Expert analysis · reviewed 5 Oct 2026
Customer lifetime value (CLV): calculating what a customer is worth without overstating retention.
Key distinctions
Three conditions before calculating.
Margin, not revenue
A CLV calculated on revenue overstates the value of every low-margin customer. Variable costs, discounts, returns and cost to serve are deducted before any projection.
The cohort, not the average customer
Retention varies with the acquisition channel, the entry offer and tenure. A global average mixes customers who do not justify the same acquisition cost.
A projection, not an asset
CLV remains an estimate conditional on future retention. It informs an acquisition or retention decision, but it is neither an accounting figure nor a valuation of the customer base.
Method
Building a defensible CLV in four steps
- 01Define the active customer and the cohort
Set the entry event, the period, the acquisition channel and the inactivity rule; in a non-contractual relationship, churn is not observed and must be modelled.
- 02Measure margin per period
Start from contribution after variable costs, discounts, returns and cost to serve, by cohort and by tenure period.
- 03Estimate retention and discount
Use observed survival for as long as it exists, document the model used beyond it, then discount each margin at the rate set by finance.
- 04Compare with an incremental CAC
Relate CLV to the acquisition cost of the same cohort, measured on the customers actually won thanks to the spend, not on every attributed customer.
The same customer, four CLV conventions
With the same data — margin, retention, discounting and CAC — the convention chosen makes CLV vary threefold and reverses the reading of the CLV/CAC ratio.
Assumptions: annual margin m = €120, retention r = 80%, discount rate d = 10%, CAC = €150.
| Convention | Formula | Implicit assumption | CLV | CLV / CAC |
|---|---|---|---|---|
| Naive LTV | m ÷ (1 − r) | Constant retention, no discounting | €600 | 4.0 |
| Discounted retention | m × r ÷ (1 + d − r) | Margin received at the end of each period | €320 | 2.1 |
| Initial margin included | m × (1 + d) ÷ (1 + d − r) | First margin earned at purchase | €440 | 2.9 |
| Cohort observed over 3 years | Σ m × S(t) ÷ (1 + d)^t | Measured survival, truncated horizon | €190 | 1.3 |
The CLV/CAC ratio only makes sense with its convention, its horizon and an incremental CAC measured on the same cohort.
The CLV/CAC ratio depends first on the convention chosen
Illustrative example: annual margin per customer of €120, annual retention of 80%, discount rate of 10% and acquisition cost of €150.
Naive LTV = 120 ÷ (1 − 0.80) = €600, i.e. 4.0 times the CAC. Discounted CLV, margin at the end of each period = 120 × 0.80 ÷ (1 + 0.10 − 0.80) = €320, i.e. 2.1 times the CAC. Cohort observed over three years, with survival of 80%, 60% then 48%: 87 + 60 + 43 = €190, i.e. 1.3 times the CAC.
The same customer looks highly profitable, profitable or barely profitable depending on the convention. The acquisition decision rests on the observed, prudent CLV; the long-term projection remains a scenario to be confirmed.
Acceptance conditions
What must be true to act.
- 01
Calculate CLV on contribution margin, never on revenue.
- 02
Publish the horizon, the discount rate and the timing of the first margin with every CLV.
- 03
Cap the CAC on the observed CLV of the cohort as long as projected retention is not confirmed.
- 04
Revise the CLV as soon as the observed retention of a cohort departs from the assumption used.
Limits
What this analysis does not prove.
- The constant-retention formula ignores customer heterogeneity and the rise of retention with tenure.
- In a non-contractual relationship, churn is inferred by a probabilistic model whose assumptions must be checked.
- CLV captures neither referrals, nor network effects, nor the option value of a customer.
- A high CLV/CAC ratio does not prove that the acquisition spend is incremental.
References