martinibuster - 11:28 am on Jun 13, 2013 (gmt 0)
Can a drop from 1.x% to .5% really be called a crash? Not really. I don't think so. Expressed as a percentage, 50% looks pretty bad. Expressed in actual clicks the story changes. The reality is often less drastic than the percentages seem to indicate.
50% fluctuations in low performing ad units are not unusual. In fact, one should expect 50% flucutations in CTR in low performing ads.
You can remove ads from the top of the page and be surprised by seeing a 200% jump in CTR from ads on the bottom of the page. When discussing low performing ads, percentages can exagerrate the situation.
Not to minimize your predicament. But the OP has been offered reasonable explanations by swa66. If it's possible to improve the Earnings (notice I didn't say CTR), then experiment with different sized ad units, color schemes (blend/no blend) and number of ad units. Limiting ad units tightens the competition and could increase the earnings. Keep in mind that I just listed four different variations that can be made to how the ads are displayed. Four variables to change means you can start with 24 experimental changes to test. The actual number of experiments is far higher. It's important to do it properly, so be sure to use your channels.