Almost everyone calculates LTV the same way. You take the revenue a customer brings in over their life and subtract the variable costs that come with it.
It is clean, it fits on a slide, and it is wrong by about a third. It is wrong because those tidy categories leave out what it actually costs to serve a real customer over time. Four costs in particular go missing, and once you put them back in your real LTV comes out lower but finally becomes a number you can act on. Here is the walkthrough.
The textbook number leaves out the cost of serving the customer
The standard formula counts the money a customer brings in and the margin on the goods, and then it stops. It treats the customer as a revenue stream with a product cost attached and nothing else. Real customers are not that tidy. They return things, they email support, and they redeem rewards, and every one of those costs lands on the same P&L your LTV is supposed to describe.
There are four costs the textbook number ignores: reverse logistics, restocking, support, and loyalty point liability. Each one looks small on its own, and together they invert the math.
Boring to measure, which is exactly why they get skipped
None of these four costs arrives with a customer's name on it. They sit in aggregate ledgers, split across teams, and nobody has done the work to figure out what a single return or a single ticket actually costs end to end. That missing research is the reason the costs get left out. The work is tedious, it crosses departments, and no one owns it, so the number never makes it into the formula.
To put them to use you extend the equation, then extrapolate each cost down to the customer.
The real formula Real Contribution LTV = Contribution LTV − (reverse logistics + restocking + support + loyalty point liability) |
| ||
| ||
| ||
|
The fastest way to push these costs down to the customer is to average them, and an average quietly assumes every customer costs the same to serve. Simple approach, but this fuels the whole problem.
The customer who returns half of what they buy and opens a ticket a month is nothing like the one who does neither, even at identical spend. So read this walkthrough as a directional tool for comparing cohorts, not a verdict on any single buyer. Getting to the real per-customer difference is where this goes next.
Let’s put real numbers on it. Take a typical apparel brand with a $100 order and a textbook contribution margin of 25-30%, so the slide says every $100 of customer revenue throws off $30.
Now layer in the four costs using published benchmarks:
That is $14 of cost against a $30 contribution, so the number falls to $16. Return-heavy categories like apparel and furniture push past a third. Lighter ones like electronics and beauty land lower.
Measuring the costs is the floor. A system is what lifts the number.
A lower LTV number is not bad news, it is an actionable one. Each of the four costs is a lever you can pull, and you cannot pull a lever you refuse to measure. So start by measuring.
Three things worth pulling this week:
| ||
| ||
|
Here is the catch. Done by hand, cohort by cohort, this is heavy work, and when you finish you have only measured the problem. Measuring cohorts tells you the number. It does not move it.
Lifting the number takes a system that reads these costs at the individual customer level, surfaces the buyers quietly eating your margin, and changes how you treat them while it still matters. A tighter return path for a serial returner. A sharper offer for a heavy discounter. Support routed by customer value, not by queue position. Averages bury those customers inside a blended number. A system pulls them out, and managing them one at a time is what lifts the overall LTV.
One bet for the road My wager is that your real LTV, with the cost to serve inside it, sits at least a third below the number on your current slide. Which means you are not as profitable per customer as you think. Two things follow from that gap. First, the four costs underneath your LTV are efficiency work you are not doing. Every dollar you drive out of reverse logistics, restocking, support, and unredeemed points lands on the customer’s profit line, and almost nobody is pulling those levers hard enough. Second, and this is the one that hurts, LTV is only half of the ratio you use to decide what a customer is worth acquiring. If your real LTV is a third lower than the slide, that side of the LTV:CAC math is already inflated. And the CAC side is almost certainly understated. Most brands under-count what acquisition actually costs, leaving out fully loaded ad spend, agency fees, tooling, and the discounts that close the first order. So the ratio is wrong on both ends. Real profit is lower, real acquisition cost is higher, and you can afford far less marketing than the slide says. The brands that win are not the ones with the prettiest LTV. They are the ones who know the real number on both sides of the ratio, drive down the costs underneath it, and size their acquisition to the truth. |
About Life After CACLife After CAC is the publication for founders, marketing leaders, CX leaders, and operations leaders at mid-market DTC brands. Each issue tackles one operational dimension of the LTV problem. Some of this will resonate and some will start arguments. Don't be shy about pushing back since I may be wrong about some of this. When I am, tell me. I'm the founder of Onward. Our platform personalizes the purchase experience for every single customer to maximize your brand's LTV opportunity. Our intelligent AI-driven suite of software spans the entire customer journey, from loyalty to returns and everything in-between, allowing brands to evolve beyond one-size-fits-all policies to a truly tailored experience based on each customer's value to your business. You won't get pitched in these emails. If you want to know what we do, useonward.com is one click away. This space is where I argue with the numbers. |
Hit reply and tell me the one cost your LTV math leaves out that you suspect is the biggest. Returns, support, restocking, or point liability. I'll anonymize the answers and rank what operators say in a later issue.
—Josh