— net incremental value
- Targeted customers
- —
- Incremental retained
- —
- Incremental value
- —
- Campaign cost
- —
- ROI
- —
- Break-even uplift
- —
Outputs are scenario estimates, not guaranteed business results.
Methodology and interpretation
Uplift is the change in retention caused by receiving the campaign compared with not receiving it. A propensity score answers who is likely to remain or respond; uplift asks whose outcome changes because of treatment. A high-propensity customer may have stayed anyway, so propensity is not a substitute for an incremental causal estimate.
Inputs
- Eligible population: customers who could receive the campaign.
- Target percentage: the share that will be contacted.
- Uplift: estimated causal retention change in percentage points.
- Costs: delivery plus incentive cost for every targeted customer.
- Contribution value: incremental value from one additionally retained customer.
Formulas and assumptions
- Targeted customers = eligible population × target rate.
- Incremental retained = targeted customers × uplift points ÷ 100.
- Incremental value = incremental retained × contribution value.
- Campaign cost = targeted customers × (contact cost + incentive cost).
- ROI = (incremental value − campaign cost) ÷ campaign cost.
Synthetic worked example
Suppose 10,000 customers are eligible, 40% are targeted, estimated uplift is 5 percentage points, delivery plus incentive cost is $5 per target, and contribution value is $150 per incremental retention. The scenario targets 4,000 customers, retains an estimated 200 incrementally, creates $30,000 in incremental value, costs $20,000, and produces $10,000 net incremental value—a 50% ROI.
Break-even interpretation: break-even uplift equals per-target cost ÷ contribution value × 100. In the synthetic example, $5 ÷ $150 gives 3.33 percentage points. The assumed uplift must exceed that threshold for positive net value. When no customers are targeted, contribution value is zero, or the comparison has no unique solution, break-even is reported as undefined. When campaign cost is zero, ROI is also undefined rather than infinite.
Limitations
This calculator does not estimate causal uplift, uncertainty, heterogeneous treatment effects, discounting, capacity limits, cannibalization, adverse treatment effects, or downstream behavior. Prefer uplift estimates from randomized experiments or defensible causal/uplift modeling. Test sensitivity to uncertain inputs and use scenario outputs as decision support, not guaranteed results.
Continue with uplift modeling, churn prediction and uplift thinking, causal inference for product analytics, the A/B sample-size calculator, and data product patterns.