Google AdWords allows you to choose how your pay per click ads are served. You can have them served by optimization level (the Google AdWords default) or by rotating all ads evenly.
The optimize ad serving option rotates your ads based on CTR. The higher the CTR the more often the ad will appear to searchers. By contrast, the rotate evenly option delivers all ads at the same rate regardless of CTR. Jeremy Mayes at PPC Discussions wrote an interesting post a couple of days ago in which he thinks the best practice is to start with the optimized default setting then switch to rotate evenly in the second round of testing.
Essentially, the way that works is like this: You test several elements at once by creating a handful of ads and rotate them into the mix by allowing Google to choose the one with the highest CTR. Naturally, on Day 1 those ads will be rotated evenly since none of them have a CTR. But after a few iterations of the rotation schedule Google will begin to rotate 1 or 2 of those ads in more often due to the developing CTR. This is a savvy way to conduct multivariate testing on your PPC ads. Then, after figuring out which of your elements are working the best, you take the ad with the highest CTR and use it as a control in an A/B test against another ad using the rotate evenly option.
I can see where this might get you to the ad that you want to use quickly, but I don’t know if I’d call it a best practice. Whether you conduct A/B testing or multivariate testing first, I think, is a personal choice. I like conducting A/B testing on my landing pages first as this gives me the ability to test specific elements one at a time and narrow in on those elements that do well. It might take longer to go that route than to test multiple elements at once, but I think you get better results in the long run. But landing pages don’t cost me money while I’m testing except perhaps from those lost opportunities to convert while those elements that I toss out are in the batter’s box. Those moments cannot be helped and the information I gain from them helps me to improve my optimization so that I earn more from the landing pages in the long run. The question is, will that same strategy work for ads or is it better to conduct the multivariate testfirst?
I think anything that allows you to find the right ad quickly is a plus. If you are testing every element of an ad (headline, description text, display URL, destination URL) and you have more than 2 or 3 variations of each element that you are testing then perhaps the multivariate test is the best way to go. But if you only have two ads and you know those two ads are the ads that you want to test then the A/B test is probably the best way to go. In that case, I might go with the rotate evenly option so that I’m comparing apples to apples.
Thanks for the wonderful post. In pay per click advertising, to get a better result we have to do more test on the ads. The test can be either A/B test or multivariate test. As for as in PPC campaigns, the ads reaching the users at first than the landing page, so the ads should be more specific and catchy.