Total On-Page vs Total Off-Page Test

everyone: Not sure if I got bored or what it was but I decided to set up a couple of pages and see whether total on-page optmization would out Rank total off-page optimization.

Or perhaps I should say, to see how many links it takes to have the off-page optimization out Rank the on-page optimiziation.

I choose as the keyword "Bob Mutch." I did up a page called bob-mutch.html and did all the on-page optimization I could think of. You can view this page by going to http://www.seocompany.ca/tool/seo-tools.html and select the Optimization Project link from the Sponsored Links. Let me know if there is anything else for on-page that I should add. I have list 29 things that I have done in the body of the page.

Then I put 500 links toward http://www.seocompany.ca with the anchor text "Bob Mutch". This page is already Ranking 12th in Google, 73rd in Yahoo and 2nd in MSN for that key phrase form the Bob Mutch links from my blog.

Already I think it is interesting that the off-page optmized page rankings 12th on Google but 73rd on Yahoo. Showing perhaps that links count more with Google.

Here is the cached version of the Cached Page.

Well I will post from time to time and let you know how its going. Pass on any tips for on page optimization.

 

I don't think that is a fair test. You should have made a new page too for off-page too and throw links at that one to make it 'scientifically' correct.

 

TOP30: Good point. It just happens that I have been getting lots of PR5 links for my home page with the anchor text "Bob Mutch" for a SEO Blog I post on. Not to meantion that I have 2 PR8's pointed at it and lost of other links.

The way I look at it is the off-page factor page has a bit of a heard start. But I guess I am really just wanting to make a point. I know that the on-page page doesn't have a hope.

 

Nov 14th 2004:
Of the term "bob mutch" the following stats:

Results on search of "bob mutch": Google 42K; Yahoo 17K; Beta MSN: 16k MSN: 3K

On seocompany.ca (620 IBL with "bob mutch" in the anchor text NO on-page optimization)
Google #4
Yahoo #1
Beta MSN #1
MSN #2

On seocompany.ca/bob-mutch.html (On IBL at all ALL on-page optimization)
Google #5
Yahoo #>500
Beta MSN #2
MSN Not in the index

I found it strange that bob-mutch.html did so well in Google which is known to have more Ranking weight on links that Yahoo, while on Yahoo bob-mutch.html didn't even rank in the top 500 (yes it is in the Yahoo index I checked this by searching for the first link in the file).

It was nice to see that Beta MSN was also giving bob-mutch.html #2 which tells me it is counting on-page optimization as high.

 

interesting

from the UK using google.com I get:

www.seocompany.ca/ at #4

This thread as 5 and 6

go figure

 

So what value is this test?

1. You are from your own words expecting to prove that off-page SEO is more important than on-page SEO.
2. To prove this, you select a page that already exists with several high PR backlinks to serve as your off-page SEO candidate -- this will be compared with a new page with no backlinks constructed especially for the test as the on-page SEO candidate.
3. You acknowledge that the off-page SEO page has a considerable head start but defend this by saying you expect the on-page SEO page to lose anyway.

My question:

Why bother to (a) do this test at all or (b) tell anyone at all about it or its results? Your null hypothesis (the one you are trying to "disprove") is that the on-page SEO candidate will win. You acknowledge that the "experiment" is set up in such a way that it is highly unlikely that the outcome can be anything but a rejection of the null hypothesis.

This isn't an experiment. It's more like a political advertising campaign.

 

minstrel: The value of the test. Clearly the value of the test is to show how a page with ALL on-page ranking and NO off-page ranking will compare with a page that has NO on-page ranking and ALL off-page Ranking.

1. Yes I personally expect that the page with all the links will ranking the highest, but as you probably know not everything one expects happens.
2. That is correct, I selected a current page that already has a number of IBL. It also have 620 links with the anchor text "bob mutch". The interesting point is that in Google and Beta MSN the bob-mutch.html page Ranks right below the off-page optimized page.
3. Yes the off-page has a head start but this will not mean a lot given time. No I don't defend that my saying the off-page will win anyway. Given time the heard start will not mean anything.

a) Why bother to do the test? Clearly I want to see the results.
b) Why tell anyone about the test? So I can get feed back and pointers.

I am not trying to disprove anything. I am doing a test to see what will happen. I just happen to have an opinion of what will happen.

Rejection of the null hypothesis. Well with Google it is quite well known that a on-page optimized page can't out Rank an off-page optimized page. But it remains to be seen on Yahoo and Beta MSN.

After I see the end results of this experiment over a period of time I will adjust the inbound links into the page with the anchor text "bob mutch" and see if that will make a difference.

I am also considering creating a number of other pages and linking them so they will have the same PR vote and then giveing them differnet inbound links so I can seeing how many inbound links with the anchor text "bob mutch" needs to be given before it will under Rank or over Rank the all on-page optimized page.

"This isn't an experiment. It's more like a political advertising campaign." I can see where it would look that way to you with you asking some questions first. I trust the above information will give you a better idea of what I am claiming to do.

 

Bob, I think you have to decide:

1. the experiment isn't needed because the answer is already well known; or
2. the experiment is needed, in which case you want to do it in such a way that it will provide some answers to the question posed: the way it is now, your "experiment" won't do that

How do you know that? How much time? As TOPS30 said, do it right or don't bother doing it.

But you obviously are -- see my comment about the null hypothesis above, plus your own comments. I admit this is a bit of a pet peeve with me -- I have seen numerous "experiments" and "tests" that are claimed by their authors to "prove" this or that and don't do either. This is just one more that has design flaws which will prohibit any real conclusions. Such "tests" don't clarify anything. They just muddy the waters even more.

 

minstrel:
1. I have already decided the test is need as I am doing it. If I didn't think it was needed I wouldn't be doing it.
2. I believe that the way I am doing it will answer the question. As noted in my last post this orgainal test is not going to be the whole test.

How do I know the head start will not make any difference. It takes may pages 1 to 2 days to get crawled by Google and about the same for Yahoo. Google puts my pages into the index within 1 year and it takes about 2 to 3 weeks to get them into Yahoo.

The only other issues will be PR which at this point it looks like a toolbar PR update will be going to every 3 months. As the pages are both on the same site the sandbox effect will effect them both the same.

So given 3 or 4 months their should be no differences.

As stated before I pretty well know what the end results will be. Then what I will do is start removing the IBLs for "bob mutch" from the home page or move to another page.

I am not trying to disprove anything. "But you obviously are..."

You have the right to refuse to believe my verbal declaration. Perhaps you think you can read my mind and I can't read my own. If that is the case there is nothing I can offer you to change you mind on that one.

 

You're missing the point completely, Bob. It's not a matter of mind-reading. It's simply the nature of the beast.

If you are doing a "test" or an "experiment", you are obviously, by definition, trying to prove or disprove something. In this case, what you are trying to prove or disprove is pretty obvious from your own posts -- go back and read them.

My point is that I don't think the way you are doing it will answer your "question" one way or the other.

 

In the words of our friends on the other side of the pond-

Mistrel - don't get your knickers in such a twist.

you made you point, now move on and let him do his test anyways, maybe he wants to prove it to himself, but wants some pointers and feedback which you did.

Double blind placebo tests are never done on more than one drug at a time, yet how many medications are people taking? usually 5 to 8 different drugs untested for drug interactions between that many drugs. My question than is - are these double-blind tests flawed? or is it the application of the results?

 

Minstrel: With all due respect it is my verbal declaration that I am not trying to prove or disprove anything. Having said that you do have your right to believe as you will.

Just because I expect the test to go one way or the other, does not indication I am out to prove or disprove. I also don't need to go back and read my posts, I am well aware of my intentions.

I will however agree with you to some degree that the way I started the tests will cause some questions as to meaning of the end results. That is why after TOPS30 made his points I decided that I would at some point need to created another page what would be equel in PR vote as the bob-mutch.html page and point the 620 "bob mutch" links to that page.

First I wanted to see what will happend with the way I have it setup.

 

The .html page ranks #5 because it's an additional result. It's probably not #5 at all but only there because of the #4 result. Google groups pages together from the same domain. This bumps that one up. Delete the #4 page and you'll see the #5 disappear.

 

TOP30: So possible I should have these 2 pages on different domains you think?

 

For the test, yes. For ranking high, no.

 

TOPS30; Yes of course not for ranking high. I was refering to fairness of testing.