About ad copy testing in an ad group

If you want to move beyond comparing ad performance metrics row by row, set up an ad copy test and let Search Ads 360 do the comparison for you. An ad copy test can determine which ads within an ad group are best at helping you to achieve your conversion goals by starting all the ads at the same time and displaying them evenly. The test automates the process and equalizes other variables, such as the device type, that impact ad performance. This focuses the test on accurately measuring and comparing ad performance based on the ad copy.

When the test completes, Search Ads 360 rates the relative performance of each ad in the test. You can pause poor performing ads or, if there isn't a significant difference, you can let all of the ads continue to run. 

Benefits of ad copy testing

An ad copy test provides the following benefits:

• Automates the process of testing and comparing the performance of ads in an ad group, saving you time and effort.
• If an ad copy test achieves your desired statistical significance threshold, the statistical analysis provides validation to enable you to confidently make decisions about your ad copy.
• The worst performing ads are culled from the ad group, enabling the ad group to optimize performance towards a goal.
• Helps you figure out whether your customers respond better to one phrase than another in that ad group. For example, does the phrase buy today perform better than purchase now?

• Focuses on the ad copy by equalizing other variables that impact ad performance such as:

  • Ad start dates. For example, if ad A has achieved more conversions than ad B or ad C, is it because ad A has been running longer than ad B or ad C?
    In an ad copy test, the performance of the ads in the ad group start on the same day and display for the same length of time.
  • Device type. Do the ads appear on the same types of devices? Does ad C appear on mobile and desktop computers while ad A displays only on mobile devices?
    The ads in an ad copy test have the same device-preference setting.
  • Ad display frequency. Did ad A display more often than other ads?
    Depending on the engine, the ad copy test automatically changes the ad rotation settings.
    Google Ads and Yahoo! Japan: The ad copy test changes the ad group-level setting to Rotate indefinitely.
    Microsoft Advertising: The ad copy test changes the ad group-level setting to Rotate.
    Baidu: The ad copy test changes the campaign-level setting to Rotate indefinitely.

Overview of the ad copy testing process

1. Select or create an ad group that receives, at a minimum, 10000 impressions per week.  See the next section for more information about the minimum number of impressions.
2. Decide on the goal such as click-through rate or action rate, that you want the ad group to achieve.
3. Choose the statistical significance that you want to the ad copy test to achieve.
4. Check the results of the ad copy test to compare ad ratings, observed value, confidence interval, and metrics such as clicks, impressions, and cost.

Search Ads 360 continues to report keyword metrics during an ad copy test.

Ad copy test requirements

To run a successful ad copy test, the ads and ad group must meet the following requirements:

• Ads included in the test are active and the device preference is the same, either all devices or mobile-preferred.
• The ad group contains a minimum of two active all-device or mobile-preferred ads. There isn't an upper limit to the number of ads that can be included in a test, but keep in mind that the ad group should receive enough impressions, an average of 5000 per ad per week, to measure statistical significance.  For example, if your ad group contains 10 ads, the ad group as a whole should receive a minimum of 50,000 impressions. One ad can receive 7000 impressions, while another ad can receive 2000 impressions, and so on.
• Test Google Ads expanded text ads in an all-device ad copy test. You can't test expanded text ads in a mobile-only ad copy test or with standard text ads. 
• Ads are included in only one ad copy test at a time. 
  You can simultaneously test two different sets of standard text ads in one ad group only if the tests target different devices.  That is, one test targets "mobile preferred" and the other targets "all devices".  

  If you're testing expanded text ads, only one test is allowed per ad group.
• Ads aren’t edited during the test. Changes to the ads in the test make the test results inaccurate, but the test will continue to run. 
• Each ad contains only one landing page URL.
• The ad rotation setting for the campaign or ad group is set to rotate ads evenly. When you create an ad copy test, you are prompted to let Search Ads 360 change the ad rotation. Note that for Google Ads, Yahoo! Japan, and Baidu, the change applies to all ad groups in the campaign. For Microsoft Advertising, only the ad group ad rotation setting is changed.

Best practices for running ad copy tests

To get the best results from an ad copy test, you may want to implement the following best practices.

• Test higher volume (at least 10,000 impressions per week) ad groups because your tests will achieve better results. To identify high-volume ad groups, first check the weekly impression total.
• Create a name for the test that includes the ad group name and a description of the test.
• When a test is running, don’t change the ads that are included in an ad copy test because the test results won't be accurate.
• Verify that all ads that are included in the ad copy test are active and approved. Disapproved ads won’t collect data.
• Check the test results every few days. If enough impressions are achieved in less than a week, let the test run longer if more testing is within your budget. A longer test reduces the effects of anomalies or seasonality on test results.
• Test one component of ads copy per test. Use the information gained from one test and test a different component in the next test. That is, keep experimenting to find the best ad copy. For example, in test 1, you vary the headlines. Ad 1 performs best. Then, you create test 2, including Ad 1. But in test 2, you vary the display URL, and Ad 5 performs best. In the next test you use the headline from Ad 1 and the display URL from Ad 5 in one ad, and so on.