It’s been said a million times before, you need to keep testing in affiliate marketing. You just never know what may or may not work until you actually commit some money to testing. We’ve all heard it before, but here’s an interesting case in point.
I was testing a campaign that has been running on Google for about 8 months now, just chugging along. I had 5 ads that I was testing for a particular Adgroup, and in one ad I noticed that I had accidentally left out a word in the description that made it read like total nonsense. It had ran that way for about 2 weeks and the funny thing is, that ad performed about 23% better than the other ads. I have no explanation for why that is, but you can bet I left that messed up ad in place 
Perhaps people click it out of curiosity?