How Well-Designed Schemas Jump-Start Analytics
In some recent surveys and interviews, several of our customers shared their thoughts on the subject.
Key TakeawaysThe key takeaways from customer feedback about Fivetran schema design include:
- Good schemas mean analysts don’t need to worry about the underlying structure of the data
- A data integration solution should feature comprehensive, quality schemas
- Clean schemas provide a good starting point for derived tables and other transformations by analysts
We noticed that the data quality and structure for some of the connectors (JIRA & Hubspot) were superior with Fivetran versus Stitch. Since we are using Looker and only rely on its PDT (persistent derived tables) feature for transformations, starting with clean schemas is extremely useful.
I love the consistency of Fivetran managed schemas. One of our most impactful projects on my BI team was to automate our month-end close process for finance. This required significant engineering and manipulation of data coming from Salesforce. Because we had faith in the consistency of the SFDC schema, we were able to build numerous automated processes on top of Salesforce without having to worry about the underlying structure of the data. We could instead spend our time on the actual analytics of the project. The same for data coming from the JIRA and Greenhouse connectors. We could essentially "set it and forget it," focusing our efforts on analytics instead of engineering and managing a schema.
Another great thing about the consistency of Fivetran schemas is that at my new company, I don't have to re-learn underlying data structures. The SFDC schemas at New Relic look very similar to the SFDC schemas at ZoomInfo. Accordingly, I can be productive and have meaningful conversations about data flow, data pipelines and data engineering needs from day one instead of on day 30 (or however long it takes to learn new data flows outside of Fivetran). My time to insight and adoption curves scale rapidly.
I evaluated and tried Stitch before choosing Fivetran and found specific schemas incomplete and not as well documented. The documentation was poor, with no changelog. Being able to see how a schema evolves and the changes applied over time to a connector was a deciding factor.
The documented data models are an incredible asset. We love them as we feel like we're adding more value to our data consulting practice when we actually extract insights and build data products. We like to treat data ingestion as a commodity.
Designing a good schema can require a great deal of work and investigative chops. To experience for yourself how Fivetran schemas can help you, sign up for a personalized demo with one of our product experts, or start your free trial today.
The schema wireframes in the documentation were fantastic and implementing Fivetran was very straightforward. Once we turned it on we were amazed by how quickly our data was in our warehouse.