Definition
A Dataflow Step is an individual transformation operation within a CRM Analytics dataflow that processes data from one or more source datasets. Each step performs a specific function such as filtering, joining, aggregating, or calculating values, and the output feeds into subsequent steps to build the final analytical dataset.
Real-World Example
At their company, a business intelligence manager at Apex Analytics leverages Dataflow Step to transform raw Salesforce data into actionable business intelligence. After setting up Dataflow Step, leadership has real-time visibility into pipeline health, team performance, and customer trends, enabling faster and more confident decision-making.
Why Dataflow Step Matters
A Dataflow Step is an individual transformation operation within a CRM Analytics (formerly Tableau CRM, formerly Einstein Analytics) dataflow. Dataflows are pipelines that read data from source datasets or Salesforce objects, transform it through a series of steps, and produce one or more output datasets that feed analytics dashboards. Each step performs a specific function: sfdcDigest (reading from Salesforce), filter, augment (joining datasets), computeExpression (calculated fields), sfdcRegister (registering the output), and many more.
Steps are chained together so the output of one step becomes the input of the next, forming a directed graph of transformations. Designing a dataflow well means breaking the desired output into discrete, single-purpose steps that are easy to debug and maintain. CRM Analytics also offers Recipes (a more visual, modern alternative to dataflows) for similar transformation use cases, and many newer projects use Recipes instead of dataflows. But dataflows remain widely used in production deployments and remain the more flexible option for complex transformations.
How Organizations Use Dataflow Step
- •MarketPulse — Built a dataflow with separate steps for filtering, joining, and aggregating their pipeline data. The discrete steps made it easy to debug when one transformation produced unexpected results.
- •Apex Analytics — Uses computeExpression steps to derive calculated fields like deal velocity and win rate, then passes the enriched data to downstream visualization steps.
- •SilverLine Corp — Migrated newer dataflows to Recipes for a more visual development experience but kept their complex existing dataflows in place because they were stable and well-documented.