A/B testing is when you test two (or more) different situations: situation A vs. situation B. A/B testing with AdSense is important for finding what ad designs will earn you the most revenue. For example, you can test how the 728x90 leaderboard fares against the smaller 468x60 banner.
So how do you A/B test? There are two important parts of running an A/B test:
- There can be only one difference between each situation
- You must run both situations simultaneously
These two restrictions will allow you to identify the impact of each change. For example, if you change the color of your ads in mid-December, it'll be difficult to tell whether an increase in clicks is due to the new colors, a new relevant ad campaign, or simply users shopping for Christmas. Instead, it would be better to run 50% of your traffic with the old colors and 50% with the new colors, so that you can directly compare the results. To easily compare results, be sure to use a different custom channel on each format.
You can find a tutorial for implementing A/B testing on the Inside AdSense blog.