A/B testing goes back to the early days of market research, 100 years ago. The idea is to test two different ads, markets, or other variables, and see which works better.
Originally it was applied only in direct mail, where it was easy to measure results by how many people sent in their money order to purchase the item.
If you tested too small a number, normal variations could drown out actual differences.
And you had to control for conditions, since a change in weather or a difference in income levels or other factors could affect results.
Fast forward to the 21st century, when Google AdWords (click ads) make A/B testing easy. You simply run two ads to the same search terms at the same time and see which produces a better response.
Run it long enough to produce a clear-cut difference. Then throw away the lower producing ad. The higher producing ad is the new "A". Write a new "B" ad and test that to see if it produces better.
Over time you can usually more than double the effectiveness of the ad. You'll also improve your quality score with Google. Both of these will reduce your cost per click.
One caution: Sometimes a higher click-through-rate (CTR) doesn't mean a better response, if you are getting more "garbage" clicks. So ideally you have conversion tracking set up (not only how many click, but how many of those buy or fill out your contact form).
Now you know what your market will respond to and you can use that same copy in your website, sales pitches and other marketing material.
This is so slick that one big-time marketer I know uses click ads to develop copy for his infomercials. He doesn't even try and sell anything through the click ads, it is for research purposes only.