Skip to main content

Analyzing your Data

Sundial allows a flexible canvas called an 'Analysis' to explore your data both governed and ad-hoc. Analyses are ideal when you're answering unexpected business questions, debugging data quality issues, or validating a hypothesis—without setting up a full project.

Step-by-Step Instructions​

  1. Start a new Analysis
  • From the workspace home, click New Analysis.
  • Choose your preferred interface—Spreadsheet, SQL, or Python. You can freely switch or mix modes later.
  1. Pull in pre-modeled metrics, dimensions, and entities
  • Use the metrics dropdown to select existing metrics, dimensions, or entities from the semantic layer.
  • These are reusable, and scoped to your organization’s data model.
  1. Write SQL as Needed
  • Want to reference a staging or transformed table from your ETL layer? Use SQL cells within your analysis to query them directly.
  • You can join these tables with modeled entities or filter existing metrics for ad-hoc deep dives.
  1. Explore and Visualize
  • Apply filters, create quick group-bys, or run aggregations directly in the interface.
  • Highlight any result set to visualize trends using built-in charts like line charts or tables.
  1. Iterate, Save, or Share
  • Keep iterating until you've found the signal in the noise.
  • Save it as a "Report" to revisit later
  • Share your analysis with others
  • If your work becomes recurring, save it as a "Dashboard".

Tips​

  • Combine modeled metrics with custom SQL to bridge governed and ad-hoc analysis seamlessly.
  • Use AI Assist to draft queries, explain columns, or generate follow-up questions.
  • Analyses are ephemeral by default, perfect for lightweight exploration, but you can persist them when needed.