Lifetime value = the present value of all contribution a customer will generate before leaving. Every phrase in it counts:
- Contribution, never revenue. A customer paying ₹5,400/year at a 22% contribution margin is worth ₹1,188/year of margin. Revenue-based "LTV" would call it ₹5,400. That version is astonishingly common in decks, and it overstates by 1/margin, which works out to 3–5×.
- Before leaving. The lifetime comes from churn, and specifically from mature-cohort churn, with the geometric caveat owned.
- Present value. Money five years out isn't money now. You learned that with the time value of money, and it was never optional. Most startups skip this correction, and it isn't small.
How is customer lifetime value calculated?
You build it in two steps.
Naive (undiscounted): LTV = m / c, annual contribution per customer m divided by annual churn c, which is the same as m × lifetime. With m = ₹1,200 and c = 20%, LTV = ₹6,000.
Advertisement
Discounted, the version to trust: treat the customer as an annuity that survives each year with probability r = 1 − c and discount at rate d. Expected contribution in year t = m·r^t, using the end-of-period convention. Sum the geometric series and you get:
LTV = m·r / (1 + d − r)
That's the standard corrected formula (Gupta–Lehmann). Derive it once yourself from the geometric series you met with the time value of money, so it's yours: Σ m·r^t/(1+d)^t, t=1…∞ = m·(r/(1+d)) / (1 − r/(1+d)) = m·r/(1+d−r). With m = ₹1,200, r = 0.80 and d = 12%, LTV = 1,200 × 0.80 / (1.12 − 0.80) = 960 / 0.32 = ₹3,000.
Correcting for discounting and survival cut LTV in half, from 6,000 to 3,000. That's a big haircut for one line of algebra. The naive formula implicitly assumes the customer's distant, ever-less-likely contributions are worth face value today. And the overstatement grows with the lifetime. At 10% churn and 12% discounting, naive LTV = 10.0×m while the discounted version = 4.09×m, a 144% overstatement. It's worst where decks brag loudest ("our customers stay ten years!").
What rules keep an LTV estimate honest?
Five rules of engagement. (1) Margin-based, always. (2) Churn from mature cohorts. (3) Discount at a rate that reflects the riskiness of the retention forecast. The company's cost of capital is the floor, and for a two-year-old startup's cohort claims the right rate is well above it. (4) Credit expansion (NRR > 100%) only if cohort data proves it, and never extrapolate expansion forever. (5) Cap the horizon for young companies. Crediting years 8–15 of a customer relationship to a company that is 4 years old is narrative dressed up as analysis.
LTV is a forecast wearing a formula's clothes. Trust the formula as far as the cohort data behind it, and no further.