Tackling BI sprawl and scaling data models with dbt: Element Biosciences' journey from 0 to 600 models in 18 months from Coalesce 2023

"We ended up getting much more consistent reporting across departments. We are able to drastically reduce the amount of duplicate data entry, and we are able to get to a single source of data for a lot of our reporting."

Matthew Hoss, senior manager of business systems at Element Biosciences, explains how Element Biosciences uses dbt to manage and organize their data. He also shares the company's journey from zero to 600 models and discusses the challenges of conforming data across different departments.

Consolidating and harmonizing data can improve business performance

At Element Biosciences, the team has been grappling with the challenge of managing and consolidating vast amounts of data, from manufacturing data to telemetry data. However, they found that simply gathering and centralizing the data was not enough.

Matthew explains, "It's become remarkably easy to get the data into a data warehouse, but there's always that missing piece... How do you get from having data to making decisions based on data?" He highlights the need for departments to align on terms and harmonize the data. "As you go through this data maturity pathway, it's about everyone who's in the red and the oranges, and getting them into the greens."

By consolidating their data and aligning on terms, Element Biosciences was able to achieve more consistent reporting across departments, reduce duplicate data entry, and have a single source of data for reporting. This harmonization of data enabled them to focus on making data-driven decisions.

Data consolidation requires executive buy-in and employee cooperation

Consolidating data not only requires technical solutions but also a significant amount of cooperation and agreement from various departments within a company. In Element Biosciences, getting everyone to agree on terms and definitions was a crucial part of their data consolidation process.

Reflecting on the process, Matthew says, "You have to have two different groups in the room, and everyone's got to be aligned. The first one is the people doing the work that are entering the data... but you also need the executive buy-in."

This process of getting everyone to agree on terms and align on data sources helped reduce confusion and disagreements, leading to more trust in the data and more data-driven decision-making.

Data-driven approach can lead to more efficient processes

Element Biosciences' data-driven approach has led to more efficient processes and informed decision-making. By focusing on specific data sets that drive the company forward, they were able to better allocate their resources and support their data models.

Matthew explains, "A lot of times we'll pull them into the meetings, and if we can get that, then that becomes one of our spokes of support, and the more of those we can identify and define in the different departments…I think that is what we're trying to build for our data strategy through the departments."

By focusing on specific tables and data sets, Element Biosciences was able to ensure that their data was reliable and trustworthy, allowing the company to make more data-driven decisions and improve their overall business performance.

Insights surfaced

• The company went from having no data models to 600 models in a year and a half
• The data stage involved setting up tools, loading data, and preliminary reporting. However, the company wasn't data-driven due to lack of unifying data
• The people stage involved getting departments to agree on what terms mean, which is crucial for data governance and analytics
• Conforming dimensions across departments led to more consistent reporting, reduced duplicate data entry, and a single source of data for reporting
• The company faced challenges with different departments having their own data models and reports, leading to discrepancies. Consolidating data sources and aligning reports across departments solved these issues
• The company aims to have an AI engine that can look at the schema and write some of the SQL for them in the future