The Customer Lifetime Value Formula, and the One Term That Bends It

A board deck says lifetime value is $430. The finance model says $180.60. Nobody made an arithmetic error. The two numbers came out of the same four inputs, and they disagree because one multiplied revenue and the other multiplied what survived the cost of serving the order. A gap that size decides whether a $60 acquisition cost looks like a bargain or a slow leak, so it is worth knowing which version you are holding before you set a bid against it.

Most pages that rank for this formula print it, work one example, and stop. The formula is the easy part. The hard part is that three of its four terms are measurements and the fourth is a forecast, and the forecast is the one that moves the answer fastest.

The formula, and what each term has to be

Written out, with the margin term where it belongs:

LTV = AOV x contribution margin % x orders per year x (1 / annual churn rate)

Take the worked example this site uses throughout. Average order value $86, contribution margin 42 percent, so $36.12 of contribution per order. Customers buy 2.5 times a year and 50 percent of them are gone by the end of it. That gives $90.30 of contribution a year, an expected life of 1 / 0.50 = 2 years, and a lifetime value of $90.30 x 2 = $180.60.

Run the same customer on revenue instead and you get $86 x 2.5 x 2 = $430. That is the board deck number. It is not wrong as a description of the revenue a customer will produce, it is simply not spendable. The ratio between the two, 430 / 180.60 = 2.381, is 1 / 0.42. It is the margin, put back.

The first two terms come from your order data and are the least argued about. Average order value is a division. The margin term is where the disagreements start, and the fourth term is where the number stops being a measurement at all.

The margin term decides whether the number is real

Three margins can sit in that slot and they are not interchangeable. Gross margin on the $86 order is 60 percent, or $51.60. Contribution margin, after payment processing, pick and pack, shipping and the returns provision, is 42 percent, or $36.12. Net margin, after the fixed costs, is whatever is left once the warehouse lease has been paid, and it is not a per-order number at all.

Use gross margin and this customer is worth $51.60 x 2.5 x 2 = $258.00 instead of $180.60. That is 42.9 percent more lifetime value for no change in customer behaviour, purely from leaving out costs that the order genuinely causes. The trap is not academic: a $60 acquisition cost looks comfortable against $258 and much tighter against $180.60, and the second number is the one your bank balance agrees with. The argument in full is in contribution margin versus gross margin, and the contribution margin calculator will strip the per-order costs out for you.

One caution on the average order value term. Contribution margin is not constant as order value rises, because some per-order costs do not scale. A $172 order does not cost twice as much to pick or ship as an $86 one. That works in your favour on large baskets and against you on small ones, and it means you cannot lift the AOV term in the formula without recalculating the margin term underneath it.

Lifespan is 1 divided by an assumption

The fourth term is usually written as customer lifespan in years, and people fill it in by guessing three, or five, because the number has to be something. The defensible version is 1 / churn, which is not a guess but a consequence: if a constant fraction of the cohort leaves each year, the expected life is the reciprocal of that fraction. At 50 percent annual churn it is 2 years. At 40 percent it is 2.5. At 60 percent it is 1.667. The churn rate calculator does the conversion in both directions.

Two things follow, and only the first is widely known. The first is that churn is the highest-leverage input in the formula. Move average order value, margin or purchase frequency 20 percent in either direction and lifetime value moves 20 percent with it, because those three terms multiply. Move churn 20 percent and it does not. Dropping churn from 50 to 40 percent takes lifetime value from $180.60 to $225.75, a rise of 25.0 percent. Letting it slip from 50 to 60 percent costs you $30.10, a fall of 16.7 percent. Same size of input move, different size of output move, and the good direction pays more than the bad direction costs. That asymmetry is what the reciprocal does, and it is the reason a retention programme can beat an acquisition programme even when the acquisition programme looks cheaper per unit of effort.

The second thing is the catch. Expected life of 1 / churn holds only while churn is constant, and almost nobody's churn is. Real cohorts lose a large share in the first period and then settle, because the customers who were never going to come back leave early. Blend those two phases into one annual rate and the formula will quietly hand you the right total for the wrong reasons.

Two cohorts, identical lifetime value, different businesses

Here is that failure in numbers. Cohort A churns at a steady 50 percent a year: $90.30 of contribution in year one, $45.15 in year two, $22.58 in year three, and so on. Total, $180.60.

Cohort B loses 70 percent of its customers in the first year and then churns at 30 percent. Year one is the same $90.30, because everyone is present for at least part of it. Year two is only $27.09, because just 30 percent of the cohort survived. Year three is $18.96, year four $13.27. Total, also $180.60, to the cent.

One lifetime value, two businesses. Cohort A has collected $135.45 by the end of year two, 75.0 percent of its lifetime value. Cohort B has collected $117.39, 65.0 percent. The missing $18.06 per customer is working capital you have to find somewhere, and it is invisible in a formula that only reports the total. Money later is also worth less than money now: discount both at 10 percent a year and cohort A is worth ($90.30 x 1.1) / 0.6 = $165.55 while cohort B falls to $158.03, because more of its cash sits further out.

So the total is not the whole answer. Run the LTV calculator on your own figures, then look at when the money actually lands.

What the number is allowed to do

Lifetime value sets a ceiling on acquisition cost, not a budget. The $180.60 arrives over several years, and you pay the acquisition cost today. Monthly, this customer produces $90.30 / 12 = $7.525 of contribution. A $60 acquisition cost is repaid in $60 / $7.525 = 7.97 months, call it 8.0. If your cash position only lets you wait six months, the amount you can actually spend against is 6 x $7.525 = $45.15, a quarter of the lifetime value, and the other $135.45 is real but not yet available. The payback period calculator and the note on CAC payback period cover the cash side properly.

Two habits keep the number honest. Recompute it per cohort and per channel rather than sitewide, because a blended lifetime value is an average of customers you would happily buy more of and customers you should stop buying, and the average hides both. And recompute churn before you recompute anything else, since it is the only term in the formula that pays back more than you put in.

Do this next

Take your last twelve months of orders and calculate one figure before anything else: what share of customers who bought in the same month a year ago have bought again since. That is your annual retention, and one minus it is the churn term. Put it into the formula alongside contribution margin rather than gross margin, and compare the result with whatever lifetime value your current model reports. If the two are more than 20 percent apart, the model is running on a guessed lifespan or the wrong margin, and every acquisition cost target built on it is off by the same amount. Fix the churn term first, then check your customer acquisition cost against the corrected ceiling.

Related reading: the LTV to CAC ratio and the CAC formula.