread use of nofollow on all outbound links was becoming a default standard, and the resulting consequence on the natural search was that no one wanted to link out to anyone else for fear of either losing their PR, or appearing to be selling links. This is clearly not good for the links-as-votes model that Google uses. So by changing the consequences of using the nofollow attribute, Google may expect to see a greater willingness to share links, perhaps reversing the existing trend. We'll see how that plays out.
The "Xeeneexee" Link Authority Experiments
We have been enthusiastic adapters of the nofollow techniques starting in late 2007, as a result of several experiments that were designed to test the claims put forth by Google. The one below involves 3 PR3 sites with varying numbers of active links pointing to the target page. Based on Cutts' original claim we were able to see that the nofollow attribute did indeed preserve PR. The experiment is still live and we should expect to see some reordering in the ranks as a result of this recent change. We're not seeing any proof yet of this latest claim, but this experiment should either confirm or refute what was just made public once enough time passes:
http://www.bobseo.com/aaa/link-authority-071231.php?password=seo
More reading:
http://www.webpronews.com/topnews/2009/06/03/google-evaporating-excess-pagerank
http://www.seroundtable.com/archives/020141.html
http://www.seomoz.org/blog/questions-answers-with-googles-spam-guru
re1y.com
Enterprise SEO
Google Penalty Solutions
Automation & Search Compliance