When payback outlasts the customer
Two numbers sit in most board decks, usually pages apart. One says payback is 18 months. The other, buried in a retention tab, says half of a given cohort is gone by month 14. Read separately, each looks survivable. Read together, they describe a company paying to acquire customers who leave before they have covered what it cost to win them.
That gap is one of the quieter ways unit economics fail. Nobody is lying. The CAC is real, the gross margin is real, the churn is real. They are just never placed next to each other, so the arithmetic that connects them never gets done.
Payback is a claim about the future
A payback period is a bet that the customer will still be around to make the payments. When you say a cohort pays back in 18 months, you are assuming eighteen months of gross profit actually arrive. If the median customer in that cohort churns at month 14, four of those months are hypothetical. The cohort pays back on paper and loses money in fact.
This is why payback and lifespan can never be assessed in isolation. A 14-month payback is healthy against a four-year customer and fatal against an eleven-month one. The number itself carries no verdict. Only the pair does.
Payback is a curve, not a point
The single "18 months" figure hides its own shape. Acquisition cost is recovered gradually, as each month of gross profit stacks up against what you spent up front. Early on, the ratio of value returned to cost paid climbs steeply. Then, as customers drop off, the curve of cumulative gross profit per customer flattens. Payback is simply the month that curve crosses the line where cumulative gross profit finally equals CAC.
Plotted this way, the question stops being "what is payback" and becomes "does the curve get there before it flattens." Some cohorts cross the line at month 9 and keep climbing. Some crawl toward it, stall at 0.8, and never arrive. A cohort that never crosses has no payback period at all, not a long one. A blended headline cannot tell those two cases apart. Watching each cohort's ratio age can.
The retention curve is the other half
Lifespan is not a single number either. It is a curve of how much of each cohort is still active at each month of age. The month a cohort falls to half its original size is a reasonable read on the median customer's life, and it sits right there on the retention curve if you look for the crossing.
Line the two curves up and the diagnosis is immediate. If a cohort hits fifty percent retention well before its value-to-cost ratio reaches 1.0, the average customer is gone before the average dollar is recovered. Growth accounting will still show new logos landing every month. The top line will still rise. Underneath, each new cohort is quietly underwater, and the P&L will not say so for a year, because the spend is booked now and the shortfall shows up only later, as revenue that never materializes.
Why the round number is dangerous
"Payback is 18 months" invites a comparison to a benchmark and a nod. Eighteen is under two years, two years sounds fine, move on. But the benchmark is the wrong question. The right question is whether your customers live long enough to make the payment, and that answer is specific to your cohorts, not to a rule of thumb. A company with an 18-month payback and a six-year customer is in far better shape than one with a 10-month payback and a nine-month customer. The single number ranks them backwards.
What Levian puts in front of you
Levian reports the month a cohort's gross-margin-adjusted value first covers its acquisition cost, and shows that ratio for each cohort as it ages, so the cohorts that cross the line and the ones that flatten below it are both visible. Logo retention by cohort shows how much of each cohort is still active at every month of age, so the point where a cohort halves lives on the same report as the point where it pays back. Reading the two together is the whole exercise, and it is done before an investor does it for you.