Learn
What is dbt?
A tool that lets your data team write, test, and document data transformations using SQL – turning raw data into business-ready tables with version control, like code reviews for your numbers.
What it is
dbt (data build tool) is software that manages the "T" in ETL – the transformation step where raw data gets cleaned, combined, and shaped into useful business metrics. Instead of analysts writing one-off SQL queries or building calculations in spreadsheets, dbt lets them define transformations as reusable, testable, documented code. Think of it like the difference between a chef who cooks from memory (different every time, no recipe to hand off) and a kitchen that runs on documented recipes with quality checks. If the chef leaves, the recipes stay. dbt works with your existing warehouse – BigQuery, Snowflake, Postgres, Redshift – and handles the logic of turning raw tables into the metrics your dashboards consume.
Why it matters for your business
Most mid-market companies have critical business logic buried in spreadsheets, stored procedures, or a single analyst's head. When that analyst goes on leave, nobody knows how the revenue report is calculated. dbt solves this by codifying every calculation: "revenue = invoice amount minus credits, excluding internal transfers, recognised on settlement date." Each rule is defined once, tested automatically, and documented. A $60M services company we worked with discovered that their three most important reports used three different definitions of "active customer." After consolidating the logic in dbt, the executive team finally agreed on their customer count – which turned out to be 18% lower than marketing had been reporting.
How we approach it
We structure dbt projects following the medallion pattern: staging models clean raw data, intermediate models handle joins and deduplication, and mart models deliver business-ready tables. Every model includes automated tests – uniqueness checks, not-null checks, accepted-value checks – so data quality issues surface during the build, not in a board meeting. We write documentation alongside every model so your team (or a future hire) can understand what each table represents and how it was derived.
Related Services
Key Takeaways
- •dbt turns your data transformations into tested, documented, version-controlled code – not tribal knowledge.
- •Every business metric is defined once in dbt and reused across all dashboards and reports – no more conflicting definitions.
- •Automated tests catch data quality issues during the build, before bad numbers reach your leadership team.
- •dbt works with your existing warehouse – it does not replace your infrastructure, it organises the logic on top of it.