A/B Testing of AdSense

What is acceptable?

[dataguy]

Since joining WebmasterWorld, probably the most beneficial advice I have received is to constantly do A/B testing to determine the best performing ads. The best way that I've found to do this is to place competing ads on a page, and use a random number generator to display each ad 50% of the time. Using this testing method, I have increased my ad revenue substantially.

I have stayed away from performing these tests on my AdSense pages since I recall reading something in the AdSense TOS about dynamically generating the ad code being against policy (though this isn't really dynamically generating the code). I can't seem to find this anywhere in the AdSense policy now and I was just wondering what the consensus was here, and how others are doing A/B testing with AdSense. Because I'm displaying competing products (though not at the same time as AdSense) I don't think it would be appropriate for me to ask the Google team about what is acceptable.

Any thoughts?

[hdpt00]

If you're using some type of PHP server side script or what not to choose randomly to display skyscraper or horizontal ads, lets say, and the code that it inserts is exactly how it should be, I doubt google would 1) know you're doing it and 2) even care.

As long as the code is remaining the same after the randomizer script runs I think you'll be fine. I use some php include statements so I can modify one file and change the type of ads I show. The code inserted is exactly how it would be if I hand coded it, so when the page source is viewed, it looks exactly how it should.

Am I getting what you're asking correct?

[dataguy]

Yes, I use the same include-page method so I can change the ads across a web site with much less effort, and you're correct that if done correctly it will look exactly like it was hand-coded. The difference is that half the time the pages will display ads from AdSense, and the other half of the time the ads will be from a competitor.

[hdpt00]

I am pretty sure as long as they aren't on that the same time it should be fine. I could be wrong on this though. Maybe someone like jenstar with more knowledge on the TOS can explain. My gut instinct is there should be no problem, especially since you can't find anything in the TOS saying it isn't allowed.

I don't think they could restrict you from doing that, much less could they figure it out. They will just think you get half the page views you actually receive.

[richmondsteve]

dataguy, this is perfectly acceptable. It's not against the AdSense Terms and Conditions. I due random server-side rotation b/w AdSense, affiliate ads and internal site promotion by allocating each type of ad a percentage of impressions and in some cases taking into account ads displayed to a user in order to keep repetition below a certain threshold. I've shown 7-figure AdSense impressions over the past 55 weeks and haven't had any concern that Google would have a problem with it.