Do Responsive Search Ads Improve Performance?

Microsoft Advertising recently announced the beta launch of responsive search ads. Microsoft’s RSAs allow advertisers to input up to 15 headlines and four descriptions into a single ad unit. Microsoft will then mix, match, and test the various combinations to obtain the best performance. Up to three headlines and two description lines can show at a time. Microsoft says more than 32,000 ad combinations are possible.

Google launched a similar offering last year.

Microsoft doesn’t define “best performance.” Likely it’s ads with higher click and conversion rates.

Microsoft doesn’t define “best performance.”

Microsoft’s introduction of RSAs follows the automation trend in paid search. Both Google and Microsoft have multiple automated bidding strategies that make real-time bid changes. The platforms use secret algorithms to determine which users are most likely to convert. Advertisers can make bid adjustments based on time and device, but not for:

• Operating system,
• Browser history,
• Ad creative,
• Language.

The platforms use those four variables to set bids. The same is true for RSAs, which, we are told, use machine learning for better decisions. Advertisers still need to provide the right parameters and emotions that machines don’t. (See a helpful post on Search Engine Land on the role of humans in automated pay-per-click .)

This post focuses on RSAs.

The principal benefit of RSAs is efficiency. Advertisers can input many messages, and the platform (Google or Bing) will find the winning combinations. Instead of manually testing, say, five ads per ad group, an RSA does the testing for you. The drawbacks are a loss of control and transparency.

For example, Google’s RSAs show only the number of impressions that each headline, description, and combination has received. Thus we can see the number of impressions, but not the number of clicks and conversions. We have to take Google’s word that it shows the best performing combinations. If we could see all of the data for each combination, we could then create new text ads using those headlines and descriptions.

Presumably Microsoft will report RSA performance similarly.

To be sure, we can view RSA performance as a whole. For example, we can see how the ad unit compares to other ad types, such as expanded text ads and dynamic search ads. Seeing the overall performance of the RSA is valuable. But knowing which combinations drive the highest conversion rates would be helpful, too.

Testing RSAs

For most of 2019 I have tested Google RSAs against regular text ads for three clients, as follows.

• RSAs are in every ad group of every campaign.
• RSAs contain four to 15 headlines and three to four description lines.
• Each account is a different vertical.
• Both Google and Bing allow the pinning of headlines and description lines. For example, the call to action can always show in headline 2. I did not pin any headlines or description lines.

The first testing metric is the click-through rate. My initial hypothesis was that the CTR would be higher for RSAs since Google claims to push better performing ads.

Instead, the RSA click-through rate was slightly lower for two of the three clients. Client number two, however, experienced a much higher CTR (6.64 percent) with RSAs.

Next is the average cost-per-click. I assumed the CPC for RSAs would be lower. This assumption was correct.

CPCs for RSAs were lower for all three clients. One potential reason is the quality score, which is largely based on relevance. The better the alignment of the search query, matched keyword, ad copy, and landing page, the higher the quality score. And higher quality scores produce lower CPCs as advertisers are rewarded for better continuity.

Since Google RSAs theoretically show the most relevant ad combination, it makes sense that the quality scores are higher.

Likewise, the cost per conversion for each client was lower with RSAs.

RSAs Only?

The three advertisers in my Google Ads test experienced better results with RSAs — notwithstanding the lack of transparency. It’s worth testing Microsoft’s new offering, too.

A good way to start with RSAs is to analyze your existing headlines and descriptions. Create your RSA from the headlines and descriptions with the highest click-rough rates, lowest cost per conversions, and highest conversion rates. Try at least eight headlines and three descriptions — the more of each the better.

I wouldn’t be surprised if Google allows only RSAs in a year or two. Microsoft would likely follow. Be prepared.