Data modeling is essential but time-consuming. Data engineers and analysts can spend a lot of time transforming raw data into analytics-ready data sets. With Fivetran, the raw data is normalized to make modeling easier, but further transformations are still needed to cleanse and prepare data and to organize it into reports and dashboards. For centralized data teams that prepare all data across the organization, data transformations can take a long time as they familiarize themselves with sources they have little experience with.
To help reduce time to insight, Fivetran offers a number of pre-built data models that can be used by any dbt user. Our data models are open-source dbt projects that contain SQL logic written to work out of the box with data you have ingested through Fivetran. Fivetran develops and maintains data models, which transform data from single-schema and multi-schema data sources.
Jacob Mulligan, Head of Analytics at Firefly Health, had this to say about leveraging the packages:
The Fivetran [data models] have been invaluable. We were able to prioritize Jira with the turnkey approach, which we wouldn’t have been able to do otherwise.
Fivetran offers two types of data models: (1) source packages expose, document and standardize the underlying Fivetran schemas being created in the destination, and (2) transform packages produce analytics-ready models for each source.
Our packages produce clear, comprehensible tables to power your reports, visualizations and dashboards. If you see data models for your Fivetran connector, you can use the packaged SQL in your own project.
How to use Fivetran data models to accelerate analytics
Here is how our customers have been leveraging Fivetran pre-built data models to accelerate analytics.
Complete out-of-the-box reporting
For certain types of reporting, such as financial reporting, where there is essentially only one way to build out a balance sheet, Fivetran data models for financial data can be used on their own. That means with a couple of clicks you can set up the models to run. This greatly reduces the amount of time it takes to manually code transformations for common reporting requirements.
James Ayoub, who works in data operations at Dandelion Energy, saved weeks of time by using a Fivetran data models for finance data:
The Quickbooks package has easily saved me weeks of development time. I can now devote more time to actually analyzing the data rather than worrying about whether or not it's modeled correctly or whether it's been properly tested.
Use with downstream transformations
For other types of reporting, Fivetran data models address some of the requirements, but you may need some additional complex or unique business logic. In these instances, Fivetran data models can be used in conjunction with custom models. dbt users can orchestrate transformations by setting up dependencies between models and scheduling transformations to run in the appropriate order. This is made easy with Fivetran’s integration for dbt Core. Schedule your dbt Core models to run automatically, following the completion of upstream data loads by Fivetran.
A templated guide
Some of our users love being able to hand-code their data transformations from scratch or use highly complex transformations. For these users, they dive into the Fivetran data models, inspect the packaged SQL and use the code as a guide to build their own custom jobs.
The best thing about the data models is that they are open source. A number of our customers add their own code. These community contributions help us produce future iterations of packages to increase their value for all users.
Try Fivetran data models
Fivetran data models accelerate transformations and leverage data modeling principles and code modularity to effectively transform data for exploratory, predictive or prescriptive analytics.
Are you already using Fivetran data models? Provide your feedback and request a package.