hmm, bad grammah on my behalf.

page A, fully optimised for phrase "web design"; links from participants being <a href="http://the.page.com">web design</a>

page B, no optimisation, not mention of key terms in content or source; links from participants being <a href="http://the.other.page.com">ipsum</a>

The purpose of the control group B isn't for publication, it's to isolate the effect of hotlink on SERPs from SEO webpage optimisation. Without it, the experiment won't specifically target any aspect of SEO, just SEO in general. There's many aspects of SEO page optimisation that play a part in SERPs. The whole point of the experiment is to test hotlink text.

When the results of page A and B come in, we will have a good idea of just what proportion of page A's SERPs can be attributed to incomming links and the proportion attributed to webpage SEO.

For the record, I don't know why ppl in this forum call it a 'hotlink,' that's a promotional term. Why don't you just calll it a 'link.' What's so hot and s e x y about it?

How can you compare how one approach works in one set of serps against how another approach works in another set of serps? Seems far from a control test to me.

If you are going to compare something it needs to be the same domain, same testing ground (serps) to exclude the questions of 'oh one domain had factors in it's favour' or 'those serps are treated differently because they are in a more compeditive arena.'

.............????????!!!!!!!!!!!!!!

Originally Posted by Dominic

If you are going to compare something it needs to be the same domain, same testing ground (serps) to exclude the questions of 'oh one domain had factors in it's favour' or 'those serps are treated differently because they are in a more compeditive arena.'

Perhaps. But maybe we could go about this with two almost identical domains, thus still having that factor under control:

What if Page A were broken down further into a test page and a control page--they will be exact copies of one another, except for the domain names, which will be similar enough to not matter (example: asdfwebdesigna.com and asdfwebdesignb.com). We will put the exact same page on each of these sites; this page will contain the term "web design" a few times in the content, in a header, in the title, etc.

We will then conduct the experiment by having many people link to one of these pages, exactly as metho mentioned:

Originally Posted by metho

page A, fully optimised for phrase "web design"; links from participants being <a href="http://the.page.com">web design</a>

These pages will be indentical in every way possible, thus eliminating all possible factors that could influence the sites' rankings (except, of course, for the incoming links).

We would then perform the analysis as dominic and metho have both been discussing--monitoring the rankings of these two pages over the course of a certain amount of time and comparing the difference in rank between them in each of the three major search engines.

What do you all think of this? It would take a while for the sites to be fully indexed (as they are going to be newly registered domains), but this would also give us the benefit of being able to recruit more links over time.

Go for it Webdev, I'm bowing out. We clearly wish for 2 different goals of the experiment. Feel free to start another thread with a clearly defined structure to avoid confusion.

Dominic, 'hotlink' is used in the context of SEO as a link which contributes to SERPs by using key terms, as opposed to a 'coldlink' which only contributes to PR when no key terms are used. This seems to be a fairly popular use of the term in the blogs/articles I've been reading of late.

Hi guys - I am very interested in helping out.
I think we need to figure out 2 things.

1. What phrases will we target?
2. Who will lead this experiment? Obviously we already have disagreements in the experiment's execution. Lets figure out which people will actually do this and those people can decide on how to build these websites (with the input of fellow board members).

I'd be up for a democratic vote if need be, but this would mean we'd actually need people to participate in such a vote in the first place.

I'm up for targeting a high-profile term such as "web design", "web development", etc. It doesn't even necessarily have to be web-related; we could target "cat food" just the same. I would be willing to set up and purchase the hosting and domain names for the two sites once we have enough support for this project that I can see it will work. I'll also develop the content for each, which will consist of a page relating to the keyword that we are targeting.

myshtern, are you interested in participating in this?

Making two pages exactly the same or even similar will thrip the duplicate content filter.

Dang; good point. What do you propose is the best way to go about this?

Actually--we could have page that have jibberish content with the exception of the keywords that we are looking for. Let me know if you think the following would work:

Page A
Test random words banana apple web design money tea pot doorknob design with and the web because design...

Page B
Fork badger great yellow grape web design spoon art the happiest design plan red rob web dwindle design...

The text would be, overall, different. However, the ratios of the targeted keywords to the rest of the content (in terms of both number of characters and number of words) would still be equal.

Any other ideas?

Still two different testing grounds (different serps).
Don't you want to test the difference on page factors achieve?
Also you will need to use natural language as semantics are in play.

I would be willing to participate in this scenario:

Page A: This page has no mention of the keyword phrase in the content but has the anchor text 'web design' pointing to it in the exact same instances as:

Page B: This page does contains the keyword phrase, also linked to in exact same way as Page A.

They can be on the same site or different sites, but it must be the same anchor text going into each page so we can compare the difference in their performence in same set of serps (same test environment).

Okay. So we randomize the order of the two links for each page that is going to be posting links up, correct? This would help to eliminate any possible error due to weight change depending on positioning of a link within a page.

Do you want to be Page A and I'll make a page for Page B?

OK, was it 500 words? Can't remember...

Since this has been done more than once, a better experiment might be to try and break the sandbox effect on a new fake commercial site. Go with a site, built from scratch for the experiment. Sell something very competitive like SEO or web hosting.
If you can get it to number 1, sell it and give half the $$ to charity.

Originally Posted by DevelopedWeb

myshtern, are you interested in participating in this?

I am very interested in participating in this.

IMHO, we need to just figure out who will do this.
These people can get in a chat or phone conference and discuss the details of this experiment.
Otherwise, nothing will get decided with this many people involved.

I'm up for debating everything openly, at least until things get too confusing. The purpose of this thread is more to get the details of the project set up than to actually promote linking.

I went ahead and made a very basic dummy site to target the "web design" keyphrase with--a ZIP file of the three HTML documents is attached to this post.

Dominic, if you want to do what you had suggested earlier and keep the domain name constant (combined with the fact that your domain is already spidered by Google), I would recommend posting these files in a "webdesign" folder on your site.

You'll notice that there is no actual contact information on that dummy site. This is intentional, as the goal of this experiment isn't to provide anyone with some easy free promotion; it's just to see how certain variables affect search engine rankings.

Let me know your thoughts, everyone!

Looks good
I would create a 2nd site, but I dont have a home computer at the moment.

I forgot one thing though
What are we testing for?

Quality links vs. Quantity of links?

We're testing relevancy of pages targeted with inbound links in two scenarios - one with a page that is targeting the given phrase ("web design"), and one that is filled with completely unrelated text, thus only achieving a rank due to the inbound links.

This will test if inbound links help rankings (and how much they do, more importantly), and it will also test the question of importance of the inbound links alone by testing it as the only possible ranking factor in a certain situation.