How to work out lifetime value without years of data
Estimate customer lifetime value from a few months of orders using gross margin a capped horizon and early cohorts rather than an optimistic forecast.
You need the number before you have the history
Lifetime value is the figure that decides whether your acquisition cost is affordable. The difficulty is that it describes behaviour over years and most businesses need it after about twelve months of trading. Waiting for real data means making budget decisions without it for two or three years.
The way out is to stop treating lifetime value as a prediction and treat it as a bounded estimate. A short horizon figure built on gross margin and your earliest cohorts is defensible. An infinite horizon figure built on revenue is a wish with arithmetic attached.
What an early estimate excludes
An estimate built on twelve months of data excludes everything after twelve months. It cannot see the customers who lapse in year two or the ones who return in year three. Any model claiming a five year value from one year of orders has assumed the answer rather than calculated it.
It also excludes cost to serve beyond the cost of goods. Support time delivery exceptions and returns all reduce the real value of a customer. Say so when you present the figure.
The work costs an hour or two with your order data if you can export orders with a customer identifier and a date. If you cannot then the first task is recording repeat purchase properly. No model repairs missing order history.
Who should model this and who should not
If you sell a genuine one time purchase then do not build a lifetime value model at all. A wedding photographer or a visa service or a borewell contractor earns margin once. The honest figure is the margin on that single sale and your model is one line. Dressing it up as lifetime value tends to produce an inflated number by quietly counting referrals as repeat orders.
Do not build one either if it will be used to justify spending money you do not have. Lifetime value arrives over months. Acquisition cost leaves your bank account this week. A business with four weeks of cash in hand needs a payback calculation instead. Only one of those two questions keeps the lights on.
The exercise is worth it for subscriptions consumables diagnostics education and anything with a natural repurchase cycle. It matters most where the first order loses money by design.
Three ways to build a figure you can defend
The simplest is a fixed window. Pick a horizon you have data for such as ninety days. Take the customers acquired in one month and total the gross margin they produced within that window. Divide by the number of customers. That is a measured figure rather than a forecast and acquisition cost can be compared against it directly.
The second method extends slightly. Work out average order value multiplied by orders per customer per period multiplied by gross margin percentage. Then cap the horizon at twelve or twenty four months and state the cap beside the number. The cap is the discipline that keeps the figure honest.
For subscriptions the mechanism is churn. The reciprocal of your monthly churn rate gives an expected lifetime in months so a business losing a twentieth of its customers each month has an expected lifetime of twenty months. Multiply that by monthly gross margin per customer. Early churn figures are unstable on small numbers so treat the result as a range. In every method split the figure by acquisition channel. Customers won on a deep discount behave differently from customers won on search and a blended figure hides that.
The check that matters and the ratio that flatters
Track realised contribution margin per cohort at fixed checkpoints. Where a cohort stands at ninety days and at one year tells you whether your model was close. That is the only validation available and it costs nothing once the cohorts exist.
The ratio of lifetime value to acquisition cost is the flattering figure. It can be made to look healthy by extending the horizon by a year and nothing in the business changes when you do that. Revenue based lifetime value flatters the same way because it ignores the cost of delivering the product. Use margin. State the horizon. Report payback period alongside it.
The common mistake
The expensive error is building lifetime value on revenue with no horizon and then raising acquisition spend to match. The number is large enough to justify almost any campaign. For a few months the dashboard looks excellent because new customers keep arriving and the margin shortfall has not surfaced yet.
What follows is a cash problem rather than a marketing problem. The money left the account in month one and the value arrives across two years. Businesses in this position usually discover it when a supplier payment is due. Margin and a capped horizon are not conservatism. They are what makes the figure usable for a decision about this quarter's budget.
Put a date on the estimate
Record the figure with the date it was calculated and the cohorts it came from. Then recalculate each quarter as those cohorts mature. A model that never changes is one nobody is checking.
Share the horizon whenever the number is quoted. A ninety day value and a two year value are different claims and confusing them is how an affordable acquisition cost becomes an unaffordable one.
This guide explains how something works. It is not legal, tax or financial advice, and rules change. Confirm anything regulatory with a qualified professional before you act on it.