Churn Prediction & Automated Upsell Automation
Churn Prediction & Automated Upsell Automation
Available from: v1.0.3
SaaS Factory includes a fully autonomous revenue protection and expansion stack. This covers two tightly integrated capabilities: churn prediction (identifying customers at risk of leaving) and upsell automation (identifying customers ready to expand and acting on that signal without human intervention).
Both capabilities run continuously with no human required at any step.
Churn Prediction
How It Works
Churn risk is computed per customer using a rolling ML/AI scoring pipeline that ingests behavioural usage signals:
- Login frequency — declining session activity relative to the customer's own baseline
- Feature adoption — breadth and depth of feature usage over time
- API call volume — drops in programmatic usage
- Support ticket sentiment — negative sentiment trends in support interactions
These signals are combined into a single churn risk score (0–100) and bucketed into a churn risk tier:
| Tier | Score Range | Meaning |
|---|---|---|
low | 0–24 | Healthy, no action needed |
medium | 25–49 | Early warning, monitor closely |
high | 50–74 | At-risk, automated intervention triggered |
critical | 75–100 | Imminent churn, escalated win-back sequence |
Score Recalculation
Scores are recalculated:
- On a scheduled cadence (rolling window, daily minimum)
- On-demand when a significant usage drop event is detected (e.g. API call volume falls >50% week-over-week)
Customer Schema Fields
The following fields are available on the customer record:
churn_risk_score Integer (0–100)
churn_risk_tier Enum: low | medium | high | critical
Dashboard Visibility
Churn risk scores and tier trends are surfaced in the Customer Management section of the platform. Each customer record displays:
- Current score and tier
- Score trend over the last 30 days
- The primary contributing signals
Automated Upsell Automation
Upsell Opportunity Detection
The upsell engine continuously monitors customers for expansion signals:
- Plan limit proximity — customers approaching seat, API call, or storage limits
- Usage velocity — customers whose usage is growing fast relative to their current plan ceiling
When an opportunity is detected, the customer is flagged as expansion-ready and an upsell sequence is initiated automatically.
Trigger Rules Engine
The rules engine dispatches upsell actions without human input:
- In-app prompts — contextual upgrade nudges served within the product at relevant moments
- Email campaigns — automated outreach tied to the specific limit or feature driving the opportunity
All upsell events are logged to the deal pipeline for full revenue attribution and reporting.
Dunning-to-Win-Back Workflow
The billing engine's dunning flow extends beyond basic past_due invoice retries. Customers who churn through non-payment enter an automated win-back sequence:
- Re-engagement email cadence — timed series of emails with account recovery messaging
- Discount offer generation — automated generation of personalised discount offers where appropriate
- Outcome tracking — reactivated or lost outcomes are recorded and fed back into the churn model as training signal, improving future predictions
Dunning vs. Win-Back
| Stage | Trigger | Action |
|---|---|---|
| Dunning | Invoice past_due | Payment retry + notification emails |
| Win-Back | Subscription churned via non-payment | Re-engagement cadence + discount offer |