Unit Economics Payback Calculator

Paying to acquire a customer is a purchase. You hand over the acquisition cost today, and in return you receive a stream of monthly margins that lasts as long as the customer stays. Seen that way, the questions sort themselves out. How much does the stream add up to? When does it repay the price? What return does the purchase earn? Those are the questions any investment answers, and a single ratio printed on a dashboard answers none of them well.

The usual shortcut is lifetime value over acquisition cost, often called LTV to CAC. Customer acquisition cost is, in its simple form, the money spent on winning new customers divided by the number won. Lifetime value is defined as the present value of the cash a customer relationship is expected to produce, which also makes it an upper limit on what acquiring that customer is worth. The definition already contains the two words most shortcuts drop: present, and cash.

Margin, not revenue

Start with what a customer actually leaves behind each month. It is not the price. A contribution margin is the marginal profit on each unit sold: what remains after the costs that rise with that sale. For a subscription that means the price times the gross margin, less whatever it costs to serve one more account. A widely repeated mistake is to compute lifetime value from total revenue or gross margin, which can overstate it several times over.

Take a fictional plan priced at 50 a month with an 80 percent gross margin and a cost to serve of 6. The contribution margin is 50 times 0.8 minus 6, which is 34. Suppose the average customer stays 24 months and acquisition costs 400. Revenue times lifetime says the customer is worth 1,200, three times the cost, and payback looks like eight months. Margin times lifetime says 816, about twice the cost. The first figure passes the popular three to one test. The second does not.

That three to one test deserves a moment, because it is often treated as a law. It is a rule of thumb. Common guidance calls a ratio near one a loss once the cost of service is counted, calls three a healthy level, and reads anything above three as room to spend more on growth. Those thresholds carry no derivation. They summarise experience, and they assume the lifetime value underneath was computed honestly.

Two leaks: leaving and waiting

Lifetime value shrinks for two separate reasons, and they compound. The first is that customers leave. In the simplest model a fixed share of active customers renews each period and the rest are gone for good, a leaky bucket. If 95.8 percent stay each month, the chance that a new customer is still paying in month k falls by that factor every month. Converting the stated 24 month lifetime into a retention rate is a one line step: retention is one minus one over the lifetime, because the expected number of months paid is one over one minus retention.

The second leak is waiting. Money received later is worth less than money received now, because it could have been earning a return in the meantime. Discounting expresses that by multiplying each future amount by one over one plus the rate, raised to the number of periods away. A 10 percent annual rate becomes a monthly rate by taking the twelfth root of 1.10 and subtracting one, about 0.797 percent, which is the inverse of the usual effective annual rate formula. Practitioners often skip this step, and the result is biased high.

Put both leaks together and the stream of expected margins becomes a geometric series: each month is the last one times retention and divided by one plus the discount rate. A geometric series whose ratio is below one in size converges to its first term divided by one minus the ratio. For a customer who pays from the first month, that gives the margin divided by one plus the discount rate minus retention. In the example, 34 over 1.00797 minus 0.95833 is 684.92. Discounting alone took about 131 off the undiscounted 816.

Both leaks are needed. Retention without discounting treats a sale in year five like cash today, and discounting without retention pays a customer who has already left.

Textbook versions of this formula multiply by retention once more, because they assume the first margin arrives only after the customer has renewed once. Neither convention is wrong. They describe different moments of counting, and the difference here is 684.92 against 656.38. What matters is stating which one you used, and whether each month’s margin lands at its start or its end. Federal guidance on benefit and cost analysis makes the same point about years: discount factors assume amounts arrive at year end unless you shift them to mid year on purpose.

Payback is the cash view

Value tells you how much. Payback tells you when. The payback period is the time it takes for cumulative cash to climb from minus the investment back to zero. It is popular because anyone can follow it, and it is criticised because the plain version ignores the time value of money and says nothing about what happens afterwards. The discounted payback period fixes the first problem by accumulating discounted amounts instead, finding the crossing by interpolation within the period where the total turns positive.

Draw that cumulative discounted curve for the example and it starts at minus 400, rises steeply, and bends as retention and discounting wear each new month down. It crosses zero at month 17.4. The undiscounted curve crosses earlier, at 15.8, and the old eight month answer from revenue alone is not even close. Lengthen the lifetime to 36 months and the discounted crossing moves to about 15.1. Payback is the number a cash constrained business lives by, because until the crossing every new customer is a loan.

So the honest answer for this customer is a little over seventeen months of waiting, not eight. That gap is what the shortcut was hiding.

Payback has a blind spot of its own. It ignores everything after the crossing. A plan that pays back in month 14 and then churns out completely can look better than one that pays back in month 18 and keeps paying for years. That is why payback belongs beside value, not instead of it, and why the curve is worth looking at as a whole rather than at one point on it.

The return on buying a customer

Because acquisition is an investment, it has an internal rate of return: the discount rate at which the net present value of the purchase is exactly zero. For a constant margin and constant retention with no horizon the rate has a short closed form, the margin over the acquisition cost plus retention minus one. In the example that is 34 over 400 plus 0.958 minus 1, or 4.33 percent a month. Cut the stream off at 60 months and the rate falls to 4.28 percent, about 65 percent a year.

The internal rate of return comes with a warning label that most calculators leave off. When cash flows change sign more than once, there can be several rates that all make the net present value zero; a stream of minus 10, plus 21, minus 11 is satisfied by both 0 and 10 percent.

Descartes’s rule of signs bounds the count: the number of positive roots is the number of sign changes, or fewer by an even number. A customer who costs money to serve in the early months of a ramp and pays later can produce exactly such a stream. When that happens the honest answer is that the rate is not unique, or that none exists, rather than whichever root a solver happened to find first.

The inverse questions are often more useful than the forward ones. The most you can pay to acquire a customer and still break even by month M is simply the discounted value of the first M months. In the example, a target of 15 months allows a CAC of 363.81, so the current 400 is about 36 too much. Alternatively, keep the price and ask what monthly retention would bring payback inside 15 months; with every other input fixed it is 97.3 percent, an expected lifetime of about 37 months.

Reading the tornado

Every input above is an estimate, and some matter far more than others. A tornado diagram makes that visible. Each input is moved alone to a low and a high value while the rest stay at their base, the output is recomputed, and the bars are stacked with the widest swing at the top. Guidance on public cost benefit work recommends exactly this kind of recomputation for the important inputs and the discount rate.

Moving each input of the example by 20 percent, price and gross margin top the payback chart with a swing of almost 16 months, because they scale the margin directly. Acquisition cost comes next, then churn, then cost to serve. The discount rate barely moves payback at all, under a month, because payback arrives early, before discounting has had time to bite.

The chart has a structural limit worth stating plainly. Changing one input at a time cannot reveal interactions: a lower price combined with higher churn can be far worse than either bar suggests. Plus and minus 20 percent is a convenient choice, not a standard. And none of these figures predicts anything. They show what the typed numbers imply for one average acquired customer, which is exactly as good as the margin and retention you feed in.