The platform team said no — dbt won with “yes” - story of governance, autonomy, and scale
At Delivery Hero, dbt adoption came from customer teams pushing past platform caution, not from a platform mandate.
Overview
At Delivery Hero (70+ countries, 50+ teams, 60+ PB of data, 300k+ tables), domain teams run their own pipelines while needing central governance. A live discussion between a platform-governance and a domain-team practitioner recreates the real tensions of adopting dbt, covering accessibility, migration, lineage, and testing.
What you'll take away
- 01
dbt adoption at Delivery Hero was pushed up from customer teams, not down from platform leadership: teams built their own tooling (dbtflow) to bridge dbt manifests into Airflow before the platform team acted.
- 02
At massive scale, about 300,000 tables, 60 petabytes, and 250 dbt projects across 45 decentralized teams, requiring an owner tag on every model, enforced in CI, fixed the "who do I contact when this breaks" problem.
- 03
The team's guiding lesson: embrace tooling change deliberately ("evolving with intent") rather than resisting it or chasing every new stack, and treat a platform's golden path as a living standard, not a fixed one.
Session speaker
M. Waqas Shahid
Delivery Hero