Churn Prediction & Retention Engine
A classification layer per customer — churn probability with confidence bands, risk-driver attribution (why the model thinks this customer is at risk), and retention-action recommendations tied to the predicted risk drivers.
- Engagement
- 6–10 week build · ongoing operation
- Built for
- Heads of growth · Customer success leads · CMOs
Churn is the slow-moving alpha — most businesses know their aggregate churn rate but don't predict which customers will churn next quarter or what the underlying risk driver is.
What this is
A churn-prediction-plus-action layer over the customer base. Three components:
- Churn prediction. Per-customer probability of churn in the relevant future window (30 / 60 / 90 days depending on the business). Confidence bands.
- Risk-driver attribution. SHAP-value-based per-prediction. The customer success team sees the why, not just the what.
- Retention-action library. Per-driver action recommendations grounded in the business's historical retention success.
How it's built
LightGBM with monotonic constraints for interpretability, SHAP for attribution, retention-library curated during the engagement with the customer success team. Integration into the business's CRM and support stack for surface.
What you get
- The churn-prediction model.
- The risk-driver attribution layer.
- The retention-action library.
- CRM and support-stack integration.
- Quarterly model refresh.
Engagement is shape, not list.
Length and price are functions of the data and the destination. The shape below is the typical engagement.
- Length
- 6–10 week build · ongoing operation
- Lead
- Bogdan
- Cadence
- Async, weekly
- Bar
- Production
Scoped during the discovery call against the actual data and the operation it integrates with.
Principal engineer. Architecture and most code ships through one keyboard.
Written updates between, calls when the decision needs the room.
Async correctness, capacity under burst, observability at every boundary.
Products this composes with.
Same suite, or vertical-specialized versions in another.
- Same suite · Operations Algorithms Suite
Customer Clustering & LTV Engine
An unsupervised clustering layer over the client's own customer-and-transaction data — surfaces behavioral segments using RFM and broader signal sets, attaches a lifetime-value forecast per segment, feeds marketing and retention decisions.
- Same suite · Operations Algorithms Suite
Real-Time Personalization API
An API that returns the next-best-product, next-best-content, or next-best-offer per user — fed by the client's own behavior data, augmented (with consent) by Subscription Economy Benchmarks contextual signal where the use case warrants.
What buyers ask about this one.
Doesn't every SaaS now have built-in churn prediction?
Built-in churn prediction often works for the SaaS's intended use case (annual SaaS subscriptions with usage-driven signals). For businesses with more complex buying patterns — multi-product subscriptions, B2B with renewal-decision committees, transactional retail with intermittent purchase — the built-ins underperform. Custom models capture the structure the built-ins miss.
How does the risk-driver attribution work?
SHAP-value-based attribution per prediction — which features drove the model's risk assessment. The customer success or marketing team sees not just 'high risk' but 'high risk because: usage declined 30% over 60 days, support tickets unresolved, account-owner changed.' Actionable, not just descriptive.
What about the retention-action recommendation layer?
Per-risk-driver, the engine attaches the retention actions that have worked historically — for the 'usage declined' driver, that might be a usage-revival outreach; for the 'support tickets unresolved' driver, an executive escalation; for the 'account-owner changed' driver, a re-onboarding cycle. The retention library is per-business; we build it during the engagement.
How is this different from the Tenant Intelligence Engine in the RE Suite?
Same shape, different domain. The RE version is built for residential/commercial tenant data; this is built for general customer data. Both use the same modeling backbone; the difference is feature engineering and the retention-action library. We list both because the buyers are different (property operator vs. CMO).
Pricing?
Scoped to data complexity and the integration depth. Discovery call covers both.
If the deliverable matches the gap, the next step is one call.
We'll scope length and price against your data and the operation it integrates with. No retainer, no fishing.
Bogdan and team · async-first · OP—2026