One for Colin here:

http://www.seo-theory.com/2011/07/08/why-seo-experiments-are-almost-always-invalid/

QuoteNeither test was sophisticated and both fail to prove or disprove anything

It does get on my tits when people claim fact with terrible stats and no control etc. And I suspect my annoyance is minimal comparred to Captain Colin.

Damn right - so many variables that people seem to ignore when testing.

I think the main point is that there are very few 'definites' that can be proven by testing - but what you can get is useful data which, if you are prepared to think in terms of 'probabllities' rather than true/false, will help along the way.

Bad testing is pretty much worthless - but not as bad as not testing at all.

You know, this place would be perfect for an seo test bed setup. We could determine which tests would be most important, best way to set them up, and then each participant manage their own % of tests to run. Keys, urls and info would be shared out in the open, ongoing, so all participants could see setup, results, control, etc.

Thats a good idea - if enough people want to do some testing

... of course, if it is just you and me, then maybe we should only let the others see the data if they send us a drink

I'm in

Couple of quick thoughts:

1) someone needs to take the helm. (not me)

2) to see the data you need to participate (private area only for those who are doing the testing) We can share general results in inner core if we want. This encourages people to participate, and we don't have to expose urls to those who aren't doing the same.

Trying to test factors and variables that occur offsite is probably futile.  There are lots of simple tests that can be run on site.  We've been having a lot of success with G Site Optimizer combined with Double Click for Publishers and their fine geotargeting capabilities.

QuoteTrying to test factors and variables that occur offsite is probably futile

I don't agree James - you can devise tests that pretty much 'rule out' certain types of links and techniques - of course, nothings perfect, and as time is a factor, and the tests take time, you are never going to get a definitive result - but that is not what we need - all we need is 'an edge', a probability that certain techniques work better than others, and a probability, say, that mass spamming directories has very little effect.

So, that's only 3 of us interested?

I'm interested but I don't have much which available to help at the moment. If I can, I'm happy to.

Me too if I can

I'm game

I absolutely think this is a good thing to do, and will do what I can to support the cause.  Unfortunately, I cannot lead something like this...and we need a leader atm.

Well, that's 8 but no leader.  Usual SEO group scenario eh? 

<everyone looks at Col>