Building the recipe — ingest, join, transform

The recipe is where raw objects become an analysis-ready dataset. We build it node by node.

The recipe pipeline

• Input — Opportunity + OpportunityHistory + User
• Join — attach owner name to each opportunity; join history for stage timing
• Transform — clean nulls, standardise stage names, cast types
• Aggregate — roll up history to one row per opp with per-stage durations
• Output — write the funnel dataset

Join types matter

Use a lookup join to enrich opportunities with user data, and an aggregate node to collapse OpportunityHistory into stage-duration columns.