by Dan Wachowski –
Responsive ads are relatively new in Google Display, so the Lever Interactive media team developed a study to test the effectiveness of responsive ads in comparison to traditional image ads. Since not every ad type performs the same across all verticals, two Lever Interactive clients from separate industry verticals were chosen for this study.
Before implementing, the Lever Interactive media team set the stage for testing. First, the team reviewed their clients’ objectives to make sure what they were about to test was in line with their goals, not only for a specific channel, but overall. After thinking about the goals and objectives, the media team was able to identify that the objective was to interact with familiar users of the brand.
With the objective in place, the next step was to develop a hypothesis. Since responsive ads can show in many different formats, the media team hypothesized that responsive ads would lead to increased traffic and better click-through rates for both verticals.
With the objective and hypothesis set, the final task was to create a controlled environment within the Google Display Network through targeting. Since the object is to target familiar users, the media team used remarketing targeting to show these ads to users that had visited the website previously but did not convert.
Keeping everything in mind from the review and brainstorming phase, the media team worked on implementing the testing plan by setting up campaigns, setting budgets, targeting and ads. The media team already had remarketing lists populated, as they were utilizing remarketing lists in search. When launching the image ads, the media team ensured they had image sizes for desktop and mobile, as it is important to have a strong follow-up message on all on devices with remarketing. The different image ad sizes also allowed for similar placements to be in market alongside the responsive ads. The teams used only one responsive ad per ad group. When testing ads, it is important to be mindful of the ad rotation setup as it dictates what kind of results a test could generate. The media team chose to use the setting “prefer the best performing ad rotation”, as Google uses their algorithm and machine learning to serve the most qualified ad to a particular user. With everything in place, data collection could begin.
The media team observed the test for four months before deeming the data significant to test and analyze our hypothesis. There is no time limit for such a test, as the duration is dictated on budgets, list sizes and seasonality. The data collection for the Lever Interactive test can be found below.
From the data collected, the media team was able to determine that their hypothesis was correct for the non-retail vertical. The chart above shows more traffic and a better click through rate for responsive in the non-retail vertical than image ads.
In reviewing the retail vertical results, the media team concluded that the responsive ads did drive more traffic, but did not lead to a better click through rate.
Even though the responsive ads performance differed by vertical from the hypothesis, responsive ads did lead to a higher conversion rate and a lower cost/conv. than image ads. This indicates that people who clicked on the responsive ads were more likely to convert, classifying this ad type as more cost effective for driving conversions.
Overall, it is important to note that performance may be impacted by the targeting selected, budget, ad rotation, type ads and the client’s vertical. Since there are many different levers that can be pulled to impact performance, the results generated above might not be true for every client.
The best way to find out if responsive ads are the best ad type for you is by testing. With that being said, our take is that responsive ads allow advertisers to get more valuable, efficient traffic in the Display Network through the many different sizes, especially if the campaign objective is to drive brand awareness.