Google Test Lab For Experimenting With Google's Rankings

Apr 25, 2014 • 8:33 am | comments (13) by twitter Google+ | Filed Under Google Search Engine Optimization

Rand FishkinRand Fishkin, the co-founder of Moz, announced on his personal blog a new thing he is playing with he named IMEC, aka Internet Marketing Experimental Collective.

The goal is to increase the number of people experimenting with Google's search results, so they can give more validity to the experiments they run. Rand said, "The goal is to assemble a group of marketers who care about producing high-quality, consistent, repeatable experiments in search engines and on social media to learn more about how these platforms work."

Here is an example of an experiment one would run:

click for full size

If you want to be part of this, apply over here.

I can promise you two things:

(1) The results will be interesting and

(2) The results will be debated.

Forum discussion at Google+.

Previous story: Google AdSense Electronic Payment Timeline



04/25/2014 01:12 pm

The goal is to assemble a group of marketers who care about producing high-quality, consistent, repeatable experiments in search engines and on social media to learn more about how these platforms work.

Michael Martinez

04/25/2014 01:48 pm

No need for debate. Point number 4 kills the potential value any such attempt might have had: "Even after completion, specific details of tests (e.g. keyword phrases tested in search results) should be kept private so as to limit future risk." The fact an elite group of people can repeat the results of an experiment doesn't validate the experiment. The experiment must be open to all, accepted by all, and repeatable at any point in time. And, frankly, when it comes to experimenting with search results, that last point is a really big issue. You might be right about something for a day.

Mike Weaver

04/25/2014 01:50 pm

The above comment is spam. I didn't click the link, but the comments from "Mary's" profile are all spammy... one talking about porn. Click at your own risk, especially if you're at work!


04/25/2014 04:17 pm

I agree Michael. This looks like burnt out SEO´s clutching at a new set of straws to clutch at. Selling old rope as new to people who don´t know better in most civilized cultures is fraud. How about Rand and his band read the google guidelines and instead of advocating smoke and mirrors they start explaining to people that the old way of SEO is well dead.


04/25/2014 07:37 pm

Michael and Harry--- As someone who applied for the IMEC Lab team, I can tell you that, while I don't agree with point #4, the experiment does still sound interesting and the risk of being wrong about something is a risk I personally accepted. Every experiment comes with risk. I just think an experiment like this shouldn't be only about Google --- there are search engines we have yet to explore. Harry--- if there's one wonderful thing about human life, that's curiosity. I don't know the intent of others, but I'm not just a freelancer and a marketer--- I'm also a Computer Science student and I just got started with Information Retrieval. An experiment like this is a blessing and touches many of my scientific and marketing interests. :) - Luana

Ben Guest

04/26/2014 01:06 pm

Isn't it a waste of time experimenting on something you have no control over? If it isn't a controlled experiment then you are literally peeing in the wind.

Michael Martinez

04/26/2014 03:00 pm

Experimentation doesn't have to have control over the phenomena being studied. In fact, many biology experiments are conducted in exactly that way. You "control" the experiment by using a baseline group of tests where you know what the results should be and compare those results to the results from the group(s) of tests evaluating a hypothesis (or more). This IMEC thing is about as pseudo-sciency as Internet Marketing can get, though. I don't expect much accurate information to come from it at all -- if any.

Ben Guest

04/27/2014 01:21 pm

For information dealers, this will work great. They'll be able to provide those case studies to draw traffic to their ads or product. However, this former integration specialist will tell you it's a waste of time when you can't control the versions you're experimenting against. It's my humble opinion. It's not up for debate Micheal. Love you though. ;-)

Michael Martinez

04/27/2014 02:17 pm

"'s a waste of time when you can't control the versions you're experimenting against." I've been saying as much for years. Glad to see I'm not the only one standing this watch.

Lennard Heinrich

04/28/2014 02:42 am

it is so cool to do such experiment. But i am so confused about the theory. I have test the something else like you do. I set different anchor text but which are related. And do the ranking, and they work well, The none anchor text is only part of it. DO NOT just follow the google rules, Cuz you know he don't wanna us spam.

Ben Guest

04/28/2014 12:29 pm

And you have a lot more "control" in biology as the evolution there is a lot slower than the evolution here in technology. And just maybe there is a ton more manual action within Google than automation through a set of rules and filters. *hint-hint* This may be a great tool to draw traffic to your site to see those ads though. Especially once you start reporting your "findings". lol Have fun kids!

Phil Zeo

04/30/2014 09:26 am

Testing is part of the "culture" of what we do as internet marketers. We've all tried to find the magic combination that guarantees top positions in Google's game . To be able to do it full time with a solid team, more power to Rand. Rand and his team will get great data, and in the process bring more visibility to his site or whatever site these "redacted findings" will be published. Make money while having fun with Google. Nothing wrong with that! Sign me up! Why all the criticism? Anyways he said repeatable results.

Gracious Store

05/03/2014 02:02 am

Are these experiments geared towards intellectual "gymnastics only" or will the experiment attempt to change or influence the way search engines rank sites?

blog comments powered by Disqus