How to Execute a Flawless Ad Copy Test in AdWords

The best paid search programs have one thing in common with even the smallest programs: testing leads to breakthroughs. Our goal as search marketers in many cases is to efficiently acquire a high value customer. Having a disciplined dedication to testing is at the core of achieving this goal.

The execution of a test is equally as important as the strategy that is driving it. Confidence in the outcome, backed by statistically significant data, will allow you to roll out learnings within the ad group you have tested and hopefully throughout other areas of your account.

Where to Start

Taking over a new account and trying to identify your first ad copy test can be overwhelming, especially if you’re working with hundreds or more ad groups.

Start with an area of the account that can deliver the most performance improvement quickly.

One way to do this is by simply filtering your ad groups by spend. Take your top spending ad group and start there.

Don’t worry about tanking the account’s performance due to poor performing ad copy within the test. We’ll put safety measures in place to avoid this.

Chances are the highest spending ad groups are pockets of opportunity that will gain the most efficiency through testing.

To illustrate, we’ll use the fictitious company BreesTravelSite.com. After sorting the ad groups we find that one of our highest volume ad groups (both in spend and clicks) is Italy Travel Packages.

Within this ad group there are six keywords that are driving a large proportion of the ad group’s volume. We’re going to select three of those keywords to base our ad copy test on. This will ensure that at least half of this ad group’s volume will be unaffected by the test itself — one of several safety measures.

Create a Test Campaign & Ad Group

Take the keywords that will be used for your ad copy test, and create a new campaign and ad group that will be used to house the test. We’ll call the ad group Italy Travel Packages_Test. When it’s time to launch the test, we’ll go back to our Italy Travel Packages ad group and pause those keywords — but we’ve got more work to do setting up our test ad group first.

Now that we know where we are testing, it’s time to look at the ad copy itself and determine what attributes within the copy we want to test.

Warning: Don’t test too many things at once. Keep it to a single variable so that you know exactly what caused increases or decreases in performance.

• Set a Control: You’ll want something to compare your test ads against. It’s always good to take your historically top performing ad creative and use it as your control. Our objective is to see if we can outperform our control ad by measuring CTR, conversion rate, and cost per acquisition (CPA).
• Create your Test Ads: There are many different attributes within your ad copy to test. The headline, description lines, and display URL all present testing opportunities. In this example, we’re going to take our control ad that has already undergone heavy title testing. Now we’re interested in seeing the lift we can achieve by testing multiple description lines.

In the example above we’re looking to see if the consumer will react more positively to a price point savings versus a percentage off. Or perhaps we’ll be surprised to find that “Best Price Guarantee!” resonates the most (See: “Secrets to Paid Search Success Revealed, Part 1,” for more ideas on ad copy testing attributes).

Once you’ve uploaded your control and test creatives into the Italy Travel Packages_Test ad group, there’s an important campaign-level setting that must be activated for your test to produce an accurate result. It’s important that all three ads have the opportunity to be served as equally as possible throughout the course of the test.

In AdWords, the setting is “Rotate: Show ads more evenly.”

Additionally, you need to create new uniquely tagged destination URLs so that you will be able to differentiate backend performance (conversion and CPA) from the test. It’s critical to the integrity of the test that each ad uses the exact same landing page.

The setup of these URLs will vary depending on the technology you use to measure conversions. If you’re tracking conversions through Google Analytics, your URLs will look something like this:

• http://www.breestravelsite.com/?utm_source=Google&utm_medium=CPC&utm_term=Control&utm_campaign=Test-Travel-Packages
• http://www.breestravelsite.com/?utm_source=Google&utm_medium=CPC&utm_term=Test+Copy+1&utm_campaign= Test-Travel-Packages
• http://www.breestravelsite.com/?utm_source=Google&utm_medium=CPC&utm_term=Test+Copy+2&utm_campaign= Test-Travel-Packages

Once your test campaign and ad group are set, your keywords and ad copy are loaded and you have worked through the appropriate settings — you’re just about ready to launch your test. The final step will be to pause the test keywords in the Italy Travel Packages ad group, and immediately launch the Italy Travel Packages_Test ad group.

Measuring Test Performance

Another important safety measure is to set an abandonment point. We’re going to take our control ad’s historical CTR (1.9 percent) and conversion rate 10.4 percent), and say that if any of the ads performs 30 percent less than these metrics that we’ll pause that ad immediately.

The time allotment for an ad copy test will vary greatly depending on how much click volume that particular test is receiving. We like to use a minimum of 100 “actions” per ad before we determine a winner. In this example, an action will be a click.

There are times when two or more of your ads have thin variances. In these instances you’ll want to continue with the test until more separation occurs.

Let’s look at our test’s outcome.

It looks like we’ve found a pretty compelling winner in Test 1. When comparing Test 1 to our control, we see a 42 percent lift in CTR, a 24 percent decrease in CPA, and a 30 percent increase in conversion rate. Not bad.

Leveraging the Learnings

Knowing with confidence that Test 1 is the strongest performer, we’ll add it to our Italy Travel Packages ad group with serving preference so that it garners the majority of volume. We can also apply the learning that consumers are gravitating more towards the price point version of the ad, and use this description line in other ad groups where appropriate.

Some tests will result in profound impacts on your account. Others won’t.

Regardless of the outcome, it’s important to create a process around ongoing testing within your account. Remember: Never stop testing.