Social media ads designed by A.I.

We worked with a global cosmetics brand to improve CPA on a mobile app that was being promoted via Facebook ads.


The client was spending a lot of money to drive downloads of a mobile app, and was looking to reduce Cost Per Acquisition.


The first step involved downloading and analyzing images and performance data from previous campaigns using A.I. clustering algorithms. We then scanned their library of image assets and used computer vision to identify images that were most likely to have a high click rate, based on our system’s findings from the Facebook analysis.

We then trained our A.I. engine to create designs conforming to the client's brand guidelines, iterating through the generated outputs until we established four main design approaches. We then used this trained model in combination with the A.I.-selected image assets to produce over 2000 Facebook ad creatives.

Finally, the creatives were published to iterative split test campaigns, the results were analyzed and new designs automatically generated to improve performance until the highest-performing creative was surfaced.


  • Over 93% reduction in Cost Per Acquisition





Global Cosmetics Brand


93% improvement in CPA

Interested in implementing something similar?

Contact us