Event Analytics - from unstructured to insightful
summary
Builds a subscription state machine that reconciles messy multi-source events by inserting implicit events (e.g., add a missing start or resume timestamp “a second before” a contradictory event) and reconstructs exact customer lifecycles. Implements a testable, technology-agnostic pipeline using python + SQL (PoC on Postgres then moved to Postgres → Redshift seamlessly). Outcome is accurate lifetime revenue, cash/bookings alignment, product bundling insights and a successful B2C→B2B transition.
AI-written summary · about
chapters (9)
- 0:00Event analytics journey
- 1:08Core solution principles
- 3:18Subscription data anomaly
- 7:35Rebuilding data model
- 9:21Subscription states and transitions
- 12:52Inconsistent event sequences
- 15:40Subscriptions to business insights
- 17:46Stakeholder value from data
- 20:53Q&A
related talks
18mBehavioral Data Creation for AI - Delivering Value and Insights
33mAccidental DataOps: How we matured our organization with data, and how you can too!
23mDelivering Retail Analytics & Data Projects With Data Created For Purpose
49mTransforming Analytical Data with Event-Driven Architecture
44mNear-Real-Time Event Analytics in a High-Load Project
38mPostmortem: Evolution of an Event recording solution
29mWide Event Analytics
27mActivity schema data modeling using a single table