Skip to Content

3 ways to test your responsive search ads

If you’ve been running Google Ads for a while, there’s a good chance you’re feeling bummed that expanded text ads (ETA) are no longer a thing.

ETA campaigns were easy to test, and it was easy to compare things like CTRs. Sigh.

But while it’s harder to run A/B tests using Google’s new responsive search ads (RSA), it is possible—and worthwhile, if you want to optimize your results.

You just need a different approach to testing.

Ashwin Balakrishnan offers three effective experiments to help you better understand and improve your RSA performance:

Experiment #1: Pinning vs. Mix-and-Match. RSA automatically combine various headlines and descriptions for testing, but their pinning feature lets you decide which headlines and descriptions appear, and where.

So create two RSA campaigns using the same creative, then pin a few headlines and descriptions in one ad, while leaving the other intact.

Running identical ads with ones pinned can help you determine which version gets better CTRs and conversions.

Experiment #2: Segmenting by message. This experiment helps you find out which topics enact engagement and what your customers are looking for.

Create multiple ad groups with a distinct theme centered around a particular keyword and use a broad match. You should look to target different personas, pain points, and even offers for topics.

Since everyone’s using the same Google automation, this will help you get an edge over your competitors.

Experiment #3: Pseudo-ETA with a Control RSA. You can create an ad resembling an ETA by pinning.

Create two of these “pseudo-ETAs” by pinning just three headlines and two descriptions. The third ad is RSA where you leave everything unpinned.

This helps you understand the CTR performance of pseudo-ETAs, especially if you’re in an industry where you need to display particular information at all times.

And there you have it: three experiments to try with RSA. So roll up your sleeves, enter your Google Ads laboratory, and start experimenting. Good luck!