The Optimizer’s Guide to Google AdWords: Copy Testing
For this week I’ve invited Chad Summerhill, a PPC expert, to talk about copy testing. Chad provide a great comment on one of my early posts regarding adgroup triplets. – @JRachwalski
____________________________________
Have you ever heard the saying, “Hope for the best, prepare for the worst”? This is good advice for a copy testing strategy.
Every time you run a new PPC copy test you are opening yourself up to losing just as much as winning. It makes good sense to do everything in your power to prepare for a losing test copy and to minimize your risk.
Consider this scenario: you add new test ads to most of the ad-groups in your PPC account, planning on watching the results very carefully. A week goes by before you check on the progress of your tests only to find many losers. Well at least you’re always testing. Right?
Unfortunately, every time you run a test and your test ad doesn’t win, you leave clicks, conversions, and money on the table. Luckily, there is a great way to help lessen the risks involved in ad testing.
Adding Multiple Copies of the Control to Limit Risk
If your ad-group has enough traffic, using multiple copies of the control ad is a great way to lessen the risks involved in copy testing, especially if you are going up against a strong performing champion ad.
By only using two text ad versions (the control vs. test copy) then you are exposing 50% of your traffic to an untested test copy. This may not matter if the test ad is, in fact, a winner, but what if it’s a loser? The cost of missed opportunity could be very high.
If you were to include the champion ad, three exact copies of the champion, and one test ad, now you are limiting your risk to only 20% of your impressions. You can still find a winner this way, but you are ensuring that a majority of your impressions continue to be of the champion copy. After all, it probably took lots of hard work to find the current champion.
Perhaps you have been testing a while and the current champion is a superstar. It may be wise to have four or five copies of the champion ad running against a test ad. If the challenger performs better than the control, you can start removing the control copies until you have proven that the challenger is a statistically significant winner (Performable calculator) over the original champion.
There is also a free Excel download for determining statistically significant PPC text ad winners that solves for CTR, CVR, or Imp-to-Conv.
Considerations When Using Multiple Controls
There is another added benefit of using multiple controls. What if the test results of your control copies vary widely from each other? When running concurrent versions of a control (champion) ad you should expect to see a very similar performance as your test matures. If not, you may need to investigate further. Variation in performance could be an indication that there are factors outside or inside your control that are affecting your test. For instance, maybe you started a landing page test during the middle of an ad test, maybe a holiday, etc.
Also, AdWords doesn’t ensure even and random ad serving. The best you can get is “Rotate: Show ads more evenly”.
Regardless of whether or not you use multiple copies of your control in your PPC copy tests, you should always watch the results closely. The sooner you identify a losing text copy the sooner you can turn it off and move on to the next test idea.
Chad Summerhill is author of the blog PPC Prospector, provider of PPC AdWords advice , and In-house PPC Specialist at Moving Solutions, Inc (UPack.com and MoveBuilder.com).

Follow us on Twitter
Grab the RSS feed








