Setting up a SEO test environment

[MartinGaedeke]

Hi everyone,

I am currently working on my university thesis (concerning a German project). A small but important part of the project should be onpage SEO.
Until now I learned quite a lot about SEO, but after extensive Google search I still don't have the slightest idea how to _professonally_ set up a test environment... And I am getting the feeling that this kind of knowledge is very hard to find - perhaps because of the fierce competition in the SEO field...

Well, I know how I would do it for myself, but as this is a "scientific" thesis, I can't just start off - I have to justify my decisions!

My (own) plan currently is to get some domains (preferably with different IPs) and to set up a bunch of very similiar sites [url removed] All these sites would receive one incoming link to get them started.
But giving this to my prof without any underlying sources will just make him laugh...

So, if you know any sources, or if you have any hints that might help me, I would really appreciate it!

Thank you,
Martin

[edited by: msgraph at 6:34 pm (utc) on Aug. 15, 2005]
[edit reason] url removed; see TOS #13 [/edit]

[sitemaker]

Interesting. I don't know any sources as such, however here's some thoughts -

- The ideal would be to find perhaps five different industries that have the same amount of competition (ie, the same number of results in a search engine search). Choose some keywords for each industry. Choosing keywords with less competition will enable you to gain rankings quicker, reducing the time of what could be a very lenghtly project.

- Then build a site for each, making sure they all start 'identical' - in terms of number of pages, internal link structure, length of content on each page, and keyword density in titles, content, links etc... Naturally they won't actually be identical because they will be in different industries. This will remove any chance of getting hit with duplicate content filters.

- As you are only testing on page SEO, ensure each site has the same amount of incoming links, and that they all come from on-topic sites, with the same amount of PR. This is going to be the only thing you can't really control, but also a necessary evil in order to gain rankings (and, as far as I can see, fluctuations in rankings will have to be the measure of each sites success, so you need them in the first place).

- Test one on page factor at a time, record how it affects your rankings across several search engines, then move on to another.

This isn't particularly scientific, but then unless you get your very own Internet with your own search engines, it never will be.

Ideally all your rankings need to be the same across the test sites to begin with, as simply saying 'changing this title tag on site 1 made it move up 10 places' might be a worse change if that site was ranked 100 in the SERPS, than a title tag change on a different site moving it from 10 to 9. Who is to judge this sort of thing?

[ciml]

If you base it on Google then it is possible to use PageRank as a scale to measure other factors. This lets you plot a graph, which your professor ought to like.

[sitemaker]

Except the PR viewers available bear little or no relation to the actual 'pagerank' of a site. Certainly not accurate enough to please a professor ;-)

[ciml]

My professors were normally very happy to see high precision interpretations (with accuracy margins) from low precision equipment. :-)

Under controlled conditions, it is possible to infer quite precise values from the Toolbar PR readings of some URLs.

[sitemaker]

Well okay, perhaps. But here's the thing - he wants to test onpage SEO, not link popularity ;-)

Not to mention that SERPs are just a lot more empyrical even if you do setup a controlled PR testing environment.

[ciml]

sitemaker, my suggestion is to "use PageRank as a scale to measure other factors". This can make testing onpage factors easier. Testing most types of link related factors is less easy.

I'm not sure what you mean by "a lot more empyrical".

[jimbeetle]

What you can also do is a variation on the "who's the best seo" game. Pick a nonsense phrase of two or three words that shows no results in Google or Yahoo. Create a handful of domains and use different SEO strategies. You should be able to get a wealth of information.