Back to feed
News Story
APriority78
InfoQ
1 sources

JioHotstar Publishes Engineering Overview of Ad Decisioning Workflow

JioHotstar has published an engineering overview of its ad request workflow, detailing how the streaming platform coordinates distributed services to select, deliver, and measure personalized ads during video playback. The architecture is designed to meet strict latency requirements while enabling real-time ad decisioning, supporting large-scale streaming traffic, and maintaining playback reliability.

SynthePulse Insight · AI deep readingMembers

A Thousand Ads Compete for One 30-Second Slot: How JioHotstar Makes Real-Time Decisions

Version 1 · 1 source

When a viewer hits an ad opportunity during video playback, how does JioHotstar's ad decisioning workflow select a handful of ads from thousands of candidates to form a 30-second ad pod within 100 milliseconds? Based on JioHotstar's published engineering overview, this article breaks down its waterfall-tiered approach, pacing control algorithms, and backend challenges.

  • JioHotstar's ad decisioning workflow completes ad selection and response generation within 100 milliseconds, even during high-concurrency events like critical moments in major sports matches.
  • The platform uses a waterfall-tiered approach combined with pacing control algorithms such as PID and SHALE to select a small number of ads from thousands of eligible candidates to form a 30-second ad pod.
  • Ad requests include contextual information such as content metadata, user context, device information, and available ad inventory details.
Open section navigationAd Decisioning: A Multi-Stage Flow from Request to Response

Ad Decisioning: A Multi-Stage Flow from Request to Response

When a viewer reaches an ad opportunity during content playback, an ad request begins. The request includes contextual information needed for ad selection, including content metadata, user context, device information, and available ad inventory details. The platform then processes the request through multiple backend components that evaluate eligible ads, apply targeting rules, and generate the final response returned to the video player.

JioHotstar explains that the platform uses a waterfall-tiered approach combined with pacing control algorithms such as PID and SHALE to select a small number of ads from thousands of eligible candidates to form a 30-second ad pod. These computations help balance campaign delivery requirements, ad inventory allocation, and advertiser constraints, all while completing ad selection and response generation within 100 milliseconds, even during high-concurrency events like critical moments in major sports matches.

Free for now

Read the full analysis

3 more sections of analysis, plus the full takeaway

Loading

Credibility boundary

The information in this article is primarily sourced from an engineering overview published by JioHotstar, as reported and translated by InfoQ. All specific figures and architectural descriptions are derived from that engineering overview and represent source claims that have not been independently verified.

Sources for this version

  1. 千条广告"抢"一个 30 秒位置，平台如何实时做出选择？InfoQ

Primary report

InfoQ

Primary source