How snowplow/dbt-snowplow-attribution works
Main Components
a) Data Sources Layer:
• Snowplow Unified Views (touch points source)
• Snowplow Unified Conversions (conversion data)
• Marketing Spend Data (optional source)
b) Processing Layer:
• Path Transformation Components
• Attribution Models
• Incremental Processing Logic
c) Output Models Layer:
• Core Attribution Models
• Optional Reporting Models
• Manifest Management
┌─────────────────┐ ┌──────────────────┐ ┌─────────────┐
• Unified Views │ │ Conversions │ │Marketing Spend│
• ┬────────┘ └────────┬─────────┘ └──────┬──────┘
┌─────────────────────────────────────────────────────────┐
• Path Transformation Logic │
• ┌─────────────┐ ┌─────────────┐ ┌────────────┐ │
• Create UDFs │─-->│Transform │─-->│Trim Paths │ │
• ┘ │Paths │ └────────────┘ │
• ┘
┌─────────────────┐ ┌─────────────────┐ ┌────────────┐
• Paths to │ │Campaign/Channel │ │Attribution │
• Conversion │ │Attribution │ │Overview │
• ┘ └─────────────────┘ └────────────┘
Key Relationships to Highlight
• How the unified views feed into the path transformation logic
• Connection between conversions data and attribution models
• Demonstrate how the manifest table manages incremental processing
• Relationship between channel/campaign attributions and the overview
Important Technical Details
• Supported platforms (Snowflake, BigQuery, Databricks & Redshift)
• Incremental processing flow
• Different attribution algorithms available
• Version requirements (dbt-core 1.6.0+)