Split Testing AdWords Ads the Wrong Way Costs Big Money

(Image credit: unsplash)

Running a split test on two AdWords ads at a time is inefficient and unwise. By running multiple copies of existing ads along with new ones, CTRs will rise rapidly.

Advertisers split testing AdWords ads in the standard head-to-head manner are costing themselves time and money. There is a better way to split test ads and it requires no more time or effort than is required for the standard method.

The Classic Method of Split-Testing AdWords Ads is Inefficient and Costly

The main problem inherent to this classic method is that the display position of a Google ad is based not only on keywords and bid prices, but also the past performance of an ad. Ads with higher CTRs receive higher placement on search results and are therefore more likely to get a customer’s attention.

When a new ad is written, obviously it has no performance history and its placement will be determined solely by the bid prices offered for each of the phrases on its keyword list. It will be placed at a disadvantage before it is even added to the campaign.

This leads to skewed CTR results. An existing ad may maintain a higher CTR than a new ad because of its prominent placement and not because of any superior ability to attract customers. In this way, a new ad that, in reality, is more attractive to web surfers can be passed over.

Another Disadvantage of the Classic Method of Split Testing

Both the new and old method of split testing requires the AdWords software to be set to display each ad in a campaign an equal number of times. If this is not done, AdWords will automatically display the ad with the highest CTR more often than the other ads in a campaign and the split test results will be skewed.

It takes a few rounds of split tests to find an ad that will perform even marginally well. The more tests that are run, though, the quality of the winning ad will rise and the likelihood of a newly written ad usurping its reign will drop.

New Google Ads are Statistically Inferior to Older Ads

Why would an advertiser take an ad that required multiple iterations to discover (and one that produces a lot of traffic) and relegate it to a position of equality with a random ad with no proven ability to create traffic?

This is one of the major flaws of classic split testing. Statistically, money is lost every time a new, unproven ad is displayed while an ad with a proven track record sits on the bench.

There is a better to split test that minimizes the traffic lost by inferior ads while simultaneously leveling the playing field between ads with a performance history and those without.

A Better Method of Split Testing Corrects All These Problems

To run a new split test, follow these steps:

  • Delete all ads from a campaign except the best historical performer
  • Create a few (2 to 4) new copies of the top performer (write the same ad again)
  • Write a new ad to be split test against the top performer and its copies
  • Any obvious failures will be apparent quickly
  • If the new ad is an obvious failure, delete it and all the copies of the top performer and start the process over again with fresh copies of the top performing ad and a newly written ad
  • If the performance of the top performer copies and the new ad is comparable, delete the copies one at a time until the new ad is running solely against the top performer
  • Let the head-to-head test run for a while and check the results
  • Determine the winner and use this free Split tester tool to determine the reliability of the statistical results
  • Delete the loser and repeat the whole process with fresh copies of the top performer and a newly written ad

Advantages of Split Testing More Than Two Ads

The copies of the top performing ad start from the same baseline as the new ad, thus their performance history is not an issue and the results will be objective.

If three copies of the top performing ad are made, the new ad will only appear 20% of the time. This will minimize the number of potential customers lost by a truly terrible ad. The copies of the top performing ad should perform comparably with the original.

This new method corrects many of the flaws inherent in the classic method of split testing ads. There is no real downside to this method. It gives better results with minimal additional effort. The money saved by employing this new method can make a significant difference in a company’s bottom line.

Did you find these tips helpful?

Got any nifty tricks or suggestions for us?

We would love to hear from you!

Posted in PPC

Leave a Reply