How DoorDash Leverages Flink to Detect User Sessions and Engage DoorDash Consumers with Real-Time Notifications
Original post Leveraging Flink to Detect User Sessions and Engage DoorDash Consumers with Real-Time Notifications
Article Summary
Chen Yang and Fan Zhang from DoorDash reveal how they turned annoying cart abandonment notifications into a conversion machine. The secret? Knowing when users actually leave, not just guessing.
DoorDash was sending push notifications to users with items in their carts, but couldn't tell if someone abandoned the app or was still browsing. Their old system blindly polled backend services at fixed intervals. The team rebuilt the entire workflow using Apache Flink to track real-time user sessions from mobile and web events.
Key Takeaways
- 6x faster notification delivery by detecting actual user inactivity
- 40% higher open rates and 2.6% boost in visit rates
- Flink manages hundreds of gigabytes of state tracking hourly user sessions
- Frontend events enable smarter triggers than backend polling alone
By switching from scheduled backend polling to real-time session detection with Flink, DoorDash dramatically improved notification relevance and drove significantly more order volume.
Continue reading
This page summarizes “Leveraging Flink to Detect User Sessions and Engage DoorDash Consumers with Real-Time Notifications” by DoorDash, published Nov 7, 2023. All credit for the original work goes to the authors.
