Case Study: Large National Media Organization achieves 200 million automated nightly predictions and 50% data-team productivity gain with Dataiku

Build an Effective Content Recommender System with Automated Data Cleansing and Scoring

Large National Media Organization, a U.S. media syndicator serving 500+ stations and direct web/mobile audiences, needed a scalable recommender system that captured nuanced user behaviors (e.g., fast‑forward vs. full listen). With a very small team and tight resources, they required an automated, maintainable solution that could be quickly tweaked and scaled—so they turned to Dataiku’s collaborative data science and analytics platform.

Dataiku automated the workflow from data cleansing through model scoring, letting the team string together 25+ processing recipes, use R and Scala where needed, and test models without heavy coding. The result: the app now automatically scores 200 million predictions every evening and the team’s productivity rose by 50%, enabling the small data team to maintain the recommender at scale and take on other projects.