Debunked: Only The 1st Anchor Text Counts With Google
A little while ago I read the following thread over at SEOmoz:
Results of Google Experimentation – Only the First Anchor Text Counts
This made me panic a bit, because it would have meant that the footer link I had on this blog was not helping my homepage rank. I think someone mentioned in the discussion that even if the 1st link was nofollowed, it would still be the only one that counted. This would have meant that I was supposed to be completely shafting myself.
I got rid of the 1st links to the homepage altogether hoping to see a rise in Google for my main phrase, but I did not. I did however rise over 60 places, within less then a week or so on MSN/Live search.
I wanted to know for sure with Google, so when I 1st changed over my design less then a week ago, I made it so that every post in this blog linked to the about us page, with my name as the anchor text. I have just checked today and it is ranking 4th on Google UK, just behind my Yahoo! post.
- The 1st anchor text always says about
- There is very little on-page for it, other then the links in these posts
- There are other pages with far more David Eaves links (external ones)
- There are no external links to that page using that anchor text
My conclusion: Google does not only look at the 1st link to a page for anchor text, MSN/Live search does and I am not too sure about Yahoo!
Good advice and practice from SEOmoz anyway though, if only for MSN. Why take a chance on it? Make sure that you have your good anchor text in the 1st links to your pages.
Update: Seeing how VanDeMar seems to be the one who started this whole thing off, here is a link to his original post:
You May Be Screwing Yourself With Hyperlinked Headers
Posted in SEO News By David Eaves, a UK search engine optimisation specialist.
If you enjoyed this post why not subscribe to the RSS feed.
June 2nd, 2008 34 Comments
Both comments and pings are currently closed.
June 3rd, 2008 at 10:17 am
Interesting findings.
Mikes tests on the Smackdown blog did look pretty conclusive I have to say http://smackdown.blogsblogsblogs.com/2007/10/09/you-may-be-screwing-yourself-with-hyperlinked-headers/.
I actually ran a couple of tests after this and found the same thing aswell – although this was sometime ago now. I know Barry did a brief test too http://www.seroundtable.com/archives/016582.html.
I think a controlled test is required.
June 3rd, 2008 at 2:06 pm
Hello.
While the test I ran was not exhaustive, I really don’t think I would label yours as a “debunking” of the phenomena… I mean, for starters, where was it ranking before you started?
PPCblogger is right, you really would need to run a more controlled test.
June 3rd, 2008 at 2:27 pm
The page never ranked anywhere at all for it before the re-design, I re-wrote it, mentioned my name on the page (like I do with every blog post) and then linked to it from every post on this blog. The homepage of this site, the blog homepage and many of the blog posts have lots of external links using that anchor text and the about us page outranks them.
The links at the bottom of these posts are definitely counting, what else could be causing the page to outrank most of the others for David Eaves?
June 3rd, 2008 at 4:43 pm
Any number of things, actually. You currently only have 56 of the 115 indexed page for this site in Google’s main index, the rest are in supplemental. It could be something as simple as the page that used to rank in it’s place went supplemental. Did you have an indented listing before? Do you know what page that was?
The main issue is that you are testing with a phrase that currently has too much noise. You would have to dissect all of the possible pages that could be ranking for your name on this site, and compare all of the of the factors that go into it, in order to determine what was actually going on. In a real world environment, using non-unique phrases like this on a live site where rankings can fluctuate with no changes whatsoever, it’s next to impossible to conclusively say, “I did A, and B happened, therefore A caused B”.
BTW – it doesn’t mean that you’re not right, or that things didn’t change, or that google.co.uk isn’t treated differently than google.com. You don’t have the same results there, and we do know that diff countries use variations of the algo (the /blog/ homepage is second after the Yahoo post on google.com). I’m just saying that this activity doesn’t actually prove or debunk the others posts.
June 3rd, 2008 at 4:59 pm
I’m going to have to agree with VanDeMar and PPCblogger.
First off, it appears that several other variables could have caused the results. Secondly, this will have to be replicated several times in order to have any statistical significance.
But I am glad that you’re looking into this topic. It’s a fairly important topic because it heavily influences interlinking strategy.
P.S. Your “Leave a Reply” format seems to be screwed up.
June 3rd, 2008 at 5:05 pm
@Hugo – I am getting the comment thing fixed soon, thanks.
June 3rd, 2008 at 8:05 pm
[...] David Eaves links (external ones) There are no external links to that page using that anchor text Debunked: Only The 1st Anchor Text Counts With Google Only the First Anchor Text [...]
June 3rd, 2008 at 8:40 pm
I have seen significant increases by adding keywords to home footer links that are not the first link back to the homepage in the source (that occurs in the top nav). Based on those observations, I find it difficult to believe that Google only looks at the first link.
June 3rd, 2008 at 9:58 pm
Of course a page doesn’t need anchor text in links to rank.
Use a random word anchor link to the page from every page other than the about page, but after the link to the about using Dave Eaves. Don’t include it on the page itself.
The problem is Google might not count a link because it is not on the page, or because it is irrelevant.
It only becomes relevant if it is on the page with a made up word.
Then you might have to take into account temporal factors – maybe the link only counts after a certain period of time, or if an external link also points to the page with the same anchor text.
They could also factor in some kind of split testing of links, or have a “Bungle and Zippy” (from Rainbow) factor – choosing which window they are going to look through on any particular day.
It is actually something quite hard to prove conclusively.
June 4th, 2008 at 5:38 am
I realized a while ago that this was not true. The only thing I have not been able to figure out is if the first link passes more ranking juice than a 2nd or 3rd link.
ie
If the first link read david eaves would you then start to rank #1?
June 4th, 2008 at 7:16 am
According to the study I have done so far; I am sure Google crawls the entire page and not just the first link. As a spider runs through every page Google comes across
June 4th, 2008 at 8:57 am
While I’m pretty sure that all of the links will count in the algorithm I’m also sure that the number of links counts for something. Having the header link back as well as several other links to the homepage will increase that pages link strength. But I do believe that you now have to divide the link strength and the strength of the text between that many links.
basically I have 3 links going to my homepage, 2 text links and one image link. Now the text is only 2/3 strength. That could be a possible linking factor.
While this hardly debunks the idea of “only 1st anchor text counts with Google” it also doesn’t prove it.
June 4th, 2008 at 11:59 am
[...] Debunked: Only The 1st Anchor Text Counts With Google – If you have two links to the same page, which one counts? Thumbs up!! [...]
June 4th, 2008 at 8:27 pm
All links count, but with more links added each new one counts less. I believe that links inside the first 250 words count more.
Check this out, please…
June 8th, 2008 at 8:25 pm
[...] deed een ander testje waaruit hij concludeert dat Google wel meerdere links volgt. In zijn” Debunked: Only the 1ste anchor text counts with Google” probeert hij aan te tonen dat dit niet het geval is…niet echt een waterdichte test [...]
June 20th, 2008 at 1:45 pm
[...] recent SEO experiments (I tend to believe though they were debated) show that you should always take care that your “best”, most targeted anchor text [...]
July 3rd, 2008 at 8:19 pm
[...] was risen at both posts. That particular facial hair elevation trend has seemingly culminated by David Eaves’ post which supposedly debunks the theory. However, some problems were found in the way David has [...]
July 4th, 2008 at 11:10 am
[...] the full story: SEOmoz | Results of Google Experimentation – Only the First Anchor Text Counts Debunked: Only The 1st Anchor Text Counts With Google Google counting only the first link to a domain – rebunked | SEO Scientist – Applying the [...]
July 4th, 2008 at 12:17 pm
[...] Debunked: Only The 1st Anchor Text Counts With Google [...]
July 7th, 2008 at 5:16 pm
[...] Google учитывает только первый линк на определенный url. David Eaves также придерживаетÑÑ Ñтого [...]
July 12th, 2008 at 3:44 am
[...] month David Eaves from SEOco.co.uk debunked the idea that if you have two links to the same place on a single page, only the first one will pass its [...]
July 30th, 2008 at 10:12 am
[...] Debunked: Only The 1st Anchor Text Counts With Google [...]
August 6th, 2008 at 9:20 pm
[...] http://www.seoco.co.uk/blog/2008/06/02/debunked-only-the-1st-anchor-text-counts-with-google/ [...]
November 27th, 2008 at 11:50 am
[...] only the first link to a domain – rebunked | SEO Scientist – Applying the scientific method to SEO Debunked: Only The 1st Anchor Text Counts With Google …then we should be making the first page/menu link to appear in the page code a keyword. The [...]
March 19th, 2009 at 4:49 am
[...] contributes some original research himself. Other relevant posts come from Rand, Mike van de Mar, Dave Eaves and Michael Martinez. Some partial information was also provided by Matt Cutts to Debra [...]
April 1st, 2009 at 2:43 pm
[...] contributes some original research himself. Other relevant posts come from Rand, Mike van de Mar, Dave Eaves and Michael Martinez. Some partial information was also provided by Matt Cutts to Debra [...]
May 21st, 2009 at 5:24 pm
[...] mid-2008 there was continued discussion at SEOmoz, it was debunked by SEO CO, and “re-bunked” by SEO [...]
June 3rd, 2009 at 4:55 am
[...] mid-2008 there was continued discussion at SEOmoz, it was debunked by SEO CO, and “re-bunked” by SEO [...]
June 5th, 2009 at 12:43 am
[...] mid-2008 there was continued discussion at SEOmoz, it was debunked by SEO CO, and “re-bunked” by SEO [...]
June 15th, 2009 at 7:08 am
[...] mid-2008 there was continued discussion at SEOmoz, it was debunked by SEO CO, and “re-bunked” by SEO [...]
June 20th, 2009 at 2:36 am
[...] contributes some original research himself. Other relevant posts come from Rand, Mike van de Mar, Dave Eaves and Michael Martinez. Some partial information was also provided by Matt Cutts to Debra [...]
August 4th, 2009 at 3:02 pm
[...] David Eaves hizo otro experimento en su propio Blog para verificar lo que Rand Fish decÃa sobre que el único link que cuenta es el [...]
January 28th, 2010 at 11:06 pm
Yeah,i don’t see this happening. Just wouldn’t make sense
January 28th, 2011 at 1:05 pm
You maybe onto something. The data looks good. Thanks