Machine Learning · Causal Inference · Decision Intelligence
AI-Powered SaaS Analytics & Predictive Intelligence
Retention analytics for SaaS churn and revenue: segment-aware churn prediction (SMB, mid-market, enterprise), causal uplift modeling, survival curves (Kaplan-Meier + Cox), cohort retention, revenue forecasting with scenario modeling, and a budget-constrained portfolio optimizer for CS teams.
Advanced AI Capabilities
Built on your own data. From survival curves and segment-aware churn models to causal uplift estimation and a portfolio optimizer for CS budgets.
Churn Prediction
Segment-aware models with 30/60/90-day probabilities and confidence intervals
RetentionLens trains separate models for SMB, mid-market, and enterprise segments — each with distinct feature weights and coefficients. Predictions include 30/60/90-day churn probabilities with confidence intervals, and all scores trace to model version and data confidence state so you can trust what you act on. Kaplan-Meier survival curves and Cox proportional hazards regression provide the survival-analytic foundation.
Revenue Forecasting
Scenario modeling with churn delta and expansion delta sliders
RetentionLens projects forward revenue from your NRR, GRR, MRR waterfall, and expansion dynamics. Interactive scenario sliders let you model the revenue impact of different churn and expansion assumptions — useful for board planning, capacity modeling, and CS prioritization.
Customer Health Scoring
Composite 0–100 score with Red / Amber / Green classification
Every customer gets a health score that rolls up payment history, product usage signals, tenure, plan trajectory, and support interactions into a single 0–100 composite. Scores are classified Red/Amber/Green and feed directly into the churn risk list and decision engine recommendations. Data contract validation (across billing, product usage, CRM, and support sources) flags data readiness so scores degrade gracefully when signals are missing rather than silently becoming unreliable.
Decision Intelligence
Causal uplift modeling, budget-constrained optimization, and full auditability
RetentionLens goes beyond correlation-based churn scores to causal inference: T-Learner heterogeneous treatment effect (HTE) estimation measures the incremental retention impact of each intervention per customer — not just which customers are at risk, but which actions actually move the needle for each one. A DP knapsack portfolio optimizer then allocates your CS budget across the at-risk list subject to velocity constraints (max touches per week), so the team works the highest-expected-value accounts first. Every recommendation traces back to model version, uplift source (learned vs. heuristic), and data confidence state.
learned vs. heuristic — auto-promotes when data is sufficientHybrid analytics + ML architecture
Next.js powers the UI and lightweight APIs; a Python ML service runs heavier compute (survival analysis, uplift estimation, portfolio optimization) and writes results back to Supabase.
ML & Decision Infrastructure
Segment-Aware Models
Kaplan-Meier survival curves, Cox proportional hazards regression, and segment-aware churn models (SMB, mid-market, enterprise) with separate feature coefficients per segment. All models version-tracked per org.
Causal Inference Layer
T-Learner HTE estimation accumulates outcomes as CS actions are taken. The decision engine automatically promotes from heuristic multipliers to learned uplift estimates once sample size reaches Medium confidence — and logs the source on every recommendation.
Data Contract Validation
Incoming data from billing, product usage, CRM, and support is validated against contracts before it enters any model. Timeliness gates, readiness scoring, and synthetic data detection ensure predictions degrade gracefully rather than silently misleading.
Data & governance
Models train on your own historical data where volume supports it. Every prediction includes a data confidence state and EV discount when signals are thin — no silent accuracy degradation.
AI Features by Plan
Choose the right level of AI-powered insights for your business.