Customer lifetime value
How much each customer will be worth, before they demonstrate it
The problem
Customer value is almost always measured looking backwards. How much they spent, how many times they bought, how long they have been around. It is certain information and it arrives late: it describes a relationship that already happened, when the decision that needs to be made is about the one to come.
And it hides something worse. Robert Kaplan and V.G. Narayanan from Harvard showed that the 80-20 rule does not apply to profitability by customer because it falls short. In the cumulative curve, the most profitable 20% generate between 150 and 300% of total profit, the bulk in the middle breaks even, and the bottom segment destroys between 50 and 200%. The company ends up with its 100% after subtracting the loss — and does not know it until someone looks case by case.
With an average, none of this is visible. A customer who buys frequently and returns half, or who consumes more support than they leave behind, appears as a good customer.
The solution
An expected value per customer for a defined horizon, calculated from the signals available at that moment.
The word doing the work is "expected". It is not what the customer spent: it is what they are estimated to spend. That allows something that history does not allow, which is scoring someone who has not done anything yet. A two-week customer has no history, but they have an entry channel, a first product, a purchase moment and a payment method, and that already produces an estimate. It is the difference between knowing how much to pay for a customer after a year and knowing it on day one.
Two definitions that are agreed before building anything, because they change the entire number. The horizon, which may be twelve months, twenty-four or unlimited. And whether it is estimated on revenue or on margin. The margin one is what is useful for decisions, and it is the one almost no one calculates, because it requires bringing the cost-to-serve down to the customer level.
Why an estimated value and not the historical one
Because the two decisions that depend on customer value happen before the history exists. How much to pay to acquire them and how much attention to give them are decided at the beginning of the relationship, not at the end.
There is also a survival reason. The relationship between customer value and acquisition cost is what defines whether the business sustains itself, and not all businesses with the same average endure the same. A company with a comfortable ratio absorbs a rise in acquisition cost and stays standing. One that is tight, does not. Knowing which customer is worth how much stops being optimisation when the cost of acquisition rises every year.
How it integrates
The estimated value enters the CRM as a field and, when there is advertising investment, is sent to the platforms as a conversion signal, so that bidding adjusts to the predicted value of each person instead of treating all conversions equally. That is where the model becomes money most directly.
The same model, other applications
Estimating future value from early signals applies wherever a decision today must be made about something collected later.
Lead value.
How much a contact is worth before working them, to distribute the commercial effort.
B2B account value.
Potential of an account beyond what it bills today, for the team managing the portfolio.
Channel value.
Which acquisition channel brings customers who stay, and not just those who convert cheaply.
Donor or member value.
The same calculation where the relationship is one of recurring contribution rather than purchase.