The Problem

• Businesses struggle to identify which customers are at risk of churn.
• Existing reports are static and don’t provide real-time insights.
• Customer success teams lack actionable intelligence to reduce churn.

Our Solution

• Built a Cortex Analyst POC on Snowflake using the DEMO_POC.CHURN dataset.
• Enabled natural language queries for churn metrics (e.g., “What’s churn by contract length?”).
• Delivered interactive dashboards and semantic views for analysts and executives.

Solution Architecture

• Data ingestion: Customer churn dataset loaded into Snowflake (CUSTOMER_CHURN table).
• Semantic view: DEMO_POC.CHURN.DEMO created for structured queries.
• Cortex Analyst: Configured to answer churn questions in natural language.
• Evaluation runs: Verified correctness and logical consistency of responses.
• Access via CoWork: Business users query churn insights directly.

Deliverables

• Cleaned customer churn dataset.
• Semantic view for churn metrics.
• Cortex Analyst agent for natural language queries.
• Evaluation reports showing accuracy and performance.
• Demo scripts and SQL queries for churn analysis.

Tech Stack

• Snowflake Cortex Analyst
• Snowflake Warehouse (COMPUTE_WH)
• SQL (for churn queries)
• Power BI (optional dashboards)
• Bandicam (for demo video recording)

Business Impact

• Customer success teams can identify at-risk accounts faster.
• Executives get clear churn summaries for decision-making.
• Data analysts save time by querying in natural language instead of writing SQL.
• Potential to reduce churn by 5–10% through proactive interventions.