A dbt "logic mesh" with packages: standard model and metrics across 25 autonomous organizations
Greenpeace manages 25 autonomous organizations with different tech stacks and no central mandate. Building a unified global metric language without creating a maintenance burden required a new architectural approach.
The solution: a three-layer dbt package architecture: source packages for governed inputs, a local logic layer for office-specific transformations, and a standard data model for shared metrics. With package versioning, compatibility testing, and CI-driven automated updates, a 12-person team now maintains 8+ packages across 25 projects.
The results are concrete: a new office onboarded every two months, a full data warehouse built in two weeks, and reporting delays reduced from months to daily updates. The data team went from bottleneck to innovation enabler. You'll leave with a reference architecture and a rollout checklist for scaling shared logic across decentralized environments.
Check out more sessions
- Keynote
Keynote: Level Up
Tristan Handy / Fivetran + dbt LabsElias DeFaria / dbt LabsView session - Breakout session
From 14-hour batches and poor documentation to AI-ready data: SafetyCulture's dbt rebuild
Yuna(Yunnan) Tang / SafetyCultureThiago Baldim / SafetyCultureView session - Breakout session
How WHOOP bridges dbt models and Snowflake semantic views
William Tsu / WhoopMadhura Pharande / WHOOPView session
