LTV (Customer Lifetime Value)

Customer lifetime value is the total value a business expects to earn from one customer over the whole relationship. It answers a single question: how much is a customer worth to us, on average, before they churn?

\[ \text{LTV} = \text{ARPU} \times \text{gross margin} \times \text{customer lifetime} \]

• ARPU — average revenue per user, monthly or annually
• Gross margin — the share of revenue kept after cost of goods sold
• Customer lifetime — how long the customer stays, in months or years

The formula is a simplified model, not a definition that fits every customer relationship. Revenue, margin, and expected paid lifetime must refer to compatible periods and populations. If valuable customers tend to stay longer, multiplying separate population averages can misrepresent expected value; estimate their joint contribution or use appropriately defined cohorts instead.

Define what counts as value

In an illustrative subscription, monthly revenue is 50 monetary units and gross margin is 80%, leaving 40 per month after the costs included in that margin. If the expected number of paid months is 24 and the monthly contribution remains constant, expected lifetime contribution is 40 × 24 = 960. Revenue over that period is 1,200. Subtracting an acquisition cost of 250 leaves 710. These quantities answer different questions and should not share an unlabeled LTV column.

The word “net” is used for more than one quantity, so the cost boundary has to be named. Gross LTV is expected lifetime revenue. Multiplying it by gross margin, as the formula above does, gives a gross-profit LTV: revenue less the costs classified in cost of goods sold. Net LTV, or contribution LTV, subtracts a defined set of costs attributable to serving the customer. It equals the gross-profit LTV only when all of those costs are already inside the margin; additional attributable costs such as support or payment processing give a lower value, so a contribution LTV should use a margin that reflects the chosen cost boundary.

Subtracting acquisition cost is a separate step. The CAC-subtracted version is a real quantity and a useful one, but dividing it by CAC gives (LTV – CAC)/CAC = LTV/CAC – 1, an ROI-style net-return ratio rather than the usual value-to-cost ratio. Using a CAC-subtracted LTV in place of the pre-CAC LTV therefore changes what an LTV:CAC ratio means, so the two should not be compared as if they were the same ratio. Subtracting CAC again from a value that already excludes it, as in ((LTV – CAC) – CAC)/CAC, would be a genuine double subtraction. When the term arrives from outside, the first question is which costs were subtracted: only cost of goods sold, a wider set of attributable costs, or acquisition cost as well.

Retention, payment timing, and discounting

Let m be a constant monthly contribution, r the probability of remaining for the next payment, and d the monthly discount rate. Assume the first payment arrives at the end of month one and is guaranteed; each later payment requires another retention event. Expected payments are m, mr, mr², and so on. Discounting each by when it arrives gives \(m/(1+d) + mr/(1+d)^2 + \cdots = m/(1+d-r)\). With no discounting, this becomes m/(1-r). These are geometric-series results under constant retention and contribution, not empirical laws about customers.

In this simplified undiscounted model, the LTV:CAC ratio is proportional to revenue and margin and inversely proportional to churn and CAC. A 10% relative reduction in either denominator raises the ratio by 1/0.9 − 1, or about 11.1%, holding everything else fixed. That is not a ranking of interventions. The cost of reducing churn, the customers reached, effects on margin, and changes in acquisition quality determine which action is worthwhile.

With 10% monthly churn, r = 0.90 and expected paid lifetime under this convention is 1/(1-r) = 10 months. A 1% monthly discount changes a contribution of 40 from an undiscounted value of 400 to 40/0.11, approximately 363.64. At 1% churn, the corresponding values are 4,000 and 2,000: payments far in the future contribute less today. If the customer must survive before the first payment, multiply both formulas by r. A finite forecasting horizon instead ends the sum at that horizon; zero churn does not establish an infinite customer relationship.

Variants, and what they are for

Historical LTV sums what customers have already contributed over a stated observation period. A forecast adds expected future contribution and is uncertain. Cohort-based LTV estimates the revenue, margin, and survival pattern of comparable customers as they age. It can allow churn to change with tenure, but an unfinished cohort still needs an assumption about the unobserved future. Predictive LTV uses customer information to estimate future value; evaluate its forecasts against subsequently observed outcomes, using only information available at the forecast date.

For acquisition budgeting, compare the expected contribution with acquisition cost while keeping payment timing and uncertainty visible. A high expected LTV does not solve a near-term cash shortage. For segmentation, compare groups defined consistently by channel, product, or customer need; an overall average can hide groups with very different economics. For forecasting, distinguish observed contribution from an extrapolated tail and test how decisions change under plausible retention and margin assumptions.

Improving retention, pricing, service cost, or customer adoption can change lifetime contribution, but each intervention can affect several inputs at once. A discount may reduce churn while reducing margin. Higher prices may raise revenue from customers who stay while causing others to leave. Estimate the combined change and the cost of the intervention rather than selecting the input with the largest isolated percentage effect.

References

• Stripe Support, Calculating subscriber Lifetime Value in Billing. The reported revenue/churn metric differs from the margin-based convention explained here.