How Warner Bros. Discovery uses Amazon Personalize
summary
Built a one-line deployable data layer called Prism SDK across 200+ properties to capture billions of interactions and feed 30M+ qualified events per week into Amazon Personalize via Kinesis Firehose and Lambda, enabling sub‑five‑second model updates. Promo click rates rise from ~0.5% to ~1.8–1.9% (≈4x), A/B holdouts show 11–14% lifts in cross‑brand visits, page loads and engagement time, and a working prototype with trained model and autoscaled inference launches in days.
AI-written summary · about
chapters (9)
- 0:00Context and objectives
- 1:19Building the Prism platform
- 3:42Architecture and data flow
- 5:57Discovery of personalization
- 9:14Choosing Amazon Personalize
- 11:13Data schema and implementation
- 14:02Demo and use cases
- 15:24Measuring performance impact
- 18:25Q&A and closing
related talks
26mPersonalisation at scale with AI/ML and Amazon Personalize
25mDeliver viewing experiences for super fans with Amazon Personalize
29mPioneering user experiences at StockX with Amazon Personalize
27mTransform online shopping experiences with personalized recommendations
24mAI-Powered Transformation in Fashion E-Commerce
11mMaximizing personalization and AI models with quality customer data
16mAWS On Air Voice of the Customer ft. Amazon Ads
29mUse sports data to elevate fan engagement and performance