Netflix

Netflix has 17,000+ titles across diverse genres. Without personalization, users couldn't discover relevant content, leading to churn. 80% of viewed content needed to come from recommendations.

Deployed multi-model recommendation system using collaborative filtering, content-based filtering, and contextual bandits. Personalizes not just content but artwork, trailers, and even search results.

Netflix estimates their recommendation engine saves $1B+ annually through reduced churn. 80% of viewing comes from recommendations. Personalized artwork increased click-through by 30%.