dbt as Neutral System of Record
Julia Schottenstein, Product Manager at dbt Labs, on the business model of open source
dbt’s real moat is not that it runs transformations, but that it becomes the neutral place where a company stores the rules for how its business data should be shaped and measured. That matters because large companies often use more than one warehouse, more than one BI tool, and separate teams for ingestion and transformation. Putting that logic in dbt avoids rewriting the same definitions inside Snowflake, Databricks, Looker, or Fivetran every time the stack changes.
-
dbt Core is an open source compiler and framework, while dbt Cloud sells the workflow around it, including IDE, CI checks, scheduling, hosted docs, and governance. That makes the paid product less like a simple runner and more like GitLab around Git, built for analytics engineers working in production.
-
This fits the way data teams are organized. Fivetran owns the pipes that move raw data into the warehouse. dbt sits closer to the analyst or analytics engineer who knows what customer, revenue, or churn should actually mean in SQL tables and metrics. The tool boundary mirrors the team boundary.
-
The same neutrality extends upward into metrics and semantic layers. If a company defines CAC or lifetime value inside a BI tool like Looker, that definition stays trapped there. dbt’s push is to define it once near the transformation layer, then reuse it across dashboards, notebooks, catalogs, and apps.
The direction of travel is toward dbt becoming connective tissue across a more consolidated but still multi-vendor data stack. Warehouses and ELT tools will keep bundling adjacent features, but the more enterprises run mixed Snowflake, Databricks, and BI environments, the more valuable a cross-cloud, cross-tool system of record for business logic becomes.