Unit Economics

Unit economics is the calculation for a single unit of a business. A unit can be a customer, an order, a subscription, or a distance traveled. Two figures are compared: what that one unit brings in and what it costs. If something is left over at the end, the company earns money on every additional unit. If nothing is left over, it loses additional money with every further unit. The English term has also become established in German, because it comes up constantly in investor conversations and quarterly reports.

Why growth without this calculation is dangerous

Many young companies grow very quickly and still post high losses. That need not be a problem. What matters is the question of where the losses come from. If they stem from one-time expenses such as software development or building a factory, this can pay off later. If, on the other hand, they stem from every single sale, growth makes the situation worse.

A delivery service illustrates this clearly. If a delivery costs nine euros on average and the customer pays seven, the company loses two euros on every order. If it doubles the number of orders, it also doubles the loss. No investor can rescue such a model in the long run. That is why funders often scrutinize unit economics more closely than total revenue.

Conversely: if the unit economics are positive, growth becomes self-sustaining. Every new customer then contributes an amount that helps pay the company’s fixed costs. Beyond a certain point, these fixed costs are covered, and the rest is profit. This point is called the break-even point, i.e., the profitability threshold.

What goes into the calculation per customer

Two key metrics are central. The first is customer acquisition cost, abbreviated CAC: everything spent to win a single new customer. This includes advertising, discount campaigns, and commissions for sales staff. The second is customer lifetime value, abbreviated LTV: the total profit contribution a customer delivers over the entire duration of the customer relationship.

As a rule of thumb, an LTV that is at least three times as high as the CAC applies. At a ratio of one to one, the company earns nothing at all. It is also important how long it takes until the acquisition costs are recovered. This time span is called the payback period. Twelve months is considered solid for software subscriptions, three years is considered risky.

When it comes to cost per unit, only the variable items count, i.e., those that arise anew with every sale. For an AI chatbot, these are mainly the computing costs for each response. Office rent is not included, because it is incurred regardless of the number of customers. It is precisely at this distinction that many calculations fail. Anyone who shifts variable costs into fixed costs makes the unit economics look better than they actually are.

Unit economics in AI companies and stock market news

In news about start-ups, the term almost always appears when a business model is in doubt. Ride-hailing services, delivery platforms, and e-scooter providers had to explain for years when their per-ride calculation would finally work out. In such cases, analysts specifically ask about the contribution margin per order. If an executive answers evasively, that is a warning sign.

The topic is especially current with AI products. Every response from a language model costs computing time on expensive graphics cards. A subscription for twenty euros a month can quickly stop being worthwhile for a heavy user. That is why providers are introducing usage limits or building smaller, cheaper models for simple queries. All of this represents attempts to push down the cost per query.

A common misconception is to equate unit economics with profitability. A company can earn money per customer and still make an overall loss because fixed costs are still too high. The metric therefore says nothing about the current bottom line, but about future prospects. It answers the question of whether the problem can be solved by acquiring more customers.