Google may have made it easier for marketers to turn a blind eye to text ad testing with automated ad serving, but there are still advantages for those that commit to even a simple cadence of testing ad copy in their paid search campaigns.
“Your ads and extensions are the only part of your account searchers see,” said Mike Henderson, CMO at SuretyBonds.com, in his SMX Next presentation on ad testing within the Google Ads platform. “If you’re tweaking title tags, meta descriptions and content for SEO Company, you should be testing your ad copy for PPC.”
The benefits of ad copy testing can extend to your other marketing agency efforts as well, Henderson points out, including email, display and social media campaign copy. Once you get into an ad testing rhythm, there is a “flywheel effect in which you build off of successes to continue seeing better and better results,” he noted.
PPC ad testing best practices
To get started, first you’ll need a plan. Set time aside each week for testing — reporting and iterating. Henderson marks an hour to an hour-and-a-half on his calendar Friday mornings.
You’ll also need to disable Google’s automated ad rotation and change your ad rotation settings to “do not optimize.” This isn’t 100% effective, but helps ensure a more even ad delivery split across the ads in your ad groups.
Lastly, Henderson recommends starting with making small changes to your best performing ads using the “copy and edit” function in the UI. (You can also use the Google Ads desktop editor if you want to work outside of the web interface. This lets you make and review changes in bulk before uploading them to your account.) The reason Henderson suggests starting this way is because you’ll be able to tell if the small change was the lever that made a difference, whereas if you start with all new ad messaging it will be harder to tell what’s driving performance changes. The caveat is sometimes minor changes have little effect.
A simple ad testing framework
Henderson said for the five or six years, he and his team have been using a framework that ensures ongoing, iterative ad testing and performance improvements.
The first step is to find opportunities by filtering your ad groups report either by impressions or clicks depending on your goals, volume and statistical significance thresholds. Pause the low performing ads and then start testing a change on the high performers.
Labels are your friend. Henderson strongly recommends using labels to capture the test and date the test launches on those ad groups. You can also use labels to help track performance through the funnel, notes Henderson. They make reporting much easier.
Simply filter your ad groups report by label and test the date range to the start date. Then filter by your click or impression threshold to ensure statistical significance. When you’re ready for a new test, be sure to add new labels with the new launch date.
Limited time? Try Google’s Ads variation tool
Henderson runs through using the Ads variation tool if you don’t have the time for weekly testing. “This can be great for small accounts,” says Henderson.
The tool allows you to run one test across multiple campaigns or an entire account and has a statistical significance indicator to help inform you when enough data has been collected to decide a winner.
Machine learning + human experience
If you’ve reluctant to test responsive search ads, or did once and it did not go well, it could be time. The automted ad formats aren’t going away, and there are steps to take to improve outcomes.
“You have to learn to tango with new features when they come out,” says Henderson, speaking about responsive search ads (RSAs) specifically. He says they have found a lot of success with pinning elements that worked in other text ads in their RSAs, for example.
That echoes Snaptech Director of marketing agency Services Amalia Fowler’s practice for newer accounts that she shared in a column last year. She runs ad groups with only standard expanded text ads (ETAs) “until I’m confident I can identify some high performing headlines and descriptions, and add RSAs at the end of month one or two, depending on volume.”
As I emphasized in my SMX Next keynote, to succeed in this age of automation, you need to understand how the feature and the machine learning systems are designed to work and the levers available to inform performance. “The winner is going to be the organization or paid search professional that can take the power of machine learning and combine it with their expertise and knowledge and understanding of the customer,” says Henderson.
Watch Mike Henderson’s full talk here (with free registration).