Skip to main content

Dimensions

A dimension is a categorical attribute in Sundial's semantic layer that maps to a column in a table. Dimensions are used to segment, filter, and group metrics to understand patterns across different categories.

Definition​

A dimension consists of:

  • Name: A unique identifier for the dimension (e.g., country, platform, device_type)
  • Source Column: The specific column in a table/view containing the dimension values
  • Source Table: The table/view containing the dimension column, which is also mapped to metrics
  • Values: The set of possible categorical values for this dimension
  • Description: Documentation explaining the dimension's meaning and intended use

Examples​

Common dimension columns that can be mapped to metrics include:

  • Geographic: country_code, region_name, city_name
  • Technical: platform_type, device_name, browser_version
  • User: user_type, subscription_tier
  • Business: product_category, marketing_source

Usage in Analysis​

As part of Sundial's semantic layer, dimensions enable you to:

  • Group metrics by categories (e.g., analyze revenue by country)
  • Filter metrics for specific segments (e.g., active users where platform = 'iOS')
  • Compare metrics across segments (e.g., conversion rate by device_type)
  • Track changes in dimensional attributes over time (using changing entity dimensions)

Best Practices​

When defining dimensions in your semantic model:

  • Use clear, descriptive names that reflect the column name
  • Document the meaning and valid values
  • Ensure column values are standardized and clean
  • Choose the appropriate dimension type based on whether the column values change over time
  • Maintain consistency in dimension definitions across tables and metrics
  • Map dimensions to actual columns in your tables that are also used by metrics