Business Models: What Would You Give Up to Earn More?

Most descriptions of a business model stop at how the money arrives: subscriptions, advertising, transaction fees, one-off sales.

That is the least interesting part. It tells you the mechanism and almost nothing about how the business will behave.

A more revealing question is: what would this company have to give up in order to earn more?

Every model has a lever for increasing revenue, and every lever has a cost. An advertising-funded product that shows more ads sacrifices the user experience. A subscription that raises prices sacrifices growth at the margin. A marketplace that raises its take rate sacrifices suppliers to a cheaper rival. Where that cost falls, and how large it is, determines what the company will do under pressure — which is a better predictor than anything stated in its strategy documents.

Who pays, and who is served

The first structural question is whether the person paying is the person being served. Where they are the same, incentives are relatively simple. Where they diverge, the product will be optimised for the payer.

This is not a moral point. It is a description of what a business is being paid to do.

A product funded by advertising is being paid to deliver attention to advertisers. It may also serve users well, and often does, because a product nobody wants delivers no attention. But when the two conflict — when a change would improve the experience and reduce time spent — the model applies pressure in one direction. Individual teams can resist that pressure; the pressure does not go away.

A product paid for directly by the person using it has the payer and the served aligned, which removes that particular tension and introduces others. Direct payment means a smaller audience, which means the product must serve those who will pay rather than everyone.

Neither is superior in general. What matters is knowing which one you are looking at, because it tells you which conflicts are structural rather than accidental.

Four ways to charge for the same thing

The same underlying capability can be monetised very differently, and the choice reshapes the business.

Sell it once. Simple, and alignment ends at the moment of purchase. After the sale, further support is a cost rather than an investment, which is why one-off products often feel abandoned. Revenue also has to be found again every period from new customers.

Sell access repeatedly. A subscription means the customer re-decides regularly, so the product must keep earning the renewal — which aligns the business with continued usefulness. The failure mode is that revenue can be protected by making cancellation difficult rather than by remaining useful, which is the same alignment running in reverse.

Sell the outcome. Charging for a result rather than a tool aligns interests tightly and requires the outcome to be measurable and attributable — which it usually is not, and where it is only partly measurable, Goodhart's Law arrives promptly.

Sell the attention it captures. Scales to enormous audiences at no charge to them, and places the served and the payer on opposite sides, with the consequences described above.

A useful observation: the best businesses tend to be ones where growing revenue and serving the customer better point in the same direction. That property is a design decision made early, and it is difficult to retrofit.

What actually makes a model durable

Beyond alignment, a few structural features separate models that last from those that work until someone competes properly.

Recurring revenue beats repeated acquisition. A business that must win each customer again every period spends heavily on acquisition and is vulnerable to a competitor outspending it. One with recurring revenue starts each period with a base.

Marginal cost shapes everything. Where serving one more customer costs almost nothing, scale produces margin and price competition can be survived. Where each customer carries real cost, growth does not automatically improve economics — which is why software and services behave so differently even at similar revenue.

Switching costs and network effects are what convert a good model into a defensible one. Without an economic moat, a model that works is an invitation to competitors.

Pricing power is the honest test. If you could raise prices ten per cent and lose few customers, you have something durable. If not, you are closer to a commodity than the business plan suggests.

The final thing worth saying is that models get chosen for reasons that have little to do with fit — because it is what similar companies do, because it is what investors expect, or because it was decided early and never revisited. Since the model determines which conflicts are structural, that is a large decision to make by default. It is worth choosing deliberately, and worth re-examining when the business has changed enough that the original reasoning no longer applies.