JioHotstar details real-time advertising decisioning architecture

JioHotstar has released an engineering overview detailing its complex advertising request workflow designed to manage real-time ad decisioning for its streaming platform. The architecture utilizes a cascaded hierarchical approach combined with rhythm control algorithms, such as PID and SHALE, to select a small group of ads from thousands of candidates within a 100-millisecond window. This system is built to maintain low latency and service reliability even during high-concurrency events like major sporting matches.

The workflow involves multiple backend components that evaluate content metadata, user context, and device information to generate personalized responses. Engineering leaders Pawan Nagar and Ayush Kumar noted that the primary challenges involve coordinating distributed services for inventory, decisioning, and tracking while ensuring that individual component failures do not interrupt video playback. The platform supports various formats, including pre-roll and mid-roll video ads, and adheres to industry standards like OpenRTB while implementing proprietary logic for content-aware targeting.

Separately, users can manage their Hotstar subscriptions through the Paytm platform. Hotstar, owned by Disney, offers Premium and VIP plans to its large subscriber base. Subscriptions can be recharged via the Paytm app or website using various payment methods, including UPI, wallets, and credit or debit cards.