Responsive Search Ads: What We’ve Seen So Far

Earlier this year, Google rolled out Responsive Search Ads for Beta testing in its continual push toward automation. This feature is still in open Beta and we’ve been testing it in multiple accounts over the past few months. We’ve certainly seen some mixed results so far, so as with any new feature you should test with caution.

What Are Responsive Search Ads?

With Responsive Search Ads, you provide up to 15 headlines and 4 descriptions for Google to choose from within an ad group. Over time, Google will test different combinations to learn what performs best for different queries. The big advantages Google touts with this ad type are the ability to take up more real estate, as these ads can show up to three headlines and up to two 90-character descriptions, as well as more ad flexibility.

Test Account 1: Strategy & Results

In our first account, we tested this new ad type in only one higher-traffic campaign and let it run alongside a few other longstanding, top performing expanded text ads. We wanted to give the Responsive ad a chance to run and gather data but we also couldn’t rock the boat too much. We decided to max out and put in 15 headline options and 4 descriptions and did our best to follow existing best practices. Our ad strength was given Excellent status:


We checked in on performance a couple times a week and waited until we reached the following statistically significant ITC percent difference:

For this campaign, the Responsive ad underperformed across the board, with a lower CTR, conversion rate and conversion/impression as well as a higher CPA.

You’ll notice a pretty big variance here in impressions by ad type. Unfortunately, the Responsive ads were put into Approved Limited status. However, since we had reached significance for ITC and CPA was creeping up too high we had to move forward with shutting down the Responsive ad before we could move forward with trying to get the Responsive ads out of limited status. If this account had allowed for greater performance fluctuations we would have likely worked to resolve the limited issue and let the Responsive ads run longer.

Test Account 2: Strategy & Results

Our strategy in the second account was similar to the first in that we let the Responsive ad variant run alongside some legacy ETAs. However, for this test, we used fewer inputs, with only 4 headlines and 4 descriptions. Our ad received an Average ad strength status, with Google recommending more headlines as well as making them more unique:

However, we chose our headlines based on those that had historically performed well so we stayed the course with our selection.

Although traffic was somewhat low to call the test when we did, the CPA was significantly higher with the Responsive ads than with our ETAs and conversion rate was abysmal so we had to shut off the Responsive ads.

Test Account 3: Strategy & Results

In our third test account, we also ran Responsive ads alongside ETAs. We tested the ads in both Brand and Non-Brand campaigns but our data below is just reflective of Non-Brand.

In this test, we also used fewer fields like our second test account, with 5 Headlines and 3 Descriptions. We used a similar strategy as our first two tests and used messaging in the Responsive ads that had worked well in the account. No ad strength indicator was available for this account since it is AU based. Again we see at a high-level ETAs outperforming Responsive ads:

A different data analysis approach we took here was viewing data by ad type. Overall, Responsive ads still underperformed however they did perform better than one of the ETA variants at an aggregate level:


That said, it could be helpful to view your data this way when analyzing performance to see if Responsive performance is worse than all ETA variants.

Key Takeaways

So far, I’m not overly impressed with Responsive Search ads yet. In all three test accounts, Conversion Rates and Conv/Impr were lower and CPLs were higher for Responsive ads, and in two out of three CTRs were lower. That said, we have pretty strict CPL goals we had to stay within for all three accounts which didn’t lend us to as long of test periods as we might have wanted. I’d like to think that with some more time the Responsive ad performance could have improved, but performance was poor enough that we couldn’t allow them the time they might have needed to get better.

Additionally, based on the performance of test account 2, I think I could say that there is some validity to the ad strength indicator tool; maybe if we had tried a few more headlines and more ‘unique’ variants we could have gotten results that weren’t so much worse than the ETAs. While I do appreciate this tool, it could use some improvements.

Data is especially lacking from a reporting standpoint since all you can see for the performance of your Responsive ad combinations are Impressions through the ‘view asset details’ feature from the Ads tab:

Unfortunately, Impressions by combination isn’t very helpful in giving us insight into what Responsive combinations are working best.

Our hope is that Google’s Responsive ad learnings will improve over time so we could potentially test this format again, and we’d also love to see Responsive ad performance by combination though we know that’s probably wishful thinking.

What about you? Have you tried Responsive Search ads yet? What results have you seen? Let us know in the comments section!