Supercharging CTR (Click-Through Rate) with Plug-and-Play AI Capabilities
summary
Claims a runnable, fully AI‑driven CTR pipeline that uses off‑the‑shelf cloud tools to generate creatives (generative models + genetic algorithms), estimate value with regressors and ARIMA for seasonality, reduce cardinality with multimodal embeddings, and explore with contextual multi‑armed bandits. Key specifics: the prototype runs on Google Cloud, processes up to 2 million pings per second across 27 petabytes of data, and the core flow fits in 34 lines of code. A hackathon build connects end‑to‑end and reportedly works in production.
AI-written summary · about
chapters (9)
- 0:00Intro
- 2:23Online Ads
- 7:22Content Generation
- 12:43AI Models
- 14:35MultiArm Bandits
- 19:33Monitoring
- 20:26Continuous Improvement
- 22:02Offloading Heavy Work
- 24:05Questions
related talks
24mKeynote: AI Trends in Performance Marketing
51mCloud AI: How to Get Started Injecting AI Into Your Applications
30mMobile Order Click-Through Rate (CTR) Recommendation with Ray on Apache Spark at Burger King
30mGenerative AI in action: from POC to business value
33mAdTech
18mAccelerate generative AI development with integrated data platforms
29mAI and ML for Computational Advertising
31mLarge Scale CTR Prediction Lessons Learned