On WebmasterWorld, netmeg did a test with a bunch of blended ads. Other ads, however, weren't as blended in with the theme of the site. Surprisingly, after viewing analytics, netmeg learned that blending hurts: the non-blended ads performed better than the ones that fit in too well with the site design.
netmeg is not alone. Other people notice that unblending is yielding better click throughs and cost per impressions.
How does this make sense? Here's a quote from the thread:
Ive always believed and used a 'third' colour to 'highlight' the ads as blending too much makes them look like the rest of the 'content' that gets skimmed rather than read, the third colour stands out, I tested it a while ago using CTR as a benchmark, the results were as you found netmeg.
So what do you conclude from this? First and foremost, if you're monetizing consistently, check your statistics to see how you can improve your campaign. Avoid blending a little too much. Above all, test and retest your copy and colors.
Forum discussion continues at WebmasterWorld.