approach
The Xaxis Analytics team recognized that focusing on a CPO metric meant all orders would be weighed equally, thus the campaign would optimize toward order count rather than toward order value. They hypothesized that if they used Xaxis proprietary AI, Copilot, to optimize towards high value orders instead of raw order count, the retailer’s return on ad spend (ROAS) would increase.
To test their hypothesis, the Data Science team built a custom Copilot model that used the retailer’s first-party online order data to its fullest potential, dynamically optimizing toward transactions that efficiently provided the most value to the advertiser. This test was run alongside the client’s previous programmatic audience segments for performance comparison.
