ML-powered customer intelligence at scale

Churn Prediction (Boosted Tree)

BigQuery ML Boosted Tree Classifier trained on 19 RFM features -- recency, frequency, monetary value, cancel rate, spend trend, category concentration, and order interval variance. Scores every customer with a churn probability.

RFM Segmentation

Automatic quintile scoring across recency, frequency, and monetary dimensions. Segments customers into Champions, Loyal, At Risk, Hibernating, and New tiers updated with every data refresh.

Customer LTV Forecasting

Boosted Tree Regressor predicts 12-month customer lifetime value. Classifies customers into Platinum, Gold, Silver, and Bronze tiers based on predicted future spend.

Cohort Retention Analysis

Monthly cohort retention tracking from M+0 through M+5. Compare acquisition cohorts side-by-side to identify which channels and campaigns drive long-term engagement.

Customer 360 View

Unified profile combining order history, RFM scores, CLV tier, churn probability, health score, and purchase cycle data into a single actionable view per customer.

Customer Health Scoring

Aggregate health scores with tier breakdown across your entire customer base. Combines recency, purchase frequency, CLV trajectory, and churn risk into one metric.

Purchase Cycle Analysis

Computes median inter-purchase intervals per customer. Identifies customers overdue for their next order and flags purchase frequency anomalies.

Loyalty Decay Detection

Tracks declining purchase frequency and shrinking order values over time. Surfaces customers whose engagement is deteriorating before they churn.

LTV Trajectories & Negative LTV

Projects lifetime value trajectories for each customer segment. Flags customers with negative LTV -- those costing more in returns, support, and acquisition than they generate.

Reactivation Candidates

Identifies lapsed customers with the highest reactivation potential based on historical purchase patterns, RFM scores, and predicted LTV recovery value.