The Real Story Behind Your Ad Performance
There are multiple features that make up paid media ads regardless of the channel they are run on. From images to videos to headlines to copy text - if an ad is underperforming how do we determine why? Quantifying ad performance is difficult because the platforms have uneven distribution and are subject to noise.
From this above chart, we have 4 ads live. From this, we would naturally interpret that UGC + No Coupon is our strongest performer. However, this is not the case.
When we segment it further, we more closely understand that the headlines actually return the same return. The deciding factor for these ads is UGC. not the coupon. By focusing on the ad as a whole, we are neglecting to break down our data to understand the more crucial features of our ad. We need to break these components down, to understand which feature has the strongest impact on ad performance.
Here at Blackbird, we utilize a tool that shows what variables of an ad are associated with better performance and most importantly, what variables are most critical to optimize into. In order to do so, we compare ad instances and features within a cross-correlation analysis (ran within R) that compares all features of an ad (against all instances of these features) and maps them based on an efficiency scale of 0 to 1.
We can group ad features by copy themes, headline themes, image/video themes, and many more. The more features, the more accurately you are able to understand what is actually impacting your channel's performance. Across each ad, we give them an efficiency ranking. The number 1 indicates “efficient”, at or under desired CPA (set by advertiser) while “0” indicated inefficient, above the desired CPA. We return the following chart.
Features on the top have the most efficiency variance. As you move down the funnel, the features closer to the bottom are the least likely to be correlated to the efficiency or inefficiency of your ads. In summary, features at the top we want to focus the majority of our attention into when creating and testing new ads.
In this instance, we can see that copy theme is the most important variable and New Year’s/Back in Stock are the best performers.
Now it is important to remember, this analysis does not take spend or impressions into account, so we want to ensure that there are multiple instances of each feature and theme present within the ads you are testing. The more data on themes, the better! With this information, you can find surprising results and direct your optimization process more effectively.
Takeaways
Break out ad qualities to understand what matters
Learn what qualities of creative are bringing better performance
Focus testing on variables that have higher variance
