So I'm going to try an experiment and I figured I would share it here to see if anyone has had any experience with it already. For example, I've found that multiple palettes on a single ad has better results that a single color scheme, no matter how well blended. Even subtle changes help for the most part. So I'm thinking the same reasoning may work with entire ads. Google provides some Javascript code for A/B testing. So I figure this is within TOS. Anyone try this before? PS. For clarification: Testing to see if randomly displaying different types of ads in the same spot helps overall CTR
I'll go you one better even...A year or two ago I did an A/B test on layout which required most of the page to be generated based upon a random number. It went well beyond the little modification they give you for doing it above. I cleared it in advance with Adsense by showing them a mockup of what I intended to do, invited them to check the source code, explained what I was doing and why and they gave me an almost-immediate "permission to proceed" email. I ran the test for about two weeks analyzed my results and went with one of the options I was testing. This was a while ago, so I don't know if things have changed...I have not read anywhere that they have. If you're feeling even a bit leery, just drop them a line letting them know.
Thanks for the reply. I'm not too concerned with the TOS part of this since it's fairly simple code provided by Google themselves. My main goal is seeing whether it is effective to keep changing the ad layout or to stick to a static one.
It's an interesting idea... but the trick to getting users to click is making the ads relevant to the users' search not necessarily changing positions constantly. Though, it'd be interesting to see the results.
It's not so much to be continuously changing the positions, though. It's to run the test for a while to see WHICH of those positions is better (or confirm a "sneaking suspicion" as in my case) and then run with the pages having the ad continually in that position. You're right, though, in that if the ad isn't relevant, the results won't be too good no matter where the location is.
This code only helps if you intended to check which color works best. The ideal would be changing the whole page so you could create 2 diferents layouts of the same page with the ads in diferent positions and the code would show each page 50%.
That is what I did...I asked permission from Adsense to do a complete A/B page generation (because it required a lot more than there example showed) and they approved my experiment almost immediately.
The page itself was mostly the random number generator, then depending upon what the random number was, a large chunk of the page was displayed through #includes which were mostly document.writes since this was INSIDE the javascript of the random number generator which is where the problem arose that required additional work beyond what they have in their example.