S12 · decision value simulation
1,000 retention calls - who gets one?
People, data and outputs in this scenario are simulated. Each stage is recorded when you commit; the reveal comes after.
- 5stages
- 35 minabout
- 0stages committed
The brief
A telecom operator (simulated) can make 1,000 retention calls per week. A churn model scores customers with calibrated 90-day churn probabilities. A past randomised test (simulated) measured the save rate: the share of would-be churners that a call retains. Each call costs 15 including the offer (hypothetical). Customer value is 12-month margin (hypothetical). Customers fall into four groups this week:
| Group | Customers | Churn probability p | 12-month margin V | Save rate s |
|---|---|---|---|---|
| A | 300 | 0.60 | 300 | 0.10 |
| B | 400 | 0.30 | 900 | 0.20 |
| C | 600 | 0.15 | 1,200 | 0.25 |
| D | 3,000 | 0.05 | 600 | 0.25 |
Expected value of calling one customer: EV = p x s x V - 15.
Stage 1 of 5
The current practice (simulated) is to call the customers with the highest churn probability.
How should the 1,000 calls be allocated?