Once again the non-discerning SEO community has come out of the woodwork to praise a nonsense report from SEOmoz that offers absolutely no credible information.
Danny Dover (who I am sure is probably a really great guy with lots of friends) published a very poorly written report on yet another test that supposedly “proves” PageRank sculpting “still works” (it never worked in the past but that doesn’t stop people from making unsupportable claims).
Instead of telling people which Websites were used for this test Mr. Dover only offers a vague description of the methodology employed (nothing about site structures, inbound linkage, etc.) and he caps off his test with a list of numbers that supposedly proves that PageRank sculpting works.
This kind of sham analysis doesn’t prove anything. It’s faux science being perpetrated on the SEO community and the search world in general by an amateurish approach to “openness”.
There are ways to tell people about Websites without linking to them. If the SEO community really wants to test PageRank sculpting it can share the data openly, especially for Websites that serve no other purpose than to test PageRank sculpting.
Google never denied that you can redirect the flow of your PageRank by using “rel=’nofollow’”.
What they did tell people in June 2009 was that so many Websites were screwing up their search visibibility with “rel=’nofollow’” on internal links that they felt compelled to take the extraordinary measure of changing how they handled PageRank on pages with “nofollowed” links.
A move that the Great SEO Testers of the Web (including SEOmoz) failed to detect with their tests.
This emotional attachment to proving that a really stupid idea can somehow work underscores just how weak the SEO approach from PageRank sculptors really is.
There ARE people who are challenging the report but let’s be honest here: most of the people who comment on blog stories don’t really stop to think things through. They are applauding a great read, not a great in-depth analysis. There is no such thing as a great in-depth analysis of this topic — at least not from the pro-sculpting camp.
If you want to prove that PageRank sculpting works, you MUST publish real data.
You MUST show us the Websites.
You MUST let us see exactly how the link structures work.
And SEOmoz of all companies is in no position to claim that people might screw up the test with unauthorized links. If LinkScape is really THAT good, it should be able to detect inbound links.
There would be no need for anyone to try to mess up a REAL PageRank sculpting experiment. In fact, a REAL experiment SHOULD take natural linkage into consideration (rather than artificially constructed link profiles).
What needs to be measured (besides the PageRank that cannot be measured) is the change in search visibility for all the pages of a sculpted site.
You also need to measure the change in search referral traffic for all the pages of the sculpted site.
You also need to measure the change in all the sculpted site’s pages’ ability to pass link value to other documents.
These kinds of measurements would prove or disprove the value of PageRank Sculpting.
These kinds of tests can only be effective if they are done openly, where everyone can see the results, track the changes for themselves, and verify in a shared environment that natural search processes are producing a real change.
Until that time, anyone who claims they can prove that PageRank sculpting works (particularly with “rel=’nofollow’”) is in no position to make any credible statements on the topic.
{ 14 comments… read them below or add one }
Luke Jones 12.29.09 at 11:45 am
I replied to your comment on SEOMoz but though it necessary to post on here too. It’s unfair of you to respond in a personal way but I do see your point, as do many other members of SEOMoz.
You’re shedding poor light on certain members of SEOMoz and I don’t think it’s fair to do so at all.
As someone who studied psychology, I know a lot of studies, methodology, results and so on and so forth. This test is clearly not as scientific as some people think and most are taking it with a pinch of salt.
trontastic 12.29.09 at 12:46 pm
I have to agree with you that the data provided by Danny did not support his theory. Any hope I had that his test might shed light on their position in favor of sculpting was shot down as soon as I realized they
A) weren’t going to tell us what sites they tested
B) Only 2 months? This should have been a much longer test ESPECIALLY since so much has been changing recently.
C) I had a Whopper for lunch and now I wish I could nofollow these narsty burps.
I dunno, maybe I’m just bitter because I’m working this week but it seemed like there were a ton of holes even in the updated version.
Michael Martinez 12.29.09 at 12:57 pm
Luke, I took such a hard line on Danny because we should as a community have moved past this whole “PageRank sculpting with nofollow works” issue. It has been thoroughly and completely discredited as snake oil SEO.
And yet Danny felt compelled to do the “test” and publish his “results” anyway.
He needs to be held accountable for raising the spectre of bad SEO. I don’t want people to question his ancestry or attack him on the street, but the people at SEOmoz need to stop publishing this nonsense.
Quentin6s 12.29.09 at 4:17 pm
Way to go Michael.
Glad to see someone stepping up to the trash that comes out of the SEOmoz blog. They will literally publish anything with a graph or percentage numbers on it.
Why people continue to waste their time there blows my mind.
Billy 12.29.09 at 4:56 pm
Anyone and everyone who’s main selling point still revolves around PageRank still have a long way to go in the SEO game. It is a bit disheartening to see these pros focused on PageRank, when the real problem today is personalized search pushing down every natural search result except the top 3.
underdown 12.30.09 at 10:04 am
Agree with your post – but a quick question – why do you nofollow the “seo-theory” home link up top?
Michael Martinez 12.30.09 at 10:26 am
underdown “…why do you nofollow the “seo-theory” home link up top?”
Michael We don’t. The theme apparently does.
It doesn’t seem to have helped our rankings any, though, has it?
Michael Martinez 12.30.09 at 10:29 am
Quentin6s “Why people continue to waste their time there blows my mind.”
Michael I have been an oft-time critic of some of the principles the SEOmoz team follow, but there have been many good articles come out of their group, too, and I’ll be the first to say that.
I check SEOmoz every day. I may not learn much from them but I don’t expect to learn much from any one source. Like many other old-timers in the industry, I’m down to gleaning bits and pieces here and there.
But I reserve the right to say when I think anyone’s article is smoke and mirrors, and Danny’s article is smoke and mirrors.
Theorist 12.30.09 at 5:39 pm
The fun part? Averages are worthless for small sample sizes. The “winner”, nofollow, has a standard deviation of 1.598, or a 95% confidence interval of -.75 to 5.5 . So… you can be 95% certain that it’s true position is pretty much anywhere. Those are great findings.
They should have done at least 50 tests with at least 20 pages (have the same techniques compete with each other). If nofollow really is a winner, nofollow should dominate significantly in all 50 tests, squeezing out the other techniques.
Ridge 12.30.09 at 5:54 pm
If you have nofollow sculpting already installed, don’t remove it. If you don’t have it installed, implementing it probably won’t make a drastic change but we encourage you to test this when it is responsible to do so.
^^^^ Above is what I find very disturbing. After 2 months of limited data they are seriously advising people to do this again. I was pretty stunned that he seems to very sure that it works after a 2 month test, sure enough to encourage people to continue this practice. Pretty amazing and irresponsible IMO
audit de referencement 12.31.09 at 9:52 am
Great Post Michael !
What solution do you recommand to block a page from getting juice ?
I’ve tried the js function in a .js on a robots blocked directory on this site (weekendesk.fr // “A moins de 2h de chez vous” part) but seems like it don’t work. Actually, I can see this pages (ex : weekendesk.fr/2h_de_lyon) in the internal links from Webmaster tools…
Michael Martinez 12.31.09 at 12:03 pm
You cannot block a page from receiving PageRank. And why would you want to?
AleFocardi 01.04.10 at 4:04 pm
I completely agree with you Michael, and you can also let the test run for some 3 – 5 months and gather data over that period of time, quantifiable data that you will be able to show.
When you publish your article you already came to a conclusion and can happily show the test websites since all the evidence would have been already gathered.
What I like is the fact that we need to take this at face value.
Badams 01.06.10 at 4:10 am
I applaud SEOmoz’s efforts, however flawed, to inject a more scientific approach in to SEO practices. I can see valid reasons for not publicising the websites used, especially as making the sites public could (and probably would) pollute any future tests they’d like to do on them. And considering the time & effort involved in setting up these tests in the first place, I imagine they want to minimise the risk of invalidating future experiments.
Having said that, I agree that the whole PageRank debate is kind of a moot point. I used to be a form believer in PageRank sculpting, but I’ve gone off it thoroughly as my knowledge and experience in SEO grew.
You must log in to post a comment.