In my last post I mentioned how I was testing 3 Ad Formats and it took me very little time to eliminate one of them that was very bad for my particular page. So now I am back to 2 Ad formats to compare.

To accelerate the test results and not waste potential earnings in the process, I decided to weight Ad exposures according to eCPM.

For you folks not up to speed with Ad performance, eCPM is Earnings per 1000 Ad Exposures so eCPM of 50 means $50 earned on average for 1000 exposures of a particular Ad.

So if you have 2 Ad formats in random rotation in the same spot, you will get different values of eCPM for each Ad. However it may vary over time of day and throughout the week, month or season.

So you don’t necessarily want to eliminate the lower performing Ad since it may be a better performer at various times.

So if Ad A has eCPM of 3 and Ad B has eCPM of 7 (depending on what time frame you evaluate it), the optimum way to display the Ads is to show Ad A 3% of the time and Ad B 7% of the time and adjust these ratios as more data is made available.

Basically, what this means is that you are still testing both Ads but should not lose much in earnings by still displaying the current lower-performing Ad. It’s not rocket science but it is not far off Don’t worry, I was good at math when I was at school.

The easiest way to do this is to use my script and manually adjust the eCPM values as you check your Adsense stats. Then you can choose how often you do this and work out what works best to optimize your own profits.

Here is the PHP code to integrate with your web template in the position where you want to split test your Adsense code:

As usual, please remove the space in “< ?php”.

Tags: adsense

This entry was posted on Sunday, May 18th, 2008 by Andy and is filed under Ads. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.