I came up with an interesting finding today, whilst testing whether having a link on a Google Profile passes PageRank. Writing it up prematurely, I am going to end up tainting the experiment, but since I inadvertently spotted something interesting, I decided to stop anyway and tell the world. So this post:
- Shows you how to tell whether a page really passes page rank
- Shows you why I think Google Profiles don’t pass PageRank (even though they are not nofollow) and
- Shows an error in Google’s algorithm…. clear as day.
How to tell if a page really passes PageRank
Well first… I don’t believe in Page Rank. What I am actually testing is whether a link from page A to page B actually influences the ranking of page B in the SERPS. If you think that’s the same as green fairy dust, then please go back a few steps.
My test was to put some unique anchor text on page A that did not appear at all on page B. Most people trying to check or claim a page passes any authority or relevance tend to completely ignore this important part of such a claim. If the anchor text has any similarity with the content on the landing page, then you have screwed up your test, because there could easily be (and probably are) other factors affecting whether the landing page shows on a search engine for that term. Below is my test in pictures.
Why I think Google Profiles don’t pass PageRank
I did the above test on a Google profile… My own google Profile. I used the unique text: “I live in Harlington with Marie and my two kids and I have just taken on an allotment.” The profile went live in late November. Now in THEORY, that link is not “NoFollowed” so if Google finds it, then it should take some authority from it’s parent domain (which is www.google.com!) and then in due course, the landing page would show in the Google results for either the phrase “I live in Harlington with Marie and my two kids and I have just taken on an allotment.” or just “just take out an allotment”… even though these words do not appear on page B right? Of course, just because Google doesn’t put “NoFollow” tags in its own links, doesn’t mean the links are “DoFollow”. Anyway… Here’s what has happened with the test so far.
Day 1: Google hadn’t indexed anything. So no screenshot to show you.
6th January the result was like this:
Very cool… Compete.com seems to have cached the page, and its cache made it into Google’s SEPS before the original post! However… no sign of the landing Google profile or the landing page yet.
Jan 15th the result was like this:
Google has found Page A (The Google Profile) and has indexed it… and to be fair has given it more relevance than the Compete.com Cache, even though Google indexed the Compete page before its own.
Feb 1st the result was like this:
Google managed to kick Compete out of the results altogether. But the Landing Page (page B) is still not appearing for either of my test search phrases. Now that I have published this post, my test must stop, because I have other links to the landing page. But a new test continues, I suppose, to see if THIS page passes anchor text link juice to the landing page. But I stopped the test, because the test already found two new things:
An error in Google’s algorithm…
Look closer at the above image. Is it my eyes, or does it say “No results found for ‘I live in Harlington with Marie and my two kids and I have just taken on an allotment.’ and then… right underneath it… show a page with EXACTLY that result. Syntax correct… match case correct… everything correct. So the Hazard symbol lies.
Here’s another bonus secret for reading to the end…
Most people think that Google will call the first page it indexes with duplicate content the authoritative version. However, Google clearly indexed compete’s version first… some days later, it indexed my profile… then some days later again it dumped compete.com.
So that debunks the theory that the first call on content wins.
Much to think about there. Can you add to my findings?
15 Comments
Christoph C. Cemper the Marketing Fan · 3rd February 2009 at 1:57 pm
Hey Dixon!
thanks for the great post!
I like the way you try to conduct the test, and agree with many of your points.
However I believe that the LENGTH of your anchor text (18 words) can have a significant impact on the actual test result.
I’ve been doing tests with long queries in Google for more than a year, and in fact we do that on a daily basis. This and several other related posts from fellow SEOs make me doubt that this long phrase is the perfect test setup.
Hit me up on IM, the next conference or twitter to discuss details…
Cheers,Christoph
Dixon Jones · 3rd February 2009 at 2:10 pm
Ah – a good point. In fact, the anchor text itself is only: “just taken on an allotment”. The screenshots use the whole phrase on Page A, but the results were the same for just the anchor text itself. (I didn’t see any mileage in taking too many screenshots, but I was also using the test to verify whether words around the anchor text influence the rankings… but since I never got page B ranking under the 5 word phrase it became a little acadmic.)
I’m at SES London later this month if you are there: http://dixonjones.com/speaking/
Dixon.
JonnyScott · 3rd February 2009 at 4:48 pm
Hi Dixon
I have noticed the error message on a definite match before. I thought, foolishly, this may have something to do with the number of stop words or denisty or positioning or in other cases that this has something to do with special characters. Perhaps it has something to do with string length? Why?
Well all I did was a simple checked what pages come up when I continue to extend the length of your searches:
“I live in Harlington with Marie” – 2 results
“I live in Harlington with Marie and my” – 2 results
“I live in Harlington with Marie and my two kids and I have just” – 2 results
“I live in Harlington with Marie and my two kids and I have just taken” – 2 results
“I live in Harlington with Marie and my two kids and I have just taken on” – 1 results
“I live in Harlington with Marie and my two kids and I have just taken on an” – 1 results
Why the drop off point is there? Someone clever is about to answer…..(please!)
ps. Only cause of your facebook status did i end up spending 10 minutes on this. Not sure if that is a good or a bad thing…
See you at SES
LifeDesignSEO · 4th February 2009 at 8:47 am
Hey Dixon,
Nice test – don’t think there’s an error though. If you look closely, you’ll see Google is offering you the non-exact match results because the exact match returned no results.
My 2 cents on the indexed page’s slide would simply be that Google dropped it from the main index once the ‘freshness’ factor wore off.
Have fun at SES!
Dixon Jones · 4th February 2009 at 2:16 pm
Hi LifeDesignSEO,
I did notice that, but the truth is that the match IS exact, as the search was cut and pasted directly from the profile page. It is possible, though,that Google decided that it wasn’t exact because only part of the text string was in a URL.
I went to test this… and… INTERESTING! Now that this post has been indexed, the Google Profile is absolutely nowehere to be seen – even in related searchs for “I live in Harlington with Marie and my two kids and I have just taken on an allotment.” – This page just shows as a GoogleWhack (www.googlewhack.com). I did say that when I posted this, I would corrupt the original experiment – and so it has proved – but yesterday, Google thought my Google profile had SOME relevence to the search phrase… but now it has dropped the Google Profile altogether. So yes… looks like your 2 cents was bang on the money.
Dixon.
Hobo · 8th February 2009 at 5:39 pm
The test could be flawed. You didn’t take into account a character limit on the actual hyperlink.
Your phrase is over 80 characters. A recent test I did seemed to confirm a 55 character limit (on that test alone).
http://www.hobo-web.co.uk/seo-blog/index.php/anchor-text-length/
Google Profiles do pass anchor text relevance. I did the same test. I will PM you if you want to see 🙂
Do They Pass Pagerank? Don’t know depends on other internal and external links to it I think.
Dixon Jones · 8th February 2009 at 7:45 pm
OOH! Good point Hobo! Yes – if your test rings true, then there is indeed a flaw in my test. It also correlates with Chris Cemper’s comment.
It also lends itself to Jonny’s observations.
So I don’t know about 88 characters (or the cut off length) but I’ll have to do the test again now, feeding that thought in.
I have now updated the profile, with new anchor text to a new domain. You’ll be able to see what happens – but if you’ve done the test you’ll know. (I still want to see for myself).
Dixon Jones · 16th February 2009 at 4:42 pm
OK – new findings from the new test. So new post!
http://dixonjones.com/seo/google-profiles-pass-anchor-text-relevency/
Saving Money Tips · 19th November 2010 at 2:03 pm
If I pass pagerank to my site from totally meaningless anchor text, does this still improve the ranking of the keywords from my on-page factors?
admin · 16th December 2010 at 6:02 pm
kinda – in that a link from an image is better than no link at all. However – whilst it increases the “juice” to your site overall, it spreads the focus of what your site is about. So imho no anchor text is possibly better than confusing anchor text.
The Big Bang Theory · 2nd February 2011 at 5:26 am
Just to point out it says their are no results for: “I live in Harlington with Marie and my two kids and I have just taken on an allotment” not I live in Harlington with Marie and my two kids and I have just taken on an allotment…. see the difference? the one you typed in had quotations around it. Google couldn’t find that statement in quotations because it doesn’t exist but could find it without quotation because that one does exist. Hence why after the search result it found it says “(without quotation). So no their is nothing wrong with Google’s algorithm. Google’s algorithm is a complex thing (as you know) more complex then most everyones mind can handle. Thats why when you use Google you don’t ever get viruses compared to Internet Explorer and Mozilla Firefox witch do.
admin · 2nd February 2011 at 8:09 am
I get results for both versions. The quotes are an operator you can use on Google and most search engines to ask for the exact phrase, as opposed to the words in any order.
The Big Bang Theory · 2nd February 2011 at 5:28 am
Correction it says i posed it at 5:26 am. thats wrong its 12:26 am I’m not sure why it did that.
Passing PageRank through Google Profiles » Dixon Jones: · 16th February 2009 at 4:34 pm
[…] have been testing whether a page passes PageRank over the last few months and in particular was using Google Profile for my test. I managed to […]
Link Juice - Google Profiles dá força em seu domínio · 17th February 2009 at 12:30 pm
[…] dixonjones.com Notícias […]