Promotion Optimization
“Among notification-eligible users, will they purchase within 4 days — assuming they receive a push notification?”
Book a demo and get a free trial of the full platform: research agent, fine-tune capabilities, and forward-deployed engineer support.
By submitting, you accept the Terms and Privacy Policy.
Loved by data scientists, ML engineers & CXOs at
A real-world example
Among notification-eligible users, will they purchase within 4 days — assuming they receive a push notification?
Marketing claims push notifications drive purchases, but how much is actually caused by the notification vs. would have happened anyway? Traditional A/B tests are slow, expensive, and measure only average treatment effects — missing the per-user variation that determines whether a promotion creates value or just gives away margin.
How KumoRFM solves this
Relational intelligence for optimal actions
Kumo's ASSUMING clause enables counterfactual prediction: compare the predicted purchase probability with vs. without a push notification for each user. This per-user causal uplift identifies the 30% of promotions that actually drive incremental purchases — and the 70% where the customer would have bought anyway. No holdout group needed, no weeks of waiting.
From data to predictions
See the full pipeline in action
Connect your tables, write a PQL query, and get predictions with built-in explainability — all in minutes, not months.
Your data
The relational tables Kumo learns from
USERS
| user_id | notif_eligible | signup_date |
|---|---|---|
| U-4001 | 1 | 2025-06-15 |
| U-4002 | 1 | 2025-09-22 |
| U-4003 | 1 | 2026-01-03 |
| U-4004 | 1 | 2025-03-10 |
| U-4005 | 1 | 2025-12-01 |
PURCHASES
| purchase_id | user_id | amount | timestamp |
|---|---|---|---|
| PUR-501 | U-4001 | $89 | 2026-02-20 |
| PUR-502 | U-4002 | $145 | 2026-02-22 |
| PUR-503 | U-4001 | $62 | 2026-03-01 |
| PUR-504 | U-4004 | $210 | 2026-02-28 |
| PUR-505 | U-4003 | $35 | 2026-03-05 |
NOTIFICATIONS
| notif_id | user_id | type | timestamp |
|---|---|---|---|
| NTF-601 | U-4001 | PUSH | 2026-02-19 |
| NTF-602 | U-4002 | PUSH | 2026-02-21 |
| NTF-603 | U-4003 | 2026-03-04 | |
| NTF-604 | U-4004 | PUSH | 2026-02-27 |
| NTF-605 | U-4005 | PUSH | 2026-03-01 |
Write your PQL query
Describe what to predict in 2–3 lines — Kumo handles the rest
PREDICT COUNT(PURCHASES.*, 1, 4, days) > 0 FOR EACH USERS.USER_ID WHERE USERS.NOTIF_ELIGIBLE = 1 ASSUMING COUNT(NOTIFICATIONS.* WHERE NOTIFICATIONS.TYPE = 'PUSH', 0, 1, days) > 0
Prediction output
Every entity gets a score, updated continuously
| USER_ID | True_PROB (with push) | True_PROB (without) | Uplift |
|---|---|---|---|
| U-4001 | 0.82 | 0.78 | +0.04 |
| U-4002 | 0.71 | 0.35 | +0.36 |
| U-4003 | 0.64 | 0.61 | +0.03 |
| U-4004 | 0.55 | 0.52 | +0.03 |
| U-4005 | 0.48 | 0.12 | +0.36 |
Understand why
Every prediction includes feature attributions — no black boxes
User U-4002
Predicted: Uplift = +0.36 (high causal impact)
Top contributing features
No organic purchase pattern — buys only after push (PURCHASES)
0 unprompted
35% attribution
Push open rate = 92% (NOTIFICATIONS)
92% opened
28% attribution
Signup recency < 6 months (USERS)
5.5 months
22% attribution
Peer users with similar behavior show high uplift (graph)
78% peer uplift
15% attribution
Feature attributions are computed automatically for every prediction. No separate tooling required. Learn more about Kumo explainability
PQL Documentation
Learn the Predictive Query Language — SQL-like syntax for defining any prediction task in 2–3 lines.
Python SDK
Integrate Kumo predictions into your pipelines. Train, evaluate, and deploy models programmatically.
Explainability Docs
Understand feature attributions, model evaluation metrics, and how to build trust with stakeholders.
Bottom line: Identify which 30% of promotions actually drive incremental purchases. Remove friction for the 70% who would have bought anyway. Save $2-5M in wasted promotional spend while maintaining the same revenue.
Related use cases
Explore more next-best-action use cases
Topics covered
