What is Data Build Tool? (DBT)
Data Build Tool (dbt) is an open-source data engineering tool developed and released by Fishtown Analytics to help analysts with the T step in the extract, load, transform (ELT).
It essentially allows anyone with experience in SQL to write queries to directly transform data against any platform that speaks SQL including RedShift, BigQuery, Snowflake, and more. For a full list of connectors, click here.
For Analysts, By Analysts
In comparison to other data engineering tools, dbt has an extremely easy entry-point and low learning curve. All that it requires is for a user to know SQL, which arguably the majority of analysts will know already.
dbt is ideal for companies who want to empower their analysts to own the entire analytics engineering workflow from transforming data using SQL to deployment and documentation, without the need for data engineers to get involved.
It is a tool that has been developed and optimised for analysts by analysts.
dbt takes a software engineering approach to the analytics workflow and seeks to make the analytics workflow for organisations more efficient and better organised.
Key takeaways from the dbt viewpoint are:
Analytics is collaborative
dbt introduces software engineering principles to the analytics workflow including version control, quality assurance, documentation and modularity.
Analytic code is an asset
Using dbt, analyst teams can access multiple environments and write analytics code designed for maintainability.
Analytics workflow required automated tools
dbt seeks to automate as much of the analytics workflow as possible through the use of automated workflows when it comes to data transformation, testing code and the deployment of that code.
You can read more about dbt’s viewpoint in more detail on their documentation site.
To Wrap Things Up
dbt is an open-source tool that is rapidly being adopted by teams of analysts in organisations of all sizes, designed to make the analytics workflow process more efficient.
While dbt is extremely good at the T step in ELT, it knows it isn’t an Extract or Load tool and so other tools are still required to supplement dbt in the overall ELT process.
Data must first be extracted and loaded into a data warehouse using other tools before dbt can actually transform the data.
This correlates to the overall industry trend of organisations moving away from do-it-all-one tools to a modern data stack composed of different components and products.
For example, an organisation could export their Google Analytics data to a data warehouse like BigQuery, transform the raw data using dbt and then visualise the transformed data using Google Data Studio.