Beyond Send-Time Optimization
summary
Claims an AI decisioning layer outperforms classic send‑time optimization by choosing send/defer/skip actions rather than just the statistically “best” minute to send. Demonstrates five scenarios, a 3-layer architecture (Predict, Decide, Protect), state management that persists fatigue across campaigns, and runtime constraints: millions of daily decisions at sub‑100 ms latency. Demo results show roughly 60% send 40% defer, plus GitHub code and details for reproduction.
AI-written summary · about
chapters (15)
- 0:00Welcome & What We’ll Cover: The Evolution of Send-Time Optimization
- 0:24Why “Best Time to Send” Isn’t Enough in a Noisy Digital World
- 3:27Engagement Is Flatlining: The Consequences of Inbox Overload
- 5:05Rewind: How Classic STO Works (and Why It Used to Win)
- 7:45High Probability ≠ High Wisdom: Where Prediction Breaks Down
- 10:24The Big Shift: From “Best Time” to “Best Action” with AI Decisioning
- 12:30What Signals Power Decisioning? Journey, Fatigue, Rules & Human Choice
- 13:32Side-by-Side Example: Prediction Sends, Decisioning Holds
- 15:08Business Impact: Trust, Lower Churn, and Relationship Optimization
- 16:42Hand-off to Engineering: From Marketing Concept to Technical Reality
- 17:30Engineer Intro + Core Problem: Predictions Don’t Guarantee Right Actions
- 18:32Live Demo: Five Scenarios of Send vs Defer vs Skip (Channel-Agnostic)
- 22:29Under the Hood: 3-Layer Architecture (Predict, Decide, Protect)
- 23:52Production Challenges: State Management Across Competing Campaigns
- 24:11Wrap-Up, Resources (GitHub), and Closing Thanks
related talks
32mFrom Three Days to One Minute: Architecture Trade-offs in Production Enterprise AI
08mThe Plan That Learns
26mAI-Powered Predictive Analytics: Optimize Performance
19mReal-Time AI: Low Latency Solutions for Interactive Applications
40mNext Best Action: Maximizing Customer Value
22mMicrodosing the Economy
35mShipping with Confidence in the Age of AI
25mTranslating AI Capabilities Into Better Business Results