From prototype to production: how DoorDash built a scalable analytics SDLC with dbt and ThoughtSpot
DoorDash used a dbt-defined semantic layer to take a new analytics experience from pilot to thousands of external merchant users in a matter of weeks.
Overview
A fireside chat on how DoorDash modernized its analytics infrastructure, consolidating multiple teams onto dbt and ThoughtSpot for a system serving more than 10,000 users.
What you'll take away
- 01
DoorDash splits its semantic layer from its consumption layer: dbt defines metrics once on top of Iceberg tables and Snowflake, and ThoughtSpot connects directly to those definitions rather than keeping its own parallel set.
- 02
Externally, DoorDash embeds a fully white-labeled consumption layer directly into merchant-facing tools, taking a new use case to "thousands of users within a few weeks."
- 03
DoorDash's advice: semantic modeling is hard but worth it, treat the semantic layer as code, and make the governed path at least as easy as writing an ad hoc query, or people will route around it.
Session speakers
Harsha Reddy
DoorDashSonny Rivera
ThoughtSpot