We use your customer data to predict, target, and convert.

{"user_id": 4892, "last_purchase": "2024-01-03", "ltv": null, "status": "active"}

SELECT * FROM events WHERE timestamp > NOW() - INTERVAL 90 DAY;

campaign_id,spend,clicks,conv,roas\n40221,4200,812,23,1.49

{"churn_prob": 0.73, "segment": "high_value", "days_silent": 42}

INSERT INTO predictions (customer_id, score) VALUES (8812, 0.91);

user_id | email_opens | purchase_freq | predicted_ltv\n4892   | 2           | 0.3           | $47.20

ad_set: "LAL_top2pct" | cpa: $38.40 | true_roas: 1.2x

{"event": "page_view", "path": "/checkout", "dropped": true}

0.84, 0.23, 0.91, 0.07, 0.62, 0.45, 0.78, 0.33

WITH cohort AS (SELECT user_id, MIN(order_date) AS first FROM orders GROUP BY 1)

{"model": "churn_v3", "auc": 0.89, "features": 34, "updated": "2025-12-01"}

source,medium,sessions,bounce_rate\ngoogle,cpc,12840,0.67

retention_d7: 0.41 | retention_d30: 0.18 | retention_d90: 0.06

PREDICT churn_risk USING model('vertex_churn_v3') ON customer_features;

email_id: 991204 | open: true | click: false | unsub: false

{"ltv_30d": 22.40, "ltv_90d": 67.10, "ltv_365d": 189.00}

pixel_fire: [fb, ga4, rudderstack] | dedup: false | attribution: last_click

UPDATE campaigns SET status = 'paused' WHERE roas < 1.0 AND spend > 500;

customer_8812: churn_score=0.91, action=WIN_BACK, urgency=HIGH

{"send_time_optimal": "08:42", "timezone": "EST", "confidence": 0.87}

0.02, 0.94, 0.11, 0.55, 0.38, 0.76, 0.89, 0.04

channel | attributed_rev | incremental_rev | delta\nmeta   | $84,200        | $12,400         | -85%

{"segment": "at_risk", "count": 1847, "pct_of_base": 0.26}