Lock the cohort, horizon and attribution contract
Inputs that must travel with the result
- One defined eligible customer cohort and one programme observation period. (not complete)
- A no-programme baseline or controlled comparison used to estimate incremental referred customers. (not complete)
- One conversion event, attribution rule, deduplication rule and conversion window. (not complete)
- Contribution per referred customer on a declared currency and indirect-tax basis. (not complete)
- The economic cost and trigger for each referrer and referred-customer reward. (not complete)
- Programme setup, software, support, moderation, fulfilment and review costs for the same horizon. (not complete)
Calculate programme contribution and the recovery threshold
programme contribution = (incremental referred customers × contribution per referred customer) - earned incentive cost - programme operating cost
- incremental referred customers
- Attributed referred customers above the evidenced no-programme baseline (customers in the selected cohort and window) — business observation or explicitly labelled scenario
- contribution per referred customer
- Revenue less the declared variable delivery, fulfilment, return and service boundary (CU per customer over the selected horizon) — business record or user scenario
- earned incentive cost
- Economic cost of rewards whose stated trigger was met (CU for the selected cohort) — programme and accounting record
- programme operating cost
- Included setup and operating cost for the same period (CU per programme period) — business record or user scenario
When the reward is a constant cost per incremental customer, break-even customers = fixed programme cost ÷ (contribution per customer - reward cost per customer). The denominator must be positive.
- Reconcile attributed customers and remove duplicates, ineligible events, cancellations and returns under the stated rule.
- Estimate the no-programme baseline and label the difference as observed or assumed incrementality.
- Build contribution per referred customer using the same finite horizon.
- Map each reward to its actual trigger and economic cost.
- Add programme operating cost once and calculate the signed result.
- Test lower, base and higher incremental-customer cases without changing unrelated inputs silently.
- Stop if the per-customer recovery denominator is zero or negative.
Worked example: an 18-customer referral scenario
Invented tax-excluded scenario: 18 incremental referred customers, 35 CU contribution per customer, a 20 CU earned reward per customer and 120 CU programme operating cost.
| Line | Calculation | CU |
|---|---|---|
| Incremental customer contribution | 18 × 35 | 630 |
| Earned incentive cost | 18 × 20 | (360) |
| Programme operating cost | entered for period | (120) |
| Programme contribution | 630 - 360 - 120 | 150 |
| Incremental customers | Contribution | Rewards | Operating cost | Programme contribution |
|---|---|---|---|---|
| 8 | 280 | (160) | (120) | 0 |
| 13 | 455 | (260) | (120) | 75 |
| 18 | 630 | (360) | (120) | 150 |
Check the failure modes before acting
- Counting every referral-code order as incremental.
- Using revenue or unbounded LTV as contribution.
- Ignoring rewards to both sides of a referral.
- Counting issued points at face value without an economic-cost policy.
- Omitting returns, cancellations, abuse review, support and software cost.
- Comparing cohorts with different attribution windows or customer definitions.
Referral break-even questions
- Should I use customer lifetime value?
- Use only a contribution amount bounded to an explicit finite horizon and supported cohort. A first-order view is safer when later retention evidence is weak.
- Do all referred customers count?
- No. The economic question uses customers incremental to the baseline. Attribution identifies credit under a rule; it does not prove the programme caused the customer.
- What if reward cost is at least customer contribution?
- There is no positive per-customer amount to recover fixed programme cost under that scenario. Change the economics or stop rather than report a misleading threshold.
Sources and methodology
- Customer Acquisition Cost methodology — Margin101: Aligned customer, period, cost and attribution boundaries.
- Customer Profitability Diagnostic methodology — Margin101: Contribution after service, acquisition and retention costs.
- Customer Referral Incentives and Social Media — Management Science: Supports conditional reward-design risk; no response rate or reward is imported.
Measure acquisition and downstream customer economics
- Customer Acquisition Cost Planner
Measure loaded and marginal customer acquisition cost across aligned channels
- Lead Value Calculator
Estimate contribution-based lead value and a sustainable cost-per-lead ceiling
- Customer Profitability Diagnostic
Diagnose customer profitability after service costs