Customer lifetime value calculator

Margin LTV and revenue LTV are not interchangeable

Two businesses with identical revenue per customer can have lifetime values that differ threefold, because one of them keeps most of what it charges and the other keeps a sliver. Quoting the wrong one next to an acquisition cost is how a business convinces itself that unprofitable growth is working.

• Margin LTV multiplies lifetime revenue by gross margin, so it counts money the business actually keeps. This is the figure that belongs beside CAC.
• Revenue LTV is the number most articles compute without saying so. It is useful for sizing a market and dangerous for setting an acquisition budget.
• The gap between them is exactly your gross margin. At 65% margin, an 720 unit revenue LTV is a 468 unit margin LTV, and an acquisition cost of 500 flips from comfortable to loss-making between those two lines.
• Neither figure subtracts the cost of serving and supporting the customer over time, which for anything with a support burden is a further meaningful bite.

The exact sums this page runs

A simple non-discounted model, stated in full so you can see everything it assumes. Annual revenue per customer is named separately because it is the block both lifetime figures and the payback number are built from.

• Annual revenue per customer = Average order value × Purchases per year. Order value times purchase frequency. Both lifetime figures below are built on this block.
• Lifetime value on gross margin = Annual revenue per customer × Customer lifespan in years × (Gross margin ÷ 100). The version a finance team accepts, because it counts money the business keeps.
• Lifetime revenue = Annual revenue per customer × Customer lifespan in years. The version most articles mean without saying so. It ignores cost of goods entirely, so it runs high.
• Annual gross profit per customer = Annual revenue per customer × (Gross margin ÷ 100). The payback figure: divide acquisition cost by this to see how many years a customer takes to repay it.

Getting a lifespan you can defend

Lifespan is the input people invent, and it is the one that moves the answer most. These are the ways to estimate it without guessing, in rough order of how much data each needs.

1. 1If you know annual churn, start with 1 divided by the churn rate. Losing 25% of customers a year implies an average lifespan of about four years.
2. 2For a monthly subscription, use 1 divided by monthly churn to get a lifespan in months, then convert. A 5% monthly churn implies twenty months, not twelve.
3. 3Sanity-check the answer against how long you have existed. A two-year-old business cannot observe a five-year lifespan, and quoting one is a forecast dressed as a measurement.
4. 4Cap the horizon at three to five years even when the maths suggests more. Money arriving in year eight is worth considerably less than money now, and this model does not discount it.
5. 5Recalculate by cohort rather than in aggregate. Customers acquired through discounting churn faster than customers acquired at full price, and a blended lifespan hides exactly the difference you need.
6. 6Rerun the figure whenever pricing or the product changes materially, because a lifespan estimated on the old product is evidence about a business that no longer exists.

What this model deliberately leaves out

Every lifetime value is a forecast, and this one is a deliberately simple forecast. Knowing its omissions is what stops it being quoted as a fact in a board pack.

• No discounting. A proper model discounts future cash flows to present value, which lowers LTV and lowers it most for the long lifespans that flatter the number.
• No expansion or contraction. Customers who upgrade, downgrade or buy adjacent products are all treated as flat, which understates good businesses and overstates fragile ones.
• Averages hide the distribution. Lifetime value is usually concentrated in a small share of customers, so the average describes almost nobody and the median tells a different story.
• Survivorship bias creeps in through the lifespan estimate. Measuring only customers still present at the point you look overstates how long a typical customer stays.