What a customer is worth, and why nobody knows it

Almost every account we take over is optimising toward a number nobody has checked. Working backward from contribution margin instead.

The first question on a first call is what a customer is worth to you. Most of the time the answer is a pause, then a number somebody picked in a meeting two years ago. Occasionally it is a number from a competitor's case study, which is worse, because it is confident and unrelated.

It matters because it is the only input that makes a cost per acquisition mean anything. A cost per lead of ₹400 is neither good nor bad. It is good if a lead converts at one in five and the customer contributes ₹9,000 in margin. It is a slow bankruptcy if a lead converts at one in fifty and the margin is ₹800.

The arithmetic is not complicated, it is just usually skipped. Take the revenue from an average order, subtract cost of goods, subtract shipping, payment fees and returns, and you have contribution margin, not revenue. Then decide what share of that margin you are willing to give up to acquire the customer, and whether you are buying a first order or a relationship. If people reliably buy four times, you can pay considerably more for the first one, provided the retention actually happens rather than being assumed.

That last clause is where most of these models quietly break. Lifetime value is frequently used as permission to overspend on acquisition, based on repeat behaviour nobody has measured. If you are going to price acquisition against a second and third purchase, the sequences that produce those purchases have to exist first. Otherwise the model is a forecast being used as a fact.

What this changes in practice: the target stops being a number the platform reports and starts being a number your accountant would recognise. It also gets much easier to say no. When a channel cannot deliver below the ceiling, the answer is to stop buying there rather than to raise the ceiling until the channel looks acceptable.

We agree this before an account starts, write it down, and report against it. It is not sophisticated. It is just the difference between spending money and buying customers.