Lakehouse Propensity Scoring

Overview‍

Learn how to leverage Snowplow's rich behavioral data to achieve our goal of converting website visitors into engaged prospects by applying ML and remarketing them through various channels.

We will utilize sample data from Snowplow's website to convert prospects from a top of funnel awareness stage to engaged users (e.g. request for demo). To unlock the value of the underlying behavioral data, we will focus remarketing efforts on prospects with highest propensity to convert. You will learn how to build a rules-based audience segment and ML-based audience segment to utilize in your activation platform.

Use cases

• Create compliant behavioral data using Snowplow
• Load data into DeltaLake
• Create silver tables using Snowplow's DBT package
• Perform exploratory data analysis using Databricks SQL
• Create gold table for audience segmentation
• Use Reverse ETL or Orchestration CDP for rules-based remarketing
• Train propensity to convert model using XGBoost and Mlflow
• Use model to predict propensity to convert
• Use Reverse ETL or Orchestration CDP for ML-based remarketing
• Monitor campaign performance and optimize efforts

Product details

‍-snowplow_web_page_views

-snowplow_web_sessions

-snowplow_web_users

-converted_users-campaigns

-snowplow_user_propensity_scores

For more details, refer to the embedded notebook.

Additional Insights

‍To learn about additional use cases for Snowplow behavioral data in Databricks, see our blog post here.