Magnite unveils Machine Learning tool to help publishers

Magnite, the world’s largest independent sell-side advertising company, today introduced machine learning driven recommendations for A/B testing within its Demand Manager product. Built on Prebid technology, Demand Manager empowers leading publishers with the tools, insights, and connections to grow revenue across the ever-changing marketplace of ad exchanges, formats, and vendors. This launch marks the introduction of sophisticated machine learning algorithms packaged with existing industry-leading capabilities for publishers to manage their Prebid stack.

The new feature utilizes machine learning to provide automated Prebid optimization recommendations based on Prebid and ad server auction data and session data, with the goal of revenue lift. Publishers can now activate machine generated settings into an A/B test with a single click. Initial tests showed that 80% of wrappers that ran a machine-generated experiment saw an increase in revenue compared to the existing setting.

“Publishers are confronted with a seemingly endless number of daily choices that materially impact revenue. We developed this new feature to eliminate Prebid optimization guesswork while still giving publishers full control,” said Matt Tengler, VP of Product at Magnite. “Infusing A/B testing with machine learning makes it easy for publishers to measure revenue and page performance improvements. This continues to bring publishers innovative tools that focus on revenue and efficiency at the same time.”

“We were excited to test this new Demand Manager feature to see how machine learning could improve our wrapper configurations,” said Ben Elshaw, Director of Operations at LADbible. “We were pleased to see a material rCPM increase following the implementation of the optimization recommendations. Demand Manager’s A/B testing functionality combined with machine learning recommendations is a welcome innovation that we hope to see expand in the future.”

“After using Demand Manager’s new machine learning-driven recommendations for wrapper optimization, we were pleased to see an immediate increase in revenue,” said Lewis Lee, Senior Ad Tech Specialist at REA. “The A/B testing capabilities allow us to customize our wrapper configurations based on data and quickly test more scenarios while minimizing risk. We look forward to testing additional features that utilize machine learning to help us improve our Prebid settings.”