Best Practices to Start Building Data Models in Sigma Today
Product Marketing, Sigma
Markets, competitors, and customers have moved faster every year since the dawn of digital.
Keeping pace is no joke, and as Crossing the Chasm author Geoffrey Moore once said, “Without big data, you are blind and deaf and in the middle of a freeway.”
While the business intelligence industry has enabled companies to effectively extract meaning from their data and improve data-driven decision making, there’s still a major issue holding businesses back. The technical knowledge it takes to get these insights creates a major bottleneck, slowing down the decision making process by making it difficult for business experts to explore data and get answers on their own.
To help solve this challenge, Sigma released visual data modeling earlier this year. Data modeling defines how data is related, what it means, and how it flows together. An effective model makes data approachable and consumable and ensures people use the right information in the right context—and it requires tight-knit collaboration between data and domain experts.
Read on for a high-level summary of the data modeling process, or read our free eBook for more detailed step-by-step instructions.
Step : Lay the foundation
First things first: The BI team will need to connect Sigma to the company’s cloud data warehouse. Once all the tables in your data warehouse have been indexed in Sigma, identify those with your core business data, or the basic information that’s often combined and configured to create more detailed reports. Turn these tables into Datasets, or collections of core, related data that serve as reusable bases for more specific and detailed analysis.
Step : Set your analysts up for success
Now that you’ve identified your core data, it’s time to start setting up Links. Links create a pre-defined, join pathway between two data sources in Sigma and give users the option to explore related data in a guided, contextualized way. Help give business users context by adding any relevant column or Dataset descriptions, and flagging your Datasets as Endorsed, Warning, or Deprecated.
Step : Use what you’ve got
If you have chunks of existing SQL you’ve been using for analytics, turn them into Sigma Datasets using the SQL Runner. No SQL? No problem—skip ahead to step 4.
Step : Establish permissions
Permissions in Sigma determine which data a user or group of users has access to build on. Permissions can be assigned at the connection, database, schema, or table level. There are 3 roles in Sigma, each with different permission levels: Admin, Author, Viewer. We recommend setting permissions at the group (team) level to make it easier to add and remove people.
Step : Bring in the analysts
Now that you’ve set the stage so other Sigma users can easily find their way through the database, it’s time to bring in the business analysts to add more color and context to the data. This is where the collaboration comes in, made possible by Sigma’s no-code user interface.
Business analysts can add in column descriptions, set up contextually relevant links, and provide additional information to show how different data fits together for particular business use cases. Adding calculations to Datasets is a good way to share key metrics and KPIs, as well as set up a single source of truth for common analytical questions. Anyone can also extract columns of data from JSON or other semi-structured files.
Not only does this process make future analyses faster and easier for the analysts themselves, but Sigma is now ready for less technical line of business users to start exploring data and uncovering valuable insights!
But wait, there’s more: Introducing Lineage
One of the key benefits of visual modeling is enabling non-SQL writers to create analyses that build on others’ foundational and endorsed work, as well as create their own joins via Links. Any published changes made to a Dataset at any time are immediately reflected across all corresponding worksheets, ensuring users are always working with the same, up-to-date information.
Lineage is a new Sigma feature that shows users both upstream data sources and downstream dependents for all Datasets and tables. Lineage is another step toward building a collaborative, centralized location for your company’s data. It allows modelers to quickly gauge the impact changes to the model will have on others’ documents, as well as see which objects depend on which sources.
Learn more about data modeling in Sigma
With Sigma, data modeling doesn’t have to be a top-down project. Get your business analysts and end-users involved and start collaborating! As new use cases emerge, users can build, materialize, and tag new Datasets, deepening your organization’s knowledge and revealing fresh insights.
To learn more about modeling, read our free best practices guide.