dbt – Instantly Useful, Yet Hard to Explain!?

Throughout my career, one skill has consistently proven invaluable: making technically complex topics understandable for non-technical decision-makers. But there’s one topic that I find harder to explain than usual: the value of dbt (data build tool).

Why is dbt hard to explain?

The challenge lies in the fact that dbt solves problems that are invisible to business stakeholders. Here are a few examples:

• The nervous tension of deploying a critical change to a central data model on a Friday afternoon.
• Reverse-engineering code and DAGs created by other engineers to understand lineage and dependencies.
• Outdated documentation in Confluence or GitHub that can never keep up with the pace of changes.

And perhaps the most significant point: before dbt, we were forced to build support systems like automated testing and continuous integration ourselves—or rely on a patchwork of tools. These basics have long been standard in software development but were painfully absent in data engineering.

Do Stakeholders Benefit from dbt?

Absolutely! While dbt’s value may not be immediately visible to stakeholders, its benefits are far-reaching:

• Automated Testing
  It doesn’t just prevent bugs but also gives us—and our stakeholders—confidence that critical models are functioning reliably.

• Lineage Graphs and Orchestration
  These save time when troubleshooting or implementing changes to existing models. They also make it easier for analysts to use existing models, reducing the Time to Insight.

• Documentation 2.0
  dbt’s YAML files establish a standard that tools like ChatGPT can recognize and replicate. Need to generate documentation for a staging model in seconds using a simple prompt and a link to the API reference? No problem anymore.

The Message for Stakeholders

The ultimate takeaway for our business stakeholders should be clear:
Excellence requires excellent tools.

With dbt, we’re not introducing unnecessary overhead. Instead, we’re adopting a framework that makes our data products more robust and our work more efficient.