LTV you can actually defend
Lifetime value is the most motivated number in business. It exists, in most companies, to justify spending more on acquisition, and numbers that exist to justify things have a way of growing. Here is how to build an LTV that would survive a hostile audit, and why the number next to it matters more.
A defensible customer lifetime value is contribution margin, not revenue; median, not mean; observed from your own cohorts, not extrapolated from a formula; and capped at a finite horizon your data has actually witnessed. Build it that way and the number typically shrinks by half or more, which is uncomfortable, and correct. Then put payback period next to it, because LTV without timing has bankrupted better businesses than yours.
Nobody inflates their LTV on purpose. It happens the way most measurement flattery happens: at every step where a choice exists, the choice that makes the number bigger also feels reasonable, and four reasonable choices later the number is fiction. Let us walk through those four choices, because undoing them is the whole method.
The four flatteries
Flattery one: revenue instead of margin. A customer who pays you $1,200 over their life did not hand you $1,200 to spend on acquiring the next customer. Cost of goods, delivery, payment fees, support time, discounts, refunds all come out first. What remains, contribution margin, is the only number that can honestly be compared against an acquisition cost, because acquisition is paid in real dollars, not in revenue. A revenue-based LTV compared against CAC is a category error, and it is the single most common one we see.
Flattery two: the mean instead of the median. Customer lifetimes are heavily skewed. Most customers leave early, a few stay for years, and those few drag the average up dramatically. An average lifetime of 19 months can coexist with a median of 7, meaning the typical customer you buy with your ad spend looks nothing like your LTV model's imaginary one. The mean answers "what is our base worth in total, divided by heads." The median answers "what is the next customer we acquire likely to be worth." Acquisition decisions need the second question.
Flattery three: the infinite horizon. The textbook shortcut, LTV equals average revenue per month divided by monthly churn rate, quietly assumes customers keep leaving at a constant rate forever, and sums an infinite series. Divide by a 3 percent churn rate and the formula asserts a 33-month lifetime, with a meaningful slice of the value arriving in years your business has not existed for. No operator would accept "we will collect this in year four" as support for a spend decision today, but that is what the formula smuggles in.
Flattery four: constant churn. The same formula's deeper flaw is that churn is not constant, it varies enormously with customer age, high early and low late, the exact curve we drew in our piece on cohort churn. Feeding a blended churn rate into the formula bakes the blend's distortions into the valuation. If your base is currently old and loyal, the formula overvalues the new customers you are about to buy; if your base is young, it can even undervalue them. Either way it is measuring your mix, not your future.
Four reasonable-feeling choices, each nudging the number up, and the result is an LTV describing a customer who does not exist.
Building the honest version
The defensible method needs no statistics degree and no software beyond a spreadsheet, because it is mostly subtraction and patience. Four steps.
Start from cohort survival, not a churn rate. Take your actual cohorts, customers grouped by start month, and their observed retention at each age. This is the same table from the cohort analysis, doing double duty. No formula stands in for the curve; the curve is the input.
Attach margin to each month of survival. For each month of age, multiply the share of the cohort still active by the contribution margin a customer generates in that month. Margin, again, is revenue minus the costs that scale with serving the customer. If margin varies with tenure, long-tenured customers often buy more, discount less, and cost less to support, use the tenure-specific figures; the variation is usually in your favor and this is the one honest place the number gets to grow.
Cap the horizon at what you have witnessed. Sum those margin-weighted months out to a fixed window, 24 or 36 months for most businesses. Not because value past month 36 is worthless, but because you have no evidence about it, and a valuation is only as strong as its weakest assumption. When your oldest cohorts age past the cap, extend it. Evidence first, then horizon.
Compute it per segment, or do not bother. One blended LTV inherits every mix problem the blended churn rate had. The number becomes a decision tool when it is split by the things you can actually choose between: acquisition channel, first product purchased, plan tier. An LTV of $420 blended is trivia. An LTV of $640 from organic and referral versus $230 from discount-led prospecting is a budget instruction.
A subscription box at $45 a month, 55% contribution margin after product, shipping, fees, and support. Cohort data shows the median customer lasts 9 months, with the survival curve flattening near 20% at month 24.
Formula LTV (the flattering one): $45 × 33 months (from 3% blended churn) = $1,485
Defensible LTV: sum of (survival × $24.75 margin) over 24 observed months = $310
If CAC is $180: the first number says spend more, aggressively.
The second says you make $130 per customer over two years, so the next question is timing.
Illustrative arithmetic. The gap between $1,485 and $310 is not a rounding disagreement; it is two different companies, and only one of them exists.
Payback period: the number that keeps you solvent
Even a perfectly honest LTV shares a weakness with every lifetime metric: it ignores time. Money you collect over 24 months has to be spent today, and the gap between those two moments is financed by your bank account. This is why the second number matters as much as the first: CAC payback period, the months until the accumulated contribution margin from a customer covers what you paid to acquire them.
Two businesses can carry the same LTV:CAC ratio and live in different worlds. One recovers its acquisition cost in 5 months and can recycle the same cash five times in the time the other, recovering in 24 months, uses it once. The slow one is not just less efficient; it is fragile, because every growth push stretches the same cash further and a soft quarter arrives before the payback does. Growth amplifies the difference: the fast-payback business accelerates itself, the slow one digs a hole that looks like success on the LTV slide.
The folklore benchmark, an LTV:CAC ratio of 3:1, is fine as a smell test and dangerous as a target, precisely because it is timing-blind. Our working rule for small and mid-size businesses: treat payback as the binding constraint, aim for under 12 months, and treat the ratio as a secondary check. A 2.5:1 business with 6-month payback is usually healthier than a 4:1 business at 30 months. You can put your own numbers into our free calculators to see where you land.
Sensitivity: how sure are you, really?
One more habit separates a defended number from a decorated one: show how much the LTV moves when its weakest assumption wiggles. If a one-point change in monthly churn moves your LTV by 25 percent, that is not a footnote, it is the finding. It means your valuation is a bet on retention holding, and retention should then be monitored like revenue.
| Assumed monthly churn | Formula lifetime | Implied LTV | vs baseline |
|---|---|---|---|
| 2% | 50 months | $1,238 | +50% |
| 3% (baseline) | 33 months | $825 | — |
| 4% | 25 months | $619 | −25% |
| 5% | 20 months | $495 | −40% |
That table uses the naive formula deliberately, to show the leverage hiding in one assumed number: a single percentage point of churn either way swings the valuation by hundreds of dollars per customer. Any decision built on the middle row is really a decision about whether the top or bottom row is coming. The cohort-observed method is far less sensitive, because it replaces the assumption with evidence, but even there, run the cap at 24 and 36 months and show both. A stakeholder who sees the range trusts the analyst more, not less, and the same principle runs through everything we write about measurement: confessed uncertainty beats confident fiction. It is the difference between a ROAS that flatters you and a numbers you can operate on.
What the honest number changes
When the LTV shrinks to its defensible size, a few decisions usually flip, and they flip in a healthy direction. Acquisition bids come down to what customers are actually worth, which trims exactly the spend that was buying the worst-fit customers. Channel budgets shift toward the sources whose cohorts stay, not the sources whose clicks are cheap. And retention work, onboarding, early value, the second purchase, suddenly shows its real return: if the median customer leaves at month 9, one additional month of median lifetime is often worth more than several points of conversion rate, and it is usually cheaper to buy. The early-warning signals that make that possible are the subject of the next piece in this series.
None of this requires believing retention is magic or acquisition is bad. It requires only that the two be priced with the same honesty. A business that knows its true LTV, its payback period, and its cohort curves is not guaranteed to grow. It is guaranteed something rarer: that when it spends a dollar, it knows what it is buying.
Common questions
How do you calculate customer lifetime value?
Sum, month by month, the share of a cohort still active times the contribution margin each active customer generates that month, out to a finite window your data has actually observed, typically 24 or 36 months. Use margin rather than revenue and the median customer rather than the mean. Avoid the shortcut of dividing monthly revenue by churn rate; it assumes constant churn and an infinite horizon, and both assumptions inflate the number.
What is a good LTV to CAC ratio?
Treat 3:1 as a smell test, not a target, because the ratio ignores timing. Pair it with CAC payback period, the months until a customer's accumulated margin covers their acquisition cost, and for most small and mid-size businesses hold payback under 12 months as the binding constraint. A modest ratio with fast payback usually beats an impressive ratio with slow payback.
Why is my LTV number misleading?
Check for the four flatteries: revenue instead of contribution margin, mean instead of median, an infinite horizon from the ARPU-over-churn formula, and a constant-churn assumption fed by a blended rate. Each pushes the number up. Correcting all four commonly cuts a headline LTV by half or more, and the corrected number is the one your acquisition spend should be priced against.
Want your LTV rebuilt on evidence?
We take your billing data, build the cohort-observed LTV by channel and segment, put payback next to it, and show you which acquisition spend the honest numbers still support. Some of it will survive. The interesting part is what does not.
Book a 30-min call →