Schibsted Reports Subscription sales Lift Following AI-Powered Personalization Rollout
OSLO, norway – schibsted, a leading Nordic media group, has seen a positive impact on subscription sales following the implementation of a new AI-powered personalization model, the company revealed today. The model, built on a foundation of rigorous data analysis and machine learning, identifies key user preferences to deliver more relevant content recommendations.
The project, initially undertaken by a small team, involved sifting through 158 potential data points to identify just a dozen truly impactful features for predicting user behavior. While model advancement proved relatively straightforward, scaling the necessary infrastructure presented a notable challenge, according to Schibsted’s team. The success underscores a growing trend in the media industry: leveraging artificial intelligence not just for content creation, but for driving revenue through enhanced user engagement and targeted subscription offers.
Schibsted’s approach combines static rankers with machine learning models, utilizing a custom feature store to understand individual user tastes. “We really dug into the user data we could get our hands on, and used a mathematical approach to determine what really mattered – what features are useful, and which ones are just noise,” explained Schmitz, a member of the development team.
The company is now expanding the solution across multiple newsrooms and transitioning to Tecton, a managed feature store, to streamline onboarding for its various brands. Schibsted is utilizing orchestration tools like Flyte and an in-memory database alongside AWS DJL (amazon Web Services Deep Java Library) for efficient model inferencing.”By unifying feature availability and expediting onboarding for various brands, we’re confident this change will further enhance user experiences across our newsrooms,” Schmitz said. When reflecting on the project,Schmitz stated,”Set more developers on it” would have been his primary adjustment.
